Latest from Tom's Hardware UK in Intel-core-i9-7900x

MORE: All CPUs Content

VRMark, 3DMark, AotS: Escalation and Dawn of War III

Test Notes

Unlike AMD's previous-gen Threadripper models, the WX-series CPUs include a Game Mode preset in the Ryzen Master software that disables three of four dies. Company representatives tell us this facilitates optimal performance in games. But AMD also provides toggles that allow experimentation with two- and four-die configurations. For this review's gaming benchmarks, we tested the Threadripper processors with Game Mode enabled. Ryzen Threadripper 2970WX consequently becomes a 6C/12T CPU for those tests.

Dynamic Local Mode is a new feature that debuts with the 2970WX, and it only applies to the WX-series models. We include a second bar chart for each game to quantify performance with this feature enabled or disabled at stock settings (labeled Creator/DM).

Gaming performance is measured at 1920x1080, minimizing graphics bottlenecks. Of course, as you step up to 2560x1440 or 3840x2160, the differences between processors shrink.

We have application test results with an overclocked Ryzen Threadripper 2920X, but weren't able to run that chip in its tuned state through our game benchmarks before it stopped working. Once we're able to get it back up and running, we'll update the gaming charts.

VRMark, 3DMark

We aren't big fans of using synthetic benchmarks to measure performance, but 3DMark's DX11 and DX12 CPU tests provide useful insight into the amount of horsepower available to game engines.

Ryzen Threadripper 2970WX's Game Mode drops the 24-core chip's count down to six active cores, but also prevents bandwidth-starved execution resources from handicapping performance during the DX11 and DX12 tests. Nevertheless, the 2970WX drops to the bottom of our chart. A dual-die 2920X easily beats the 2970WX, so it's clear that the quad-die CPU's topography is the issue.

UL's VRMark test lets you gauge your system's suitability for use with the HTC Vive or Oculus Rift, even if you don't currently own an HMD. UL defines a passing score as anything above 109 FPS. Precision Boost Overdrive yields a 13.6 FPS gain for the Threadripper 2970WX.

Ashes of the Singularity: Escalation

Ashes of the Singularity: Escalation is a computationally intense title that normally scales well with thread count.

Again, Ryzen Threadripper 2920X facilitates better performance than the 2970WX at stock settings. If there's a silver lining here, it's that PBO pushes today's subject beyond a stock 2990WX.

The second slide shows us that Dynamic Local Mode, which we used in tandem with Creator mode, imposes a slightly lower frame rate. Dynamic Local Mode's background process avoids shuffling threads that fully utilize the CPU's resources. Ashes of the Singularity does this, so the result isn't entirely surprising.

It is clear, however, that the 2970WX doesn’t scale linearly as cores and threads are added in Creator mode. You'll get similar (or better) performance in Game mode with just six cores active.

Warhammer 40,000: Dawn of War III

Threadripper processors perform well in this title. The 2970WX lands close to the top with PBO enabled, though the much cheaper Ryzen 7 2700 isn’t far behind.

The 2970WX in Game mode is faster than the same chip in Creator mode, even after we enable AMD's new Dynamic Local Mode. The company touted that switch's ability to improve performance "up to 49%,” but remember that those gains happen in Creator mode. You probably don't want to game in Creator mode.

MORE: Best CPUs

MORE: All CPUs Content

Far Cry 5, GTA: V and Hitman

Far Cry 5

Intel's Core i9-9900K is fastest in Far Cry 5. But Ryzen Threadripper 2970WX is also impressive after a bit of overclocking. Really, though, there aren't big differences between most of our test field.

While Far Cry 5 is one of the most repeatable metrics in our suite with a variance of less than 0.5 FPS between runs, we did notice some oddities after enabling Dynamic Local Mode. For instance, the results alternated between 60 and 65 FPS. This repeatable phenomenon persisted after several retests. We also recorded much lower 99^th percentile frame rates and uneven gameplay with the feature active. AMD tells us that Dynamic Local Mode is still being optimized for a broader selection of applications and games, so our observation will likely by rectified in the future. For now, Game mode is the go-to choice, as evidenced by the test results.

Grand Theft Auto V

Grand Theft Auto V favors Intel architectures and, more generally, multi-core designs with high clock rates.

An overclocked Ryzen Threadripper 2970WX slides past the Core i9-7900X, while the 2990WX doesn’t benefit as much from tuning. Once again, Game mode proves to be the obvious choice for gaming (despite a larger gain from Dynamic Local Mode this time around).

Hitman

Our Hitman benchmark was rendered almost useless by a patch that imposed a 90 FPS performance cap. A subsequent update restored our test to its prior glory.

Hitman responds well to high core counts and clock rates, so it isn’t surprising to find the overclocked Core i9-7960X in first place. The Core i9-9900K is impressive even in stock form, and the Ryzen 7 2700X proves to offer great bang for the buck.

MORE: Best CPUs

MORE: All CPUs Content

Shadow Of War and Project CARS 2

Middle-earth: Shadow Of War

Shadow of War leans heavier on graphics resources than host processing, so we don't see large deltas between the fastest and slowest CPUs. The same observation applies to our experiments with Creator mode, Dynamic Local Mode, and Game mode.

Project CARS 2

Although Project CARS 2 is purportedly optimized for threading, clock rates obviously affect frame rates. If you’re gaming in Creator mode, the Dynamic Local feature serves up a big speed-up. But the standard Game mode still offers the best performance.

MORE: Best CPUs

MORE: All CPUs Content

Office and Productivity

Test Notes

We tested all Threadripper processors in Creator mode during our application benchmarks.

Web Browser

The Krakken suite evaluates JavaScript performance using several workloads, including audio, imaging, and cryptography. Like most Web browser workloads, single-threaded performance reigns supreme.

AMD's second-gen Threadripper line-up goes a long way to improve the performance of lightly-threaded workloads, but Intel still leads in these tests.

Ryzen Threadripper 2970WX lags the 2990WX in our Krakken benchmark, but automated overclocking functionality does deliver a solid improvement. Meanwhile, there's little to no improvement in the lightly-threaded Krakken from enabling Dynamic Local Mode. This feature does seem to help in MotionMark and WebXPRT.

Productivity

The application start-up metric measures load time snappiness in word processors, GIMP, and Web browsers under warm- and cold-start conditions. Other platform-level considerations affect this test as well, including the storage subsystem.

Ryzen Threadripper 2970WX benefits from Dynamic Local Mode at stock clock rates, but still trails the 2990WX right out of its box. Interestingly, the 2920X and 2950X yield the best performance from any Threadripper CPU. Other desktop processors like the Ryzen 7 2700X and Core i9-9900K fare well, too.

Our video conferencing suite measures performance in single- and multi-user applications that utilize the Windows Media Foundation for playback and encoding. It also performs facial detection to model real-world usage. Again, mainstream processors offer the best value in these types of applications.

The photo editing benchmark measures performance with Futuremark's binaries using the ImageMagick library. Common photo processing workloads also tend to be parallelized. Unfortunately, the Dynamic Local Mode doesn't benefit all applications. It doesn't necessarily hurt performance, either. The 2% delta between our stock 2970WX falls within UL's 3% threshold for run-to-run variability with PCMark 10.

MORE: Best CPUs

MORE: All CPUs Content

Rendering, Encoding and Compression

Rendering

Many of these workloads stress the memory subsystem, diminishing Ryzen Threadripper 2990WX's core count advantage due to accesses from the remote memory controllers. The 2970WX benefits from higher clock rates to surpass the 2990WX in our single-threaded POV-Ray and Cinebench tests. Both are closely matched after we active PBO, though.

If you're looking for single-threaded supremacy, Intel's ninth-gen chips cannot be beaten, as evidenced by Core i9-9900K's dominance.

Threaded workloads are an ideal match to Threadripper's high core counts. But the Ryzen Threadripper 2990WX doesn't scale linearly in these types of workloads. That means the 2970WX's $500-cheaper price is attractive for high-end desktop PCs.

Encoding & Compression

Our compression and decompression metrics work directly from system memory, removing storage throughput from the equation. This workload should benefit from threading. But either memory throughput or poor software scaling holds the 2990WX back from realizing its potential in the compression test. The same issue affects Ryzen Threadripper 2970WX, though Dynamic Local Mode provides a slight performance uptick. But neither the 2970WX nor the 2990WX are a match for the 2920X and 2950X with their dual-die architectures. Conversely, the WX chips dominate our decompression tests, illustrating the performance trade-offs AMD's highest-end CPUs force you to make.

y-cruncher, a single- and multi-threaded program that computes pi using AVX instructions, is a great test to measure Threadripper’s AVX performance. Intel’s Core i9 employs two 256-bit AVX FMA units per core that operate in parallel, whereas Ryzen's Zen architecture divides 256-bit AVX operations across two FMA units per core. Intel's AVX instruction support shines during the single-threaded benchmark. However, spreading the workload across Threadripper's many cores helps improve its standing. Despite an eight-core advantage, the 2990WX offers little benefit over the 2970WX in the threaded y-cruncher test. Clearly, their cores suffer from a memory bandwidth bottleneck.

MORE: Best CPUs

MORE: All CPUs Content

Workstation Compute and Graphics

Workstation Compute

Many workstation applications scale very well with additional cores in certain workloads or with special plugins, but the result is always the sum of many factors and tasks in which the pure computing power of all the cores is important, but even so also not crucial. Often enough, the parallelizable tasks do not scale beyond a certain number of cores / threads, so IPC will co-decide. And that's not the advantage of AMD.

The dynamic mode of the Ryzen TR 2970WX is another such thing in its own right, because between the individual iterations of a benchmark (between 3 and 5, depending on the application), it sometimes comes to very clear differences. We can only explain it again with the missing memory controller, since many AVX- and SSE-optimized codes (but not only those) depend on memory bandwidth. And when a software solution such as Dynamic Mode intervenes, the well-intentioned can sometimes turn into the opposite.

Workstation Graphics

While workstation graphics are a niche for most readers, some might consider using Threadripper 2970WX's twelve cores and 24 threads for professional tasks. Really, though, there aren't many threaded applications for real-time graphics output. These benchmarks mostly benefit from high IPC and frequency, which isn’t one of Threadripper’s inherent strengths. The results are not bad, but also not outstanding.

Nevertheless, there are also applications that have to calculate in parallel and are grateful for every additional thread. AutoCAD is just an example of the clock dependence of fewer threads when it comes to pure 2D drafting or real-time 3D graphics output. The graphic performance is very reminiscent of the general result in gaming, it doesn’t matter if you use DirectX, OpenGL, or just the Windows GDI.

MORE: Best CPUs

MORE: All CPUs Content

Power Consumption

Interestingly, AMD reduced the idle power consumption of its second-gen Threadripper CPUs a bit. This is certainly motherboard-dependent, so be sure you're using the latest BIOS on your X399-based motherboard.

Just be ready for Windows 10 to bounce you back to 25-40W as background processes kick on and off (and particularly with PBO enabled for additional performance).

Ryzen Threadripper 2920X with PBO enabled hits clock rates as high as 4.3 GHz in our CAD workload, so power consumption spikes as well. The 2970WX is similar in that its high boost frequency results in a >64W measurement.

Our Witcher 3 benchmark is great for GPU power consumption comparisons. But it's less enlightening when we run it across different host processors because the underlying engine doesn't use enough cores. Assassin's Creed Odyssey is better about utilization, causing Threadripper's power to spike in excess of 100W. But that game's averages are far from reproducible, forcing us to keep it on the shelf.

Prime95 certainly isn't representative of an everyday scenario, but it's great for measuring peak theoretical power consumption. Some real-world applications pull even more power. For instance, a real Blender workload drives the 2970WX up to 230W, while we only measured 210W with Prime95! The same applies when we switch on PBO: Blender coaxes 447W from the Ryzen Threadripper chip compared to 416W in Prime95.

The smaller Ryzen TR 2920X is slightly different. It pulls 160W during the Blender workload, whereas Prime95 shows a power consumption of 180W. Activating PBO increases the delta: we measure 200W during the Blender test and 249W in Prime95. We can't explain why both CPUs react differently, but their results are reproducible.

MORE: Best CPUs

MORE: All CPUs Content

Temperatures

If you really needed to, you could equip the Ryzen Threadripper 2970WX with a potent air cooler, so long as you don't overclock manually or activate Precision Boost Overdrive. However, a large Blender workload would completely overwhelm it. Conventional heat sinks and fans just aren't up to the task.

Ryzen Threadripper 2970WX can hit 450W during everyday operation. That amount of waste heat requires more than air or a compact all-in-one liquid cooler. You need a much more capable thermal solution if you want to unlock the CPU's maximum performance potential.

AMD deserves credit for this processor's finely-tuned protection mechanisms. Even with PBO active, you can easily push Threadripper 2970WX to the limits of a weaker cooler without damaging it. But you have to give up proper performance in return.

XFR2 and PBO both work well on a platform with ample cooling. The chip adjusts its voltages based on telemetry data, and PBO is actually preferable to manual overclocking. While we're not fans of hidden mechanisms, PBO does exactly what you expect.

AMD prioritizes package temperature: all measurements and information are based solely on this T_die reading. For compatibility reasons, the 27°C-higher T_ctl value is used for fan control. AMD sets the upper limit for Ryzen Threadripper 2970WX at 68°C, which translates to a T_ctl value of 95°C.

At idle, our cooler keeps Ryzen Threadripper 2920X below 25°C. Under a real-world Blender workload, we average about 43°C with the CPU keeping all cores at 4 GHz. With PBO active, the 2920X accelerates to 4.15 GHz across all of its cores. The average T_die rises to just over 49°C. This gives us a maximum delta of less than 30°C for our potent cooling solution.

During a normal Blender workload without PBO, Ryzen Threadripper 2970WX reaches 3.55 GHz on all cores and averages 48°C. With PBO turned on, the all-core clock rate jumps to 4.025 GHz at an average temperature of 62°C.

The CPU does peak at 68°C though, meaning our sample is at its limit for full performance. Any higher and it would need to throttle back a bit.

MORE: Best CPUs

MORE: All CPUs Content

Final Analysis

Ryzen Threadripper 2970WX marches onto the HEDT scene with 24 cores and 48 threads. That's more cores than any competing Intel processor. But not all cores are created equal. Intel still gets more work done per clock cycle with its ninth-gen Core CPUs than AMD does with second-gen Ryzen. What's more, the 2970WX exhibits the same idiosyncrasies as the flagship Ryzen Threadripper 2900WX. It's an impressive performer in heavily-threaded workloads that aren't memory-bound, but it struggles in some applications that don't scale well based on core count (particularly if they're sensitive to available memory bandwidth).

More mainstream Ryzen and Core CPUs offer better value to gamers. Even the X-series Threadripper processors are smarter purchases than the halo WX models if you're an enthusiast. AMD's less expensive Threadrippers promise a better experience in the lightly-threaded apps that Intel continues to dominate.

But maybe Ryzen Threadripper 2970WX serves as a less expensive entry point for professionals able to exploit its copious core count in workstation-class software. The $1800 Threadripper 2990WX doesn’t always scale well, particularly in AVX-heavy tasks like HandBrake. If you're going to have to make compromises like that, you might as well save some money on the $1300 2970WX and get similar performance in the apps able to utilize its quad-die design effectively.

The X399 platform is expensive, and while drop-in compatibility with existing motherboards is a big advantage for AMD, you need a board with robust power circuitry. You also want a power supply with two EPS connectors. Cooling is a little easier on AMD's HEDT CPUs than Intel's competing Skylake-X chips, largely due to the Indium solder that AMD uses between its dies and heat spreader. But a beefy water cooler is still almost mandatory if you plan on overclocking.

AMD’s Dynamic Local Mode feature attempts to circumvent the performance issues endemic to its quad-die topology. Unfortunately, the mode is not as effective as we hoped it'd be. Some games do register big benefits. However, they still mostly trail the performance you get by flipping the CPU into Game mode via Ryzen Master, disabling 75% of the 2970WX's cores. AMD claims Dynamic Local Mode's background service will improve over time as the company characterizes more applications. Still, we see this as a bandage for the inconvenience of having to change modes and reboot your PC.

The competition isn't sitting still. Intel continues to get more competitive as it tries winning back the hearts and minds of enthusiasts. Its Basin Falls/Skylake-X Refresh processors should arrive next month. They'll still top out at 18 cores and 36 threads, and undoubtedly bear the company's notoriously high prices. But they are also rumored to employ Indium solder for improved thermal dissipation. That could make the new CPUs more attractive to tuners. It'd also be nice to see higher multi-core Turbo Boost bins that bolster performance in lightly-threaded workloads. Intel is also eschewing the practice of disabling PCIe lanes on less expensive HEDT processors, which is obviously a response to AMD’s practice of exposing all 60 PCIe lanes on every Threadripper model.

For now, AMD's Ryzen Threadripper 2950X, and as an extension the 2920X, offer the best value for high-end desktop PCs. Much like the Threadripper 2990WX, the 2970WX we tested today is a niche product for professionals seeking very specific capabilities. The competitive landscape is changing though, so we'd recommend waiting this one out.

MORE: Best CPUs

MORE: All CPUs Content

Cooler Master Cosmos C700M Review: A Colossal Success

Steven Lynch — Sun, 14 Oct 2018 13:00:00 +0000

Features & Specifications

Buyers are too often forced to compromise when choosing a new chassis, but adages such as "you can't have everything" aren't necessarily the case when it comes to the Cooler Master Cosmos C700M. In fact, not only does this chassis have just about everything a person could want in a full tower chassis, it also offers special features that allow the end user to reconfigure the case at a later date for a completely new look. As one might expect, a premium chassis such as this comes at a price. The Cooler Master Cosmos retails for $440.

Specifications

Type	Full-Tower ATX
Motherboard Support	Mini-ITX, Micro-ATX, ATX, (E-ATX support up to 12 x 10.7 inches)
Dimensions (HxWxD)	25.6 x 12.1 x 25.5 inches (651 x 306 x 650mm)
Space Above Motherboard	3 inches (78mm)
Card Length	19.3 inches (490mm) without HDD cage
CPU Cooler Height	7.8 inches (198mm)
Power Supply Length	Standard ATX PS2-Style PSU
Weight	52.5lbs (23.8kg)
External Bays	1x 5.25 inches
Internal Bays	5x 3.5 inches4x 2.5 inches (+ 5x converted from 3.5)
Card Slots	8
Ports/Jacks	1x USB 3.1 Type-C4x USB 3.01x audio/mic jacks
Other	ARGB control button, fan speed switch
Front Fans	3x 140mm (up to 3x 120/140mm)
Rear Fans	1x 140mm (up to 1x 140mm or 120mm)
Top Fans	✗ (up to 3x 120/140mm)
Bottom Fans	✗ (up to 2x 120/140mm)
Side Fans	✗
Dampening	✗
Warranty	Two years, limited

Exterior

The Cosmos C700M is colossal in every sense of the word. Measuring a massive 651 x 306 x 650mm (HxWxD) and absolutely crushing the scales at over 52lbs (empty), this ARGB-lit chassis will no doubt be the centerpiece of any room it occupies. The frame of the chassis is steel, and the exterior is a mix of plastic, aluminum and tempered glass. To the casual observer, the Cosmos C700M bears a striking resemblance to the Cosmos C700P but, upon closer inspection, you will notice an array of improvements both cosmetic and functional on this new chassis.

Once such feature is the ability to reconfigure the internal components from a standard or inverted layout to a chimney effect that rotates your components 90 degrees and vents heat out the top of the chassis instead of the back. We will talk about that in detail later in this review.

The plastic top panel of the C700M is covered with with a very stylistic brushed aluminum insert. Two sturdy cast aluminum bars run the entire length of the top, allowing you to lift the case from above. Much like Cooler Master's own MasterCase H500M and the older Cosmos C700P mentioned earlier, the slanted area between the top panel and the front panel is home to four USB 3.0 and one USB 3.1 Type-C ports, headphone and microphone jacks, a hard drive activity LED, a fan control switch, an ARGB lighting controller and a reset and power button. Note that if you plan on utilizing all four USB 3.0 ports, you will need a motherboard with dual 3.0 headers or an adapter cable.

The front panel matches the top with its stylish brushed aluminum cover mated to the plastic front panel. Behind the front panel are a plastic mesh filter, a mounting location for a 5.25-inch optical drive or drive bay accessory and three 140mm fans.

Behind the lower portion of the integrated center panel are six metal pins that are mated to six corresponding contact points in the frame of the chassis. This system allows the embedded ARGB lighting to work with the rest of the system without the need for wires.

Ventilation slots run from top to bottom along both sides of the front panel with dual-layer black metal mesh (a coarse perforated metal backed by fine metal screening material) built into the panel itself.

The center portion of the front panel features a slotted hinge at the base that allows the panel to tilt outwards to allow access to the optical drive slot without having to fully remove the front panel.

The lightly tinted full-coverage tempered glass side panel was created using state-of-the-art technology to add the curved shape. The tempered glass panel, as well as the steel panel on the opposite side, are hinged in the rear.

Both panels are interchangeable and can easily be removed by lifting the panels off the steel spring-loaded brackets. This is very handy when working on your system or installing components. Powerful magnets located in the leading edge keep both panels securely in place when closed.

In its default configuration, the rear of the chassis features eight standard expansion card slots, plus an opening to route cables for vertically mounted video cards, an opening for a bottom-mounted PSU and an exhaust-fan mounting location that supports both 120mm and 140mm fans. The exhaust-fan mounting location is equipped with slotted screw holes that allow you to adjust the position of the fan to fine-tune airflow or make room for system components.

There is also a removable plastic frame on the rear held in place by magnets. It should be noted that although the magnets do a great job holding the rear panel in place, the panel will pop off fairly easily if you attempt to use it as a handhold when lifting or moving the chassis.

A large plastic mesh filter that covers the entire bottom of the chassis is removable from the front. Two large flat cast aluminum bars (that match the top of the chassis) with rubber pads keep the case almost an inch off the ground.

In addition to the dual-layer black mesh lining, the openings in the front panel and large removable plastic mesh filters that cover the length of the case's front and bottom, the entire top panel is filtered. Both the front and bottom filters are removable from the front, so there is no need to move this massive chassis for maintenance and cleaning.

MORE: Best Cases

MORE: All Case Content

Hardware Installation & Test Configuration

Included accessories for the Cosmos C700M are various screws, zip ties, an extra hard drive rack, a 4-pin RGB lighting connector and a 400mm GPU riser cable. The box also contains a printed manual and alternative back panel.

The interior of this chassis is where things get really interesting. The Cosmos C700M is designed in such a way that the chassis can be configured in one of three different layouts. In addition to the traditional layout we are all accustomed to, the internals of this chassis can be rotated 90 degrees for a chimney effect or inverted for those that like your cases positioned to your left.

The interior of the chassis is painted black and has three large cable pass-through holes with rubber grommets to the right of the motherboard area. There are two slots, one at the top of the motherboard tray and one at the bottom for cable management as well. Also, regardless of orientation, this chassis can be outfitted with CPU air coolers up to 198mm in height.

The slot at the upper edge of the motherboard tray is specifically designed for routing fan cables and the ATX12V/EPS12V power cable. A large hole in the motherboard mounting plate, behind the CPU socket area, facilitates heat sink changes without removing the motherboard.

In standard and inverted modes, the eight expansion slots will accommodate long graphics cards, up to 320mm (490mm with hard drive racks removed/repositioned), even in multi-GPU configurations. If the Cosmos C700M is configured with the motherboard in the "chimney" orientation, video cards installed directly into the motherboard's PCI-e slots are limited to 305mm in length without removing the PSU shroud.

The adjustable GPU bracket allows end users to show off their GPUs by mounting it vertically, or at an angle between 0 and 90 degrees, on the PSU shroud. You can also mount the GPU bracket so that your video card is orientated top to bottom in the front of the chassis closest to the 140mm intake fans. Cooler Master includes a 400mm PCIe x16 riser cable.

With all of that out of the way, let's talk about the process of switching the layout in this chassis because, to be frank, it is not a task to be taken lightly. It is time-consuming, labor-intensive and there are a lot of screws. So, so many screws. The good thing is that, after you've done it once, it's fairly easy to do it again.

We highly recommend taking your time and weighing your choices carefully before tearing into this case. Enthusiast and experienced modders shouldn't have any issues with the swap, but novice users could find the task daunting. That said, regardless of your choice of layout, the Cosmos C700M's cavernous interior gives end users the freedom to build until their heart’s content.

There are mounting locations for a total of five 3.5-inch hard drives and nine 2.5-inch drives (five converted from 3.5-inch). Two drive racks are mounted under the PSU tunnel and two in the main compartment of the case. There is also an extra drive rack included in the accessories box. These removable drive racks can be relocated to any of the several open slots in the main compartment of this chassis or removed completely.

Two of the four 2.5-inch drive caddies are attached to the motherboard tray in the main compartment, one is attached to the top of the PSU tunnel and one behind the motherboard tray. All drive caddies, with the exception of the stamped steel unit behind the motherboard tray, are made of metal and come with rubber grommets for vibration dampening. These 2.5-inch mounting pads also double as mounting locations for water pumps and reservoirs.

The cable management area directly behind the motherboard tray can accommodate just about any system you can throw at it, from mild to wild. Cooler Master says the cable management design was inspired by server racks. Regardless of inspiration, the overall design is a breath of fresh air. The plastic cable strips organize the cable by using channels, and the Velcro retention straps keep everything in their place. The case also includes a convenient cable cover made of stamped steel that keeps unsightly wires hidden from view.

Cooling

The Cosmos C700M supports up to nine 120/140mm fans in total. It ships with three PWM 140mm intake fans in front and a single PWM 140mm exhaust fan. Fan speeds can be set to low (600rpm), medium (900rpm) and high (1,200rpm) via the fan speed switch on the front of the case.

The main compartment can easily accommodate one or more custom cooling loops, multiple reservoirs and pumps, all-in-one coolers or a mixture. Radiators and AIO coolers up to 420mm can be installed in the front and top of the chassis simultaneously with the optical drive bay removed.

As with many large full-tower ATX chassis, even though you can physically mount most AIO coolers in the front of this case because of the fixed length of the tubing, some coolers might not be able to reach the CPU socket. Despite that limitation, these locations are ideal for mounting radiators for water-cooled graphics cards.

The exhaust fan mounting location can be fitted with 120/140mm coolers in single and dual-fan configurations. You can also mount coolers and radiators up to 240mm in the bottom of the chassis.

And with the case's large size, every power supply we tested, regardless of layout, from the smaller EVGA 450 BT all the way up to larger units like the Corsair HX1200i, fit without a problem.

Test Configuration

Drivers & Settings

Chipset	Intel INF 10.1.1.42
CPU	4GHz (40x 100MHz) @ 1.1V core
Motherboard	Firmware 1.10 3/2/2018
RAM	14-14-14-34
Graphics	Maximum Fan for Thermal Tests \| NVIDIA GeForce 398.36 WHQL Game Ready Driver

MORE: Best Cases

MORE: All Case Content

Benchmark Results & Conclusion

be quiet! Dark Base 700

Cougar Panzer EVO RGB

NZXT H700i

We tested the Cosmos C700M with our new Intel i9-7900X test platform. For comparison, we brought in the Cougar Panzer EVO RGB, the NZXT H700i and the be quiet! Dark Base 700 to give you a solid idea where this case stands against other cases that are of similar size and feature sets.

Given that this chassis is equipped with three 140mm intake fans feeding a single 140mm exhaust fan, it should come as no surprise that its thermal performance easily bested the rest. At just 57 degrees Celsius under load and 47 degrees Celsius at idle, the C700M is in a league of its own. Even with the fans set to low speed (600 rpm), our overclocked Intel i9-7900X processor running at 4.0GHz peaked at 58 degrees Celsius over the ambient temperature of 25 degrees Celsius, putting it neck-and-neck with the Panzer EVO RGB. GPU temperatures maxed out at 48 degrees Celsius over the ambient room temperature, on par with the other cases we tested against.

For audio testing, we recorded sound level readings with two off-the-shelf dB meters from two different angles. For our first round of testing, we set the case fans to their maximum speed of 1,200 rpm. Our second audio test was run with the case fans on their lowest setting, 600 rpm. The tempered glass side panel and the solid case construction kept sound output to a minimum, even on the highest setting. The C700M generated noise levels on par with the Dark Base 700 and the Panzer EVO RGB. Both of those are also extremely quiet chassis.

Note that our Performance Value chart above doesn’t take features into account. Even though this chassis has great thermal and acoustic performance, it is hard to justify the $440 price tag based on performance alone. This is where the Cosmos C700M's impressive list of features, unparalleled configurability and unique modular design should be factored into the equation.

Bottom Line

We are the first to admit that this chassis' price tag might be too high for some people's taste, but we also realize that this is a specialized, high-end chassis specifically built for enthusiasts willing to spend their hard earned money to have the best. And make no mistake about it, the Cooler Master Cosmos C700M is one of the best.

MORE: Best Cases

MORE: All Case Content

Lian Li Lancool One Digital Case Review: Old School Style With RGB Flare

Steven Lynch — Sun, 07 Oct 2018 13:00:00 +0000

Features & Specifications

Combining old school styling and just a touch of modern flare, the Lian Li Lancool One Digital ($90) can accommodate a variety of system builds. Its built-in addressable RGB lighting is compatible with most major motherboard vendors, and it even has USB 3.1 Gen 2 Type-C. That said, not everything is addressable rainbows and sunshine. The lack of thumbscrews, flimsy front panel and grommets that refuse to stay in their hole keep this case from achieving true greatness.

Specifications

Type	Mid-Tower ATX
Motherboard Support	Mini-ITX, Micro-ATX, ATX, (E-ATX support up to 12 x 10.7 inches)
Dimensions (HxWxD)	18.6 x 8.6 x 17.7 inches (472 x 220 x 450mm)
Space Above Motherboard	1.5 inches (38.1mm)
Card Length	16.5 inches (420mm)
CPU Cooler Height	6.9 inches (175mm)
Power Supply Length	8.1 inches (224mm)
Weight	18.2lbs (8.25kg)
External Bays	✗
Internal Bays	2x 3.5 inches4x 2.5 inches
Card Slots	7 + 2 vertical
Ports/Jacks	1x USB 3.1 Type-C2x USB 3.01x audio jack1x mic jack1x RGB switch
Other	Tempered glass side panel, addressable RGB (ARGB) lighting
Front Fans	1x 120mm (Up to 3x 120mm / 2x 140mm max)
Rear Fans	1x 120mm
Top Fans	✗ (Up to 3x 120mm / 2x 140mm max)
Bottom Fans	✗ (Up to 2x 120mm max)
Side Fans	✗
Dampening	✗
Warranty	One year, limited

Exterior

Measuring 472 x 220 x 450mm (L x W x H), this steel chassis is painted black inside and out, features a tempered glass side and brushed-aluminum accent panels and weighs in at just over 18lbs.

The top of the Lancool One Digital is covered by a plastic/metal mesh panel that you can easily remove from the rear. Below it are mounting locations for three 120mm or two 140mm fans. At the front edge are a pair of USB 3.0 ports, headphone and microphone jacks, an HDD LED, an LED controller button, a power and reset button and a USB 3.1 Type-C connector.

Thin sheets of aluminum with beveled edges are affixed to the plastic front panel. Vertical vents line both sides of the fascia, extending from the top to the bottom of the chassis. A gap in the front panel between the two aluminum panels serves as an air intake port and also allows you to see the addressable RGB lighting. Removing the front panel is a simple matter of grabbing the lower edge and pulling out and away from the chassis.

The rear of the chassis is standard fare for an ATX mid-tower. It has expansion slots, an opening for a bottom-mounted PSU and an exhaust-fan mount fitted with a 120mm fan. That location features adjustable screw holes that lets you position the fan higher or lower to fine-tune airflow or make room for system components.

On the bottom of the chassis you will find a 6-inch plastic mesh filter covering the PSU's fan opening that is removable from the rear. The four large, rectangular, rubber-coated feet elevate the case approximately 0.5 inch.

The full-cover, tempered glass panel is slightly tinted. Lian Li decided against conventional rubber-coated locating pins and thumbscrews, opting instead for a tempered glass panel bonded to a metal frame (much like Cooler Master's MasterCase H500P and the NZXT H500i). The entire assembly attaches to the frame via thumbscrews in the rear. Oddly enough, the solid metal panel on the opposite side is secured with the type of thumbscrews typically used with tempered glass panels.

The fan filtration system consists of a mix of different types of filters and does a good job preventing most dirt and dust particles from entering your system. The lift-off metal mesh filter on the top of the case is attached by magnets embedded in the edge of the filter. The full-cover filter in the front of the chassis also attaches to the frame via a magnetic seal. The nylon filter covering the PSU mounting location slides in and out from the rear of the case. Considering the rear filter measures just 6-inches-long, servicing this filter shouldn't be an issue.

MORE: Best Cases

MORE: All Case Content

Hardware Installation & Test Configuration

Secured in one of the two open 3.5-inch hard drive bays is a small white box filled with various screws in individual baggies and a handful of zip ties. The case doesn't come with a physical manual, but you can download one from the company website by scanning a QR code on the accessory box.

The main compartment of this chassis is bland--pedestrian, if you will. There are no optical drive bays or hard drive racks, just a PSU tunnel and two 140mm fans. The overall look is extremely clean, and there is more than enough room for a variety of different system builds.

Although Lian Li claims the Lancool One Digital supports up to EATX motherboards, the truth of the matter is that since a true EATX motherboard measure 13 inches deep by 12 inches tall, there are not enough standoffs to install a motherboard of that size. Attempting to use a full EATX motherboard would leave over 3 inches of your motherboard unsupported (and likely shorting against the raised forward portion of the motherboard tray).

The interior of this case is painted black to match the exterior. There are three large cable pass-through holes with rubber grommets in the motherboard tray for cable management. The rubber grommets, while functional, tend to fall out of the holes in the motherboard tray at the slightest touch. We found it best to run all our cables and make sure everything fit before reinstalling the grommets. The holes at the upper edge of the motherboard tray are specifically designed for routing fan cables and the ATX12V/EPS12V power cable.

A large hole in the motherboard mounting plate, behind the CPU socket area, facilitates heat sink changes without removing the motherboard. This chassis can be outfitted with CPU coolers up to 175mm and graphics cards up to 420mm. The Lancool One Digital has a total of nine expansion slots, seven traditional and two vertical, and can accommodate multi-GPU set-ups up to 420mm in length and 150mm wide.

The Lancool One Digital can accommodate up to two 3.5-inch hard drives and a total of four 2.5-inch hard drives or SSDs. You can mount the two 2.5-inch drive caddies to the top of the PSU tunnel or behind the motherboard tray. Personally, we like to keep our drives out of sight for a cleaner look by utilizing the mounting slots behind the motherboard.

We tested power supplies up to 220mm in length without any issues or the need to move the hard drive racks.

Cooling

The Lancool One Digital ships with two 120mm fans. While that number of fans is fairly common, it would have been nice to see at least one more 120mm intake fan. We'll see if this is a detriment to the case's thermal performance on the next page.

You can mount all-in-one liquid coolers and radiators up to 360mm to the front and top of the chassis' main compartment (although at 38mm motherboard clearance, the top lacks the 52mm vertical space needed by most radiator/fan combos). The front of the chassis can easily support radiators sandwiched between fans in a push-pull configuration. Keep in mind that the thickness of your radiator/all-in-one cooler and fans will subtract from the overall length of GPU you will be able to use.

Due to space limitations in the top of the case, those wanting to install radiators and all-in-one coolers in a push-pull configuration will need to utilize the outermost mounting slots to avoid clearance issues with motherboard components. The exhaust fan mounting location can be fitted with 120mm coolers in single- and dual-fan configurations.

Test Configuration

Drivers & Settings

Chipset	Intel INF 10.1.1.42
CPU	4GHz (40x 100MHz) @ 1.1V core
Motherboard	Firmware 1.10 3/2/2018
RAM	14-14-14-34
Graphics	Maximum Fan for Thermal Tests \| NVIDIA GeForce 398.36 WHQL Game Ready Driver

MORE:
Best Cases

MORE: All Case Content

Benchmark Results & Conclusion

We tested the Lian Li Lancool One Digital with our new Intel i9-7900X test platform. For comparison, we used the be Quiet! Silent Base 601, be quiet! Dark Base 700, Cougar Panzer EVO RGB and NZXT H700i to give you a good idea where this case stands against other cases that are of similar size and feature sets.

be quiet! Silent Base 601

be quiet! Dark Base 700

Cougar Panzer EVO RGB

I wish I had something more exciting to say about this chassis' thermal performance, but our Intel i9-7900X processor running at 4GHz turned in numbers almost identical to all the other cases used for comparison.

The same goes for our GPU results. The most impressive aspect is that the LanCool One Digital kept up with the competition while utilizing just two 120mm fans.

As always, we took sound level readings with two off-the-shelf dB meters from two different angles. With the fan speed at default speed (1,000 rpm), our test system registered 29.9dBA at idle, putting it right in the middle of the pack. Under load, sound output increased to 32.8dBA, almost identical to that of the other comparison cases.

This chassis' relatively low $90 asking price gave the Lian Li Lancool One Digital a favorable value score in our performance-to-price comparison. There is also a standard RGB version of this chassis, dubbed the Lancool One, available for those that have motherboards that do not support addressable RGB.

MORE: Best Cases

MORE: All Case Content

Cougar Conquer Essence Case Review: A Fanless Dust Trap?

Steven Lynch — Wed, 03 Oct 2018 13:30:00 +0000

Features & Specifications

Although the Cougar Conquer Essence is sleek, stylish and has very good thermal performance, there are just too many things wrong with this chassis to recommend it at this price.

Looking for an open-air micro-ATX case that is stylish, easy to mod and stands out from the crowd right out of the box? Cougar's Conquer Essence might be right up your alley, if you can stomach all the dirt, dust and pet hair that will likely fall into the case and the price tag.

This chassis is essentially a smaller version of the Cougar Conquer chassis. The design is similar, but not only is the Conquer Essence smaller, a number of features have been removed, retooled and / or completely redesigned from the Conquer. The overall look is much more spartan than the full-sized Conquer chassis.

Specifications

Type	Mini-Tower ATX
Motherboard Support	Mini-ITX, Micro-ATX
Dimensions (HxWxD)	18.9 x 9.2 x 20.5 inches (480 x 233 x 522mm)
Space Above Motherboard	3 inches (76mm)
Card Length	13.8 inches (350mm)
CPU Cooler Height	6.3 inches (160mm)
Power Supply Length	7.9 inches (200mm)
Weight	22lbs (10.0kg)
External Bays	✗
Internal Bays	2x 3.5 inches6x 2.5 inches
Card Slots	5x vertical
Ports/Jacks	2x USB1x audio jack1x mic jack
Other	Tempered glass side panels
Front Fans	✗ (2x 120mm max)
Rear Fans	✗ (2x 120mm max)
Top Fans	✗ (2x 120mm max)
Bottom Fans	✗
Side Fans	✗
Dampening	✗
Warranty	One year, limited

Exterior

The entire chassis, aside from the tempered glass side panels, is constructed out of an aluminum alloy and is powder-coated black inside and out. The trapezoid-shaped chassis measures 480 x 233 x 522mm (HxWxD) and tips the scales at just over 22lbs. The frame of the chassis appears to be made of a single large sheet of aluminum that is laser-cut and folded into shape.

The top of the chassis is bare except for two open 120mm fan holes. The area between the top panel and the front panel is slanted at a 45-degree angle. This space is home to two USB 3.0 ports, headphone and microphone jacks and reset and power buttons

The removable front panel consists of a solid piece of aluminum that has large unfiltered ventilation holes lining both sides. There are large openings located at the top and bottom of the front panel as well. These air-intake holes provide a path for fresh air to be drawn into the chassis. Directly behind the front face are mounting locations for two 120mm fans. The front panel is held in place by four Philips head screws.

The full-coverage tinted tempered glass side panels are held in place by large thumbscrews that thread directly into metal rods, much the same as the side panels on the original Conquer and the Thermaltake Core P3 and P5 cases. Bear in mind that although rubber grommets line the holes in the tempered glass, there is nothing to hold the panel in place and prevent the glass against accidental drops. We highly recommend laying this case on its side when removing the tempered glass panels.

On a side note, there was a large amount of what seemed to be melted plastic all over both side panels on our unit. A razor blade and a bit of rubbing alcohol made short work of the mess, but was still disappointing that this wasn't caught at the factory.

While we are on the subject, we also noticed that the bracket for the five expansion cards was severely bent. Although we were able to bend it back in place, doing so caused a small tear in the aluminum alloy material at the base of the bracket.

Directly below the expansion card bracket is the mounting location for your power supply. The Conquer Essence does not feature a PSU tunnel.

Since this is an open-air chassis, there is no filtration system in place. In fact, even though the company includes a pair of fan grills, there is nothing at all preventing dust and debris from falling into your system or spilling directly onto your exposed hardware.

MORE: Best Cases

MORE: All Case Content

Hardware Installation & Test Configuration

Included accessories for the Cougar Conquer Essence are various screws, 3M cable ties and a pair of fan grills. The box also contains a printed manual.

There are seven cable pass-through holes in the motherboard tray for cable management: three large rectangular openings and two larger holes in the bottom of the motherboard tray. Additionally, there are two small holes within the top of the motherboard tray, one designed specifically for the 12v motherboard power cable. You'll also find a large hole in the motherboard mounting plate behind the CPU socket area to facilitate heat sink changes without removing the motherboard.

Again, because of the open-air styling of the the interior of this chassis, it is very similar to that of the Cougar Conquer and comparable to the Thermaltake Core P3 and P5 cases.

The five expansion slots will accommodate graphics cards up to 350mm, even in multi-GPU configurations. CPU air coolers up to 160mm can be fitted in this case, making coolers such as the Cooler Master MasterAir MA410M and the be quiet! Dark Rock 4 solid choices for this chassis.

As you can see from the photos, care must be taken when installing the PSU because you could easily run into clearance issues, depending on the length of your PSU and choice of motherboard. Our Corsair HX1200i PSU is 200mm long, and it barely fit into this case with our ASRock X299M Extreme4 micro-ATX motherboard.

Two of the three hard drive mounting locations behind the motherboard tray can support both 2.5-inch and 3.5-inch drives. One of the three supports 2.5-inch drives only. The two mounting locations in the base of the case can be fitted with both 2.5-inch and 3.5-inch drives.

Due to the open nature of this case, there isn't really anywhere good to hide cables, making cable management more difficult than usual. Even if a modular power supply is used, careful planning and routing is still a must.

When it comes to cooling, we were a bit disappointed that the Conquer Essence doesn't ship with any fans from the factory. Obviously, this can be remedied by the end user, and enthusiasts prefer to install fans of their choosing anyway. But it would have been nice to see at least a couple fans included with a case that costs $220. In total, this chassis can be equipped with up to four 120mm fans.

Radiators and all-in-one coolers up to 240mm can be mounted in the top and front of this case. Both mounting locations can easily support radiators up to 240mm sandwiched between fans in a push-pull configuration.

Test Configuration

Drivers & Settings

Chipset	Intel INF 10.1.1.42
CPU	4GHz (40x 100MHz) @ 1.1V core
Motherboard	Firmware 1.10 3/2/2018
RAM	14-14-14-34
Graphics	Maximum Fan for Thermal Tests \| NVIDIA GeForce 398.36 WHQL Game Ready Driver

MORE: Best Cases

MORE: All Case Content

Benchmark Results & Conclusion

We tested the Cougar Conquer Essence with our new Intel i9-7900X test platform. For comparison, we brought in the Cougar Panzer EVO RGB, the NZXT H700i, the be quiet! Silent Base 601 and be quiet! Dark Base 700 to give you a solid idea of where this case stands against other cases that are of similar size and feature sets.

be quiet! Dark Base 700

be quiet! Silent Base 601

Cougar Panzer EVO RGB

The lack of airflow in this chassis is due to the fact that it ships with no fans. Although we are fully aware that enthusiasts prefer to add their own fans, we still think Cougar should have included one or two 120mm fans to cover the ugly empty holes in the top and to provide some airflow into the chassis.

Temperatures were good because, let's face it, this chassis is one step away from being an open-air test bench. Our beastly overclocked Intel i9-7900X processor running at 4.0GHz peaked at 58 degrees Celsius over the ambient temperature of 25 degrees Celsius. GPU temperatures were also impressive at 48 degrees Celsius over the ambient room temperature. The addition of fans to this chassis would mean even better temperatures, but at the cost of elevated noise levels.

Speaking of noise levels, we recorded sound level readings with two off-the-shelf dB meters from two different angles. Normally, one of the benefits of a side panel made of tempered glass is its ability to reflect sound back into the chassis. Unfortunately, the open layout of this chassis allows almost all the noise your system's components make to escape the case. At idle, the system was fairly quiet at just 29.6dBA. Our test results under load were a totally different story. The fan noise from our GPU and all-in-one cooler hit an annoyingly high 38.8dBA, making it far and away one of the loudest cases we've ever tested.

Our Performance Value results here can be a bit misleading. This is due to the fact that those of you buying this chassis will have to add cooling fans and/or fans attached to radiators and all-in-one cooling solutions. Installing fans would add to the overall cost of the case, and that in turn would affect the Performance Value rating. Also, even though fans would improve overall thermal performance, it would certainly introduce more system noise and negatively impact our acoustic-efficiency rating.

Even if you carry over fans from a previous build to the Cougar Conquer Essence, there is still the over-$200 asking price to contend with. We don't have a problem with expensive cases, especially when they bring something completely new to the table or a host of new features. But the Conquer Essence, as unique and interesting as it may be, makes it extremely difficult to justify its asking price given that it lacks basics, such as fans, filters and USB 3.1 Type-C.

MORE: Best Cases

MORE: All Case Content

be quiet! Silent Base 601 Case Review: RGB-Free Tranquility

Steven Lynch — Fri, 28 Sep 2018 01:00:00 +0000

Features & Specifications

If bling is not your thing and you are looking for a new chassis for your next system build with an emphasis on quiet operation, the $130 be quiet! Silent Base 601 is certainly worth your consideration. One of the best pc cases around, this chassis doesn't feature any tempered glass or RGB lighting, instead opting for a more subdued look with a focus on acoustic performance.

Although the company claims that the Silent Base 601 is a reworked version of the original Silent Base 600, the truth of the matter is that this is almost a completely different design. The Silent Base 601 is a far more traditional chassis and a bit of a throwback to the basics.

Specifications

Type	Mid-Tower ATX
Motherboard Support	Mini-ITX, Micro-ATX; ATX (E-ATX support up to 12 x 10.7 inches)
Dimensions (HxWxD)	20.9 x 9.4 x 20.2 inches (532 x 240 x 514mm)
Space Above Motherboard	1.5 inches (39mm)
Card Length	17.6 inches (449mm) without HDD cage
CPU Cooler Height	7.3 inches (185mm)
Power Supply Format	Standard ATX PS2-Style PSU
Weight	23lbs (10.4kg)
External Bays	✗
Internal Bays	3x 3.5 inches (+ 4x optional)6x 2.5 inches (+ 8x optional)
Card Slots	7 + 2 vertical
Ports/Jacks	12x USB 3.01x USB; audio/mic jacks; fan speed switch
Other	Fan controller
Front Fans	1x 140mm (Up to 3x 120mm or 2x 140mm)
Rear Fans	1x 140mm (Up to 1x 140mm or 120mm)
Top Fans	✗ (Up to 3x 120mm)
Bottom Fans	✗
Side Fans	✗
Dampening	10mm acoustic foam on 4 panels
Warranty	Three year, limited

Exterior

There are a few aesthetic similarities between the two chassis; the overall size of the case, the all-black exterior with vents surrounded by accent colors (orange, silver and black) running down both sides of the front face and the oddly shaped ventilation slots in the rear portion of the top panel. Aside from that, the Silent Base 601 is an entirely different beast.

The chassis is constructed of plastic and steel that is painted black (inside and out) and features optional accent colors around the intake vents in the front of the case (available in black, orange and silver). It measures 532 x 240 x 514mm (L x W x H) and tips the scales at just over 23lbs. The Silent Base 601 comes in two versions; one with a tempered glass side panel and the base model you see here today.

The redesigned top panel is different from the original Silent Base 600. It's totally flat with the ventilation slots moved all the way to the rear of the panel. Directly under the panel are mounting locations for three 120mm or 140mm fans on a slide-out rack. Also, the front panel ports have all been relocated from the front panel to the top of the case. Here ,you will find one USB 2.0 and two USB 3.0 ports, headphone and microphone jacks, a hard drive activity LED, a reset and power button and a fan control switch. The power button is illuminated to match the accent color of the case.

The front of the case is a large plastic panel with a thin layer of aluminum affixed to the outside.

The reverse side of the lift-off front panel is lined with a 10mm-thick layer of sound damping material. Beyond that you will find a large, removable plastic mesh filter covering mounting locations for three 120 / 140mm fans.

Vertical vents line both sides of the front fascia, extending from the top to bottom of the chassis. It should be noted that the front panel comes off very easy. It is a simple matter of grabbing the lower edge and lifting up. Care must be taken when transporting your system, however, because you can easily drop your system or rip the front panel off if you inadvertently grab the lower edge of the front panel instead of the bottom of the case when moving your PC.

In the rear of the chassis you'll find an opening for a bottom-mounted PSU, seven expansion card slots, a standard motherboard I/O area and an exhaust fan mounting location that supports both 120mm and 140mm fans. The exhaust fan mounting location has slotted screw holes that let you adjust the position of the fan to fine-tune airflow or make room for system components.

The bottom of the chassis is covered by a 20.9-inch plastic mesh filter that is removable from the front. The filter covers the metal mesh area under the hard drive cage and the power supply fan opening. The four large, rectangular, rubber-coated feet elevate the case approximately 0.5 inch.

Both side panels on the 601 feature a push-button quick release mechanism that allows you instant access to the inside of your case. Removing the side panels is a simple matter of lifting the panel off once the release button has been pressed. Each side panel is lined with the same 10mm-thick sound dampening material seen behind the front panel.

The fan-filtration system on the Silent Base 601 consists of two full-length, removable nylon filters covering the front and bottom of the chassis. Thankfully, both filters are accessible and removable from the front of the case, meaning you no longer have to move your entire system to remove and clean your filters.

MORE: Best Cases

MORE: All Case Content

Hardware Installation & Test Configuration

Included accessories for the Silent Base 601 are various screws, zip ties, an extra removable 2.5-inch / 3.5-inch drive bay. The box also contains a printed manual.

Painted black to match the exterior, the interior of the Silent Base 601 is very spacious and features a clutter-free design. There are three large cable pass-through holes with rubber grommets (with the same color as the exterior accent trim) in the motherboard tray for cable management. The hole at the upper left-hand edge of the motherboard tray is specifically designed for routing fan cables and the ATX12V / EPS12V power cable.

The seven (plus two vertical) expansion slots will accommodate graphics cards up to 449mm, even in multi-GPU configurations. The vertical VGA GPU slots allow end users to install a PCIe x16 riser cable to show off their bright and shiny GPUs by mounting it vertically, but the riser cable is sold separately. You can use CPU air coolers up to 185mm tall.

The spec sheet lists support for "standard ATX PS2-Style PSUs." During testing, we fitted this chassis with power supplies up to 220mm in length without any issues. Those needing even more space have the option of removing the hard drive cage.

There are two mounting locations for 2.5-inch / 3.5-inch hard drives under the power supply tunnel closest to the front of the chassis. Additionally, you can mount up to six 2.5-inch hard drives under the PSU tunnel, behind the motherboard tray and in the hard drive cage included with the accessories.

Those needing more hard drive mounting locations can add four 3.5-inch or eight 2.5-inch drives by purchasing optional hard drive cages from the manufacturer. All drive caddies are made of metal and come with rubber grommets for vibration dampening.

In the past, the company included plastic block-off panels to cover the holes in the motherboard when not filled with an optional hard drive case. No block-off plates come with the Silent Base 601; since you can't see through the metal side panel, there is no need.

Even though this chassis is a targeting the quiet computing crowd, it's still a bit odd to only include a single 140mm Pure Wings 2 front intake fan. There is another 140mm Pure Wings 2 exhaust fan behind the CPU socket area. We'll see if this fan combination has an adverse impact on cooling later in this review.

There is a three speed fan control switch on the front of the chassis that allows you to adjust the fan speeds from 500 to 1,000 rpm.

Those of you that prefer liquid cooling will appreciate the fact that radiators and all-in-one coolers up to 360mm can be mounted in the top and front of this case. The mounting location in the front of the chassis can easily support radiators up to 360mm (280mm with the PSU tunnel in place) sandwiched between fans in a push-pull configuration. Radiators over 280mm require the removal of the plastic panel insert in the top of the PSU tunnel for clearance.

There is a total of 39mm of clearance for radiators and coolers in the space above the motherboard. Much like the removable fan/radiator brackets we've seen in other chassis, such as the Corsair Obsidian Series 500D and the Phanteks Enthoo Evolv, installing radiators and fans is a simple matter of removing two screws and sliding the unit out and away from the chassis. Once you've attached your cooling components to the bracket, you'll slide the entire setup back into your system and reinstall the screws. Generally speaking, if you want to install radiators and all-in-one coolers in a push-pull configuration in this location, you will need to utilize the outermost mounting slots to avoid clearance issues with motherboard components. The rear of the chassis supports 120 / 140mm radiators or all-in-one coolers in the exhaust fan location over the CPU socket area in single and dual-fan configurations.

Note that the sound dampening material and the limited airflow provided by the six slots in the top make this a less-than-ideal mounting location for dual / triple fan radiators and all-in-one coolers.

Test Configuration

Drivers & Settings

Chipset	Intel INF 10.1.1.42
CPU	4GHz (40x 100MHz) @ 1.1V core
Motherboard	Firmware 1.10 3/2/2018
RAM	14-14-14-34
Graphics	Maximum Fan for Thermal Tests \| NVIDIA GeForce 398.36 WHQL Game Ready Driver

MORE: Best Cases

MORE: All Case Content

Benchmark Results & Conclusion

We tested the Silent Base 601 with our new Intel i9-7900X test platform. For comparison, we brought in the Cougar Panzer EVO RGB, the NZXT H700i and the be quiet! Dark Base 700 to give you a solid idea where this case stands against others of similar size and feature sets. Thermal and audio tests were conducted with the fan speed switch set on low (500 rpm) and then on high (1,000 rpm).

be quiet! Dark Base 700

Cougar Panzer EVO RGB

NZXT H700i

CPU core temperatures on our Intel i9-7900X processor running at 4GHz ran a bit on the high side at 65 degrees Celsius over ambient with the fans set to low. This is no doubt due to the combination of having only one 140mm intake fan and the thick, 10mm, sound-deadening material. Flipping the fan switch to high resulted in a better CPU temperatures, but only by 2 degrees Celsius.

Out of the cases we used for comparison, the Silent Base 601 had the highest overall temperatures. We recommend to anyone running a high-end or overclocked system to add at least one 140mm intake fan.

GPU temps were also high at 54 degrees Celsius but, unlike our CPU temperatures, the graphics card temperature actually benefited a great deal from the boost in fan RPM and the close proximity of the intake fan. With the fan switch on high, GPU temps dropped by a full 5 degrees Celsius.

Now, it's time to see if running the fans on the high setting hurts acoustic performance.

As always, we took sound level readings with two off-the-shelf dB meters from two different angles. With the fan speed switch set on low (500 rpm), our test system registered just 28.7dBA at idle, easily besting the rest of the chassis used for comparison in this review. Under load, we were surprised at just how much sound escaped from the rear of the case. At 32.5dBA, the Silent Base 601 went from being the quietest case in the group to cranking out middle of the pack audio performance. The good news is that after the fan switch was set to high (1,00 rpm), there was only a 1dBA increase under load and no change at idle.

Bottom Line

Although thermal and audio performance could have been better, and the $130 price point ($140 for the windowed version) is a bit higher than other mid-tower ATX cases, the design, features and quality craftsmanship more than make up for those shortcomings. That being said, even though this chassis is less expensive than direct competitors, we believe a $120 price point would elevate this to a deal you just couldn't pass up. But still, Silent Base 601 is a very nice case, even at its current retail price.

MORE: Best Cases

MORE: All Case Content

Cooler Master MasterCase H500M Review: Expensive, but Excellent

Steven Lynch — Mon, 17 Sep 2018 13:00:00 +0000

Features & Specifications

Packed with premium features and top notch thermal and acoustic performance, there are a number of reasons to choose the Cooler Master MasterCase H500M. Its ability to accommodate a wide variety of system builds makes it easy to recommend to just about anyone, if they have the money.

With so many companies vying for your hard earned money, it seems like there isn't a brand out there that doesn't have a premium chassis packed with all the bells and whistles, including tempered glass panels, addressable RGB lighting, USB Type-C ports and water cooling options galore. One such company competing for your attention is Cooler Master with the latest entry in its MasterCase line of chassis.

At first glance, the MasterCase H500M ($199) looks very similar to the MasterCase H500P. But upon further inspection you will notice subtle design differences that set this chassis apart from Cooler Master's other mid-tower offerings. It's obvious that this design is aimed at gamers who want over-the-top looks and enthusiasts who want to showcase their system builds.

Specifications

Type	Mid-Tower ATX
Motherboard Support	Mini-ITX, Micro-ATX, ATX (E-ATX support up to 12 x 10.7 inches)
Dimensions (HxWxD)	21.5 x 9.8 x 21.4 inches (546 x 248 x 544mm)
Space Above Motherboard	1.5 + 1.5 inches (76.2mm)
Card Length	16.2 inches (412mm)
CPU Cooler Height	7.5 inches (190mm)
Power Supply Format	Standard ATX PS2-Style PSU
Weight	31.1 lbs (14.1kg)
External Bays	✗
Internal Bays	2x 3.5 inches6x 2.6 inches
Card Slots	7 + 2 vertical
Ports/Jacks	4x USB 3.01x USB Type-C (Gen 2), audio/mic jacks, RGB switch
Other	Tempered-glass side panels, included RGB / fan controller
Front Fans	2x 200mm ARGB fans
Rear Fans	1x 140mm
Top Fans	✗
Bottom Fans	✗
Side Fans	✗
Dampening	✗
Warranty	One year, limited

Exterior

This chassis is made of steel and plastic and is painted black (inside and out). It measures 546 x 248 x 544mm (LxWxH) and weighs just over 31lbs. A slightly modified version of the industrial crossbar design featured on the MasterCase H500 and H500P, the design is visually similar to Cougar’s Panzer line of cases.

The top panel is made up of a large, slightly tinted tempered glass panel. Directly under this panel is an area with mounting locations for three 120mm or 140mm fans, or two 200mm fans.

The revamped structured bar design features large ventilation holes running along both sides of the front and top panels with dual-layer black metal mesh (a coarse perforated metal, backed by fine metal screening material) embedded in the frame. These air-intake vents provide a path for fresh air to be drawn into the chassis.

The area between the top panel and the front panel is slanted at a 45° angle. This area is home to four USB 3.0 and one USB 3.1 Type-C ports, headphone and microphone jacks, a hard drive activity LED and a reset and power button. The reset button can also be used to manually control RGB lighting functions. Note that if you plan on utilizing all four USB 3.0 ports, you will need a motherboard with dual 3.0 headers or a 20-pin splitter cable.

In its default configuration the front fascia ships with a center panel made of the same dual-layer black metal mesh material embedded in the front and top panels. Cooler Master includes an optional tempered glass front panel for this chassis as well. Changing the center insert requires the removal of eight Phillips head screws and is a rather straightforward process. We conducted separate thermal and audio tests with each panel installed (page three).

Behind the mesh front panel you'll find two large 200mm RGB-lit fans. The design of the front panel requires removal of both tempered glass side panels to gain access to six clips (three on each side) holding the front in place.

The full-coverage tinted tempered glass side panels are held in place by a slotted-head cam latch mechanism in the middle of the upper edge. Turning the latch clockwise releases the panel from the frame. A metal edge along the panel's base does a great job holding the panel in place, helping to prevent the glass against accidental drops.

In the rear of the chassis, you'll find seven standard expansion-card slots (plus two vertical ones for video-card mounting), an opening for a bottom-mounted PSU and an exhaust-fan mounting location that supports both 120mm and 140mm fans. The exhaust-fan mounting location is equipped with slotted screw holes that allow you adjust the position of the fan to fine-tune airflow or make room for system components. There is a large plastic mesh filter covering the power-supply fan opening that is removable from the rear. The four large rectangle, rubber-coated feet keep the case just over .5 inches off the ground.

The fan-filtration system on the H500M is basic, but effective. The dual-layer black mesh we mentioned earlier filter the fan-mounting locations in the front and top of the chassis. Cleaning and maintenance is best done with the entire panel removed. A large, removable nylon filter covers the opening for the power-supply fan; you remove this filter from the case's rear. This, of course, requires moving your entire system to gain access.

MORE:Best Cases

MORE:All Case Content

Hardware Installation & Test Configuration

Included accessories for the MasterCase H500M are various screws, zip ties, an RGB extension cable and a cleaning cloth for the tempered glass side panels. The box also contains a printed manual.

The interior of this chassis is almost identical to that of the H500P Mesh White, with a few upgrades. The seven (plus two vertical) expansion slots will accommodate long graphics cards, up to 412mm, even in multi-GPU configurations. The vertical VGA GPU slots allows end users to install a PCIe x16 riser cable to show off their bright and shiny GPUs by mounting it vertically, but the riser cable is sold separately.

Cooler Master also includes an adjustable graphics card support arm for those running extra long GPUs that are prone to sagging.

Although the spec sheet claims a limit of 200mm, we tested power supplies up to 220mm in length without any issues. You can use CPU air coolers up to 190mm tall.

There are seven cable pass-through holes in the motherboard tray for cable management: three large rectangular openings with rubber grommets, two smaller holes with rolled metal edges in the top of the motherboard tray and two in the top of the PSU tunnel cover.

The H500M is equipped with a number of removable cable covers that keep the area behind the motherboard tray looking neat and tidy. This is a nice feature because, unlike the H500P Mesh White, this side of the chassis features a tinted tempered glass side panel as well.

Although we found plenty of room for cable management behind the motherboard tray, careful planning and routing is required if you choose to use the included cable covers.

The two mounting locations for the 3.5-inch hard drives are located under the two-piece power supply tunnel, closest to the front of the chassis. Additionally, you can mount up to six 2.5-inches hard drives on the PSU tunnel, behind the motherboard tray and on the cable cover bar just above the GPU support arm.

The two 200mm front intake fans feed a large 140mm exhaust fan behind the CPU socket area. This combination of fans should provide more than enough airflow for even the most demanding system configurations, without sounding like a jet engine at take-off.

Liquid cooling aficionados will be pleased to know that radiators and all-in-one coolers up to 360mm can be mounted in the top and front of this case. The mounting location in the front of the chassis can easily support radiators up to 360mm (280mm with the PSU tunnel in place) sandwiched between fans in a push-pull configuration.

There is a total of 39mm of clearance for radiators and coolers in the space above the motherboard and an additional 39mm for fans to be mounted in the area above the frame and below the top panel. The rear of the chassis supports 120 / 140mm radiators or all-in-one coolers in the exhaust fan location over the CPU socket area in single and dual-fan configurations.

Those of you worried about motherboard clearance can free up just over .5 inches of space by utilizing the the outermost mounting slots. Note that that by doing so, airflow will be impeded to a significant portion of the fins in your radiator or all-in-one cooler.

The H500M comes with a built in RGB controller located behind the rear cover for the PSU area for those of you with motherboards that do not feature ARGB functionality. This allows for color change and basic RGB effects such as breathing, fading and pulse, all selectable by hooking the RGB controller to the reset button on the front of the case. For those with ARGB-capable motherboards, the LED-lit RGB fans are natively compatible with MSI's Mystic Light, Asus' Aura Sync, Asrock Polycrome and Gigabyte's RGB Fusion lighting software.

Test Configuration

Drivers & Settings

Chipset	Intel INF 10.1.1.42
CPU	4GHz (40x 100MHz) @ 1.1V core
Motherboard	Firmware 1.10 3/2/2018
RAM	14-14-14-34
Graphics	Maximum Fan for Thermal Tests \| NVIDIA GeForce 398.36 WHQL Game Ready Driver

MORE:Best Cases

MORE:All Case Content

Benchmark Results & Conclusion

We tested the Cooler Master MasterCase H500M with our new Intel i9-7900X test platform. For comparison, we brought in the Cougar Panzer EVO RGB, the NZXT H700i and the be quiet! Dark Base 700 to give you a solid idea where this case stands against other cases that are of similar size and feature sets.

be quiet! Dark Base 700

Cougar Panzer EVO RGB

NZXT H700i

Not surprising, CPU core temperatures on our Intel i9-7900X processor running at 4GHz were rather impressive at 59°C over ambient. These results put the H500M in a dead heat (pun intended) with the Panzer EVO RGB and slightly ahead of the H700i and Dark Base 700. Swapping out the front mesh panel for the tempered glass unit, temperatures rose just 1°C. This is no doubt due to the fact that the large vents on either side of the tempered glass panel allow more than enough air flow into the chassis to keep temperatures under control.

In stock configuration, GPU temps reached 49°C over the ambient temperature. Retesting the H500M with the mesh panel installed resulted in a slight temperature increase to 50°C (122°F). These results mirror our CPU testing findings. The tempered glass panel had little impact on cooling performance.

For audio testing, we recorded sound level readings with two off-the-shelf dB meters from two different angles. We set intake fans speeds to their maximum speed of 950RPM. For our first round of testing, we used the standard metal mesh front panel and then retested the chassis with the tempered glass insert in place. Unlike our temperature tests, there was a noticeable impact on our test results depending on the choice of front panel. Tests using the metal mesh front panel resulted in sound level readings of 31.1dB at idle and 34.3dB under load. The H500M with the tempered glass panel in place registered 29.8dB at idle and 31.9dB under full load.

The H500M's great feature set, first-rate components and solid build quality justify its $200 asking price. For those that like the design of the H500M but feel $200 is too rich for your blood, there is always the Cooler Master Master Case H500P, also available in Mesh White, which shares many of the H500M's features, yet costs $50 less.

MORE: Best Cases

MORE: All Case Content

Enermax Saberay RGB Case Review: Missing the Mark

Steven Lynch — Fri, 31 Aug 2018 13:00:00 +0000

Features & Specifications

When it comes to buying a new case, people usually fall into one of three groups; those that want pure performance, those solely interested in the look of the chassis and those that want both. That's where Enermax's Saberay RGB chassis comes in. The company promises great thermal performance and flashy good looks for $170. Unfortunately, there are several issues with this chassis that will lead you toward other cases in the same price range.

Specifications

Type	Mid-Tower ATX
Motherboard Support	ATX, mATX, Mini-ITX
Dimensions (HxWxD)	22.3 x 8.8 x 18.8 inches (566 x 223 x 478mm)
Space Above Motherboard	3 inches (76.2mm)
Card Length	16.5 inches (420mm) w/o side radiator
CPU Cooler Height	6.9 inches (175mm)
Power Supply Format	Standard ATX PS2-Style PSU
Weight	26.1lbs (11.8kg)
External Bays	1x 5.25 inches
Internal Bays	2x 3.5 inches 2x 2.5 inches
Card Slots	7
Ports/Jacks	2x USB 3.02x USB 2.01x headphone jack1x microphone jack
Other	Tempered-glass side panelIncluded RGB / fan controller
Front Fans	3x 120mm RGB fans
Rear Fans	1x 140mm
Top Fans	✗
Bottom Fans	✗
Side Fans	✗
Dampening	✗

Exterior

At first glance, the Enermax Saberay RGB seems rather large for a mid-tower ATX chassis. Upon closer inspection, however, the interior of the case is much smaller than you'd think from looking at the case's outside. This is primarily due to the 2-inch-tall top panel, 1.5-inch-deep front panel and large feet that elevate the chassis 1.5 inches off the ground.

Measuring 22.3 x 8.8 x 18.8 inches (HxWxD) and tipping the scales at just over 26lbs, the Saberay RGB is constructed of steel, plastic and a tempered glass side panel. The black on black paint scheme is accented by two RGB light bars that span the entire top / front of the chassis. The top portion of this case is made of plastic and contains a plastic and metal mesh insert with mounting locations for three 120mm or two 140mm fans directly under the panel.

At the front edge of the top panel are a pair of USB 3.0 ports, two USB 2.0 ports, headphone and microphone jacks, HDD LED, power and reset buttons. On either side of the power button you will find buttons for controlling fan speed and RGB lighting effects. Enermax includes anti-dust plugs for each port to prevent debris from accumulating in the ports when the computer is not in use.

The front of the chassis, by default, features a tinted acrylic panel that, although it looks great, allows no airflow into the front of the case. Enermax includes a secondary metal mesh panel that is identical to the top panel.

Directly behind the front panel you will find mounting locations for three 120mm intake fans. This case also has a single 5.25-inch mounting location for optical drives or drive bay accessories made of the same polished acrylic material seen in the removable panel in front.

The left side of the Enermax Saberay RGB features a full-cover tempered glass side panel, which is held in place by rubber-coated locating pins and large thumbscrews.

Push-lock pins hold the acrylic front panel in place The panel is extremely prone to scratches.

The plastic push-lock pins on the top and front do not hold the panels close enough to the frame. The result is panels that do not line up well with the rest of the chassis. The plastic panels are marred with visible casting flaws that have the appearance of rub marks. These marks can only be seen at certain angles, but it is distracting nonetheless.

The rear of the chassis is home to seven card slots, a motherboard I/O area, an opening for a bottom-mounted PSU and an exhaust-fan mounting location for 120 / 140mm fans.

The bottom of the case has a filter-covered hole for the power supply, as well as two large, rubber-coated plastic feet.

The filtration system on this chassis is a real head-scratcher to say the least. The top filter is metal mesh and does an adequate job filtering larger dirt and dust particles. The problem is that by default, there are no fans mounted in this location.

To remove the large nylon filter in the bottom of the chassis, you have to access the case's rear, making cleaning a hassle. If you are the type of person that likes to show off your system build, you will need to stock up on canned air because the acrylic front panel does nothing to filter the air from the three included 120mm intake fans. The only other option is to use the included alternate front cover, which is made of metal mesh.

MORE: Best Cases

MORE: All Case Content

Hardware Installation & Test Configuration

A small white box filled with individual baggies of various screws is secured in one of the two open 3.5-inch hard drive bays. Here, you'll also find Enermax badges, a Velcro strap, a handful of zip ties and a manual.

Like many of the cases we've recently reviewed, including the Cougar Panzer EVO RGB and the NZXT H700i, all hard drive mounting locations are located behind the motherboard tray. This trend really opens up the main compartment for large graphics cards, radiators and other water cooling components but does so at the cost of hard drive mounting locations. Enermax's Saberay RGB can easily accommodate a custom cooling loop, dual 360mm radiators, a reservoir and a multi-GPU set-up.

The interior of the chassis is painted black and has a very clean and spacious layout, especially considering the smaller-than-average exterior. Two cable pass-through holes with rolled metal edges live to the right of the motherboard area, and two more are in the top of the power supply shroud. You can equip the Saberay RGB with coolers as tall as 6.9 inches and can accommodate multiple graphics cards up to 16.5 inches in length.

We ran into an issue during motherboard installation caused by clearance issues with the I/O shroud on the motherboard and the 140mm exhaust fan. Ultimately, we had to remove the fan and reinstall it after the motherboard was in place. We tried again with the MSI Z270 Gaming M5 motherboard but encountered the same clearance issue.

The motherboard tray is a two piece design; it splits into two sections with a 1-inch gap between the left and right hand sections. The right hand side of the motherboard tray, closest to the front of the chassis, has mounting locations for fans and radiators that is very similar to the layout of Lian Li's PC-O11 Dynamic Case.

A large hole in the motherboard mounting plate behind the CPU socket area allows CPU cooler changes without removing the motherboard. The cable pass-through holes at the upper edge of the motherboard tray is specifically designed to route fan cables and a 12V CPU power cable.

The cable management area directly behind the motherboard tray is a bit of a mixed bag. Although there is plenty of room for a high-end system build, the mounting location for radiators and fans leaves most of the system cables visible unless you spend a good deal of time and effort into cable routing. Overall, the cable management is sufficient for most system builds, though extra work is needed.

The large, non-removable PSU tunnel at the bottom of the main compartment is home to the power supply and two drive bays. This area, including the hard drive mounting locations, receives airflow from the bottom-most 120mm intake fan.

The Saberay RGB has mounting locations for a total of two 3.5-inch hard drives and four 2.5-inches drives, two of which are converted from the two 3.5-inch drive trays mounted under the PSU tunnel. Two of the dedicated mounting locations for 2.5-inch drive caddies mount directly behind the motherboard tray and are stamped steel.

Every power supply we tested, from the smaller EVGA 450 BT all the way up to larger units like the Corsair HX1200i fit without a problem.

Cooling

There are mounting locations for up to nine 120mm fans: three in the top, three in the front, one in the rear and two in the side of the motherboard tray. You can equip the chassis with up to three 140mm fans, two in the top and one in the rear. The Saberay RGB comes with three 120mm intake LED-lit RGB fans in the front and an unlit 140mm fan in the rear.

Water cooling aficionados will be pleased to know they can mount radiators and all-in-one liquid coolers up to 360mm on the top and front of this chassis. The mounting location in the front of the chassis can easily support radiators and all-in-one coolers sandwiched between fans in a push-pull configuration using the stock 120mm LED fans.

Although there is more than enough space in the top of the chassis for radiators and all-in-one coolers up to 63mm, users wanting to run fans in a push-pull configuration can do so by taking advantage of the space outside the frame directly under the top cover, all the while maintaining the same amount of clearance over the motherboard. The exhaust fan mounting location can be fitted with 120 / 140mm coolers in single and dual-fan configurations.

Test Configuration

Drivers & Settings

Chipset	Intel INF 10.1.1.42
CPU	4GHz (40x 100MHz) @ 1.1V Core
Motherboard	Firmware 1.10 3/2/2018
RAM	14-14-14-34
Graphics	Maximum Fan for Thermal Tests \| NVIDIA GeForce 398.36 WHQL Game Ready Driver

MORE: Best Cases

MORE: All Case Content

Benchmark Results & Conclusion

We tested the Saberay RGB case with our new Intel i9-7900X test platform. For comparison, we brought in the Cougar Panzer EVO RGB, the NZXT H700i and the be quiet! Dark Base 700 to give you a solid idea where this case stands against other cases that are of similar size and feature sets.

be quiet! Dark Base 700

Cougar Panzer EVO RGB

NZXT H700i

CPU core temperatures on our Intel i9-7900X processor running at 4GHz were very good considering the sheer amount of heat this processor is capable of generating under load. Given the fact that the Saberay RGB is outfitted with triple 120mm intake fans, we expected the chassis to be more than capable of keeping system temperatures in check.

We set intake fans speeds at max (1400RPM). For our first round of testing, we used the optional mesh front panel and, under load, our processor temps held steady at 62°C over the ambient temperature of 25°C (77°F). With the mesh panel in place, the Saberay RGB was in a dead heat with the Dark Base 700 but behind both the H700i and the Panzer EVO RGB. Unfortunately our CPU temperatures jumped a full 5°C with the default acrylic panel installed.

Once again, with the mesh panel installed for testing, GPU temps peaked at 51°C over the ambient temperature. Given the sheer amount of airflow and close proximity of the intake fans to the GPU, we expected results closer to those of others in the comparison group. As with our CPU results, we saw GPU temps increase by more than 10 percent on the Saberay RGB, easily the highest temperatures of the test group.

We took sound level readings with two off-the-shelf dB meters from two different angles. The Saberay RGB registered 30.3dB at idle and 32.8dB under load with the acrylic panel. Those results increased to 31.3dB and 34.4dB, respectively, with the mesh front panel in place.

Bottom Line

We really wanted to like the Saberay RGB. It's big, bold and has RGB functionality in spades. But, when it came right down to it, the poor build quality, oddball design choices and lack-luster cooling performance are simply unacceptable at this price point. Consumers would be better off with the Cooler Master MasterCase H500P Mesh for a lower price of $149.99 or spending a few extra dollars for the Cougar Panzer EVO RGB at $219.99

MORE: Best Cases

MORE: All Case Content

Asus ROG Strix X299-E Gaming ATX Motherboard Review: 44-Lane CPU Ready

Thomas Soderstrom — Sun, 05 Aug 2018 13:00:00 +0000

Layout & Features

The idea of a $300 board being mainstream might sound ludicrous, but entry-level X299 boards cost around $200, and Intel’s high-end desktop series is designed for big spenders. Mid-level might be a better term for this price, and the Asus ROG Strix X299-E Gaming (which sells for at or just under $300 when we wrote this) gets there with a combination of dual USB 3.1 Gen2 controllers, a 867Mb/s Wi-Fi module, enhanced overclocking and competitive RGB lighting features.

Specifications

Socket	LGA 2066
Chipset	Intel X299
Form Factor	ATX
Voltage Regulator	Eight Phases
Video Ports	✗
USB Ports	10Gbps: (1) Type-C, (3) Type A 5Gb/s: (4) Type A; (2) USB 2.0
Network Jacks	Gigabit Ethernet, (2) Wi-Fi Antenna
Audio Jacks	(5) Analog, (1) Digital Out
Legacy Ports/Jacks	✗
Other Ports/Jack	USB BIOS Flashback I/O-Panel Button
PCIe x16	(3) v3.0 (@44: x16/x16/x8) @28: (x16/x8/x1) @16: (x16/x0/x1, x8/x8/x1)
PCIe x8	✗
PCIe x4	(2) x4/x1 (Consumes SATA ports 5-8)
PCIe x1	(1) v3.0 (Consumes front-panel USB 3.1 Gen2)
CrossFire/SLI	3x / 3x
DIMM slots	(8) DDR4
M.2 slots	(1) PCIe 3.0 x4 / SATA, (1) PCIe 3.0 x4, (1) Key-E (Filled); (Consumes SATA port 1)
U.2 Ports	✗
SATA Ports	(8) 6Gb/s (Ports 1, 5-8 shared w/M.2, PCIe x4)
USB Headers	(1) 10Gb/s Type-C, (2) v3.0, (1) v2.0 (Shared w/PCIe x1)
Fan Headers	(7) 4-Pin, (1) Extender Card Header
Legacy Interfaces	Serial COM Port
Other Interfaces	FP-Audio, (2) RGB-LED, ADD-RGB, VROC_Key, Thermistor
Diagnostics Panel	Numeric
Internal Button/Switch	Power
SATA Controllers	Integrated (0/1/5/10)
Ethernet Controllers	WGI219V PHY
Wi-Fi / Bluetooth	RTL8822BE 802.11ac 2x2 (867Mb/s) / BT 4.2 Combo
USB Controllers	(2) ASM3142 PCIe 3.0 x2, ASM1074 Hub
HD Audio Codec	ALC1220
DDL/DTS Connect	✗
Warranty	3 Years

Unlike many mid-level boards though, the ROG Strix X299-E Gaming requires a 44-lane CPU to effectively use its third x16-length PCIe slot. Although it's a solid-enough board to encourage a few buyers, that limitation is a big hindrance to our overall recommendations.

The Strix X299-E Gaming has two antenna connections for Realtek’s 2T2R 802.11ac module and two ports (Type-C and Type A) for ASMedia’s PCIe 3.0 x2 USB 3.1 Gen2 controller on its I/O panel. There's also a USB BIOS Flashback button, which uses a special ASIC (application-specific integrated circuit) to enable flashing firmware without the need for a compatible CPU or DRAM. Other I/O panel features include four USB 3.0 (aka USB 3.1 Gen1), two USB 2.0, digital optical audio output and five analog audio jacks.

Zooming out shows the extremes Asus went to for the board to represent the ROG brand aesthetically, from the lighted sliver on the I/O connector hood, to the machined aluminum voltage regulator sink, the M.2 heat spreader integrated into the PCH sink cover and ROG lightbox badge at the motherboard's center. Two open-ended PCIe x4 slots add to the three x16 slots to boost enablement of serious connectivity.

Given the board’s support for everything from 44-lane to 16-lane processors, you’d expect to see the double-row of eight two-lane PCIe switches under its top slot. The surprise comes in the other three switches to the side of that group, as the lower PCIe x4 slot steals all four HSIO resources (Intel’s flexible SATA/USB3/SATA pathways) from SATA ports 5-8. And if you were thinking of using the PCIe x1 slot instead, note that doing so will switch off the front-panel USB 3.1 Gen2 connector. Given that you’ll probably have a graphics card covering the x1 slot, you might be un-enthused to learn that the upper PCIe x4 slot has only one pathway.

Compromises on the smaller slots are understandable to those familiar with Intel’s chipset limitations, particularly given the market’s focus on x16 slots. While a 44-lane CPU can shoot an x16/x16/x8 configuration and x16 processors spit an x8/x8/x1 configuration by borrowing a lane from the chipset, options for 28-lane processors teeter on the edge of infuriating: Asus says the top slot is locked at x16 mode with those processors, the middle slot gets eight CPU pathways, and the lower slot still gets only a single pathway from the chipset. Asus gives no excuses or explanations for why it couldn’t do a three-by-eight configuration for three-way SLI on 28-lane processors, nor does it say why the CPU’s other four lanes are ignored.

Layout is fairly good in spite of configuration limitations that are peculiar to the installation of a 28-lane CPU. All of the ports in front of the slots (eight SATA and one dual-port USB 3.0 header) point forward to allow installation of extra long cards. There are two case fan headers near the back of the board for easier reach to the back panel, and two CPU fan headers are just above the rear banks of memory to ease installation.

A perpendicular M.2 storage slot is found south of the 24-pin power connector, and there's a front-panel USB 3.1 interface just south of that. The USB interface gets a controller to itself, though that controller shares its pathway with--and is disabled by--the PCIe x1 slot. Similarly, the M.2 slot features four PCIe lanes for NVMe, but users of SATA M.2 drives are forced to sacrifice SATA port to when using that connection. A bracket within the installation kit allows builders to lock down their M.2 module.

The bottom of the ROG Strix X299-E Gaming has a front-panel audio header, one (of two) RGB LED cable connectors, an old-fashioned 9-pin Serial Port breakout plate header, a two-digit diagnostics code display, a power button, a special header for Asus’ aftermarket Fan Extension card, a second USB 3.0 front-panel header that can’t be used when a graphics card is installed in the bottom slot, two (of seven) four-pin fan headers, an addressable LED strip header and a standard Intel front-panel switch/LED group (9-pins) with adjacent connections for a beep-code speaker and legacy spaced (3-pin) power LED. The Intel VROC (virtual RAID on CPU) module header, which adds a RAID firmware module for NVMe drives connected to the CPU-based PCIe controller via the board’s x16 slots, is above the lower-front-corner connectors.

The ROG Strix X299-E Gaming includes a driver and application disc in the box, along with a manual, I/O shield, a riser bracket for the perpendicular M.2 slot, a two-channel Wi-Fi antenna, a high-bandwidth SLI Bridge for two cards, a pack of cable ties, four SATA cables (two with a right angle end), an Asus Q-connector bundling block for front panel power/LED leads, a thermistor lead, an RGB extension cable, an addressable LED strip extension cable, a sheet of adhesive cable labels, an Asus special offer card for cablemod.com, an ROG sticker sheet and a “Do Not Enter” door handle card.

MORE: Best Motherboards

MORE: All Motherboard Content

Software & Firmware

Asus' Aura software now controls a variety of things, from onboard lights to LED strips to RGB memory. It's known for its nearly universal support of anything that would connect to its motherboards' RGB interfaces.

Asus' Sonic Studio 3 modifies the audio output to make things either easier or more enjoyable to hear. It has Smart Volume to help reduce changes in volume level, Voice Clarity to increase the volume of dialogue and virtual surround sound.

Asus' Sonic Radar 3 adds a video overlay to 3D games that indicates the direction and distance of noises, which is useful in determining the position of moving adversaries.

The company's AI Suite 3 comprises its Dual Intelligent Processors 5 manual overclocking, TPU automatic tuning, Turbo App application prioritization, EPU efficiency optimization, Fan Xpert 4 fan tuning, Digi+ VRM CPU voltage regulator controls, PC Cleaner trash utility and a shortcut for its EZ Update application. The bottom bar of its application hosts additional status menus, along with adjustments for its system warning alerts.

Clicking the highlighted 5-Way Optimization button allows the board to run its full CPU and Fan optimization routine, which produced a 4.3GHz fixed clock on our hardware, along with a variable voltage of 1.17 to 1.24V under heavy loads. Manual tuning options allowed us to reach 4.4GHz at a fixed 1.15V CPU core setting.

Asus' EZ Update polls its servers for updates to included applications and drivers, GameFirst IV allows network prioritization to put game packets first, Ramcache II sets aside a hidden partition of memory to cache frequently accessed files, and Clone Drive adds drive-imaging to the impressive software mix.

Firmware

The Strix X299-E Gaming's firmware opens to an Advanced Mode GUI, where the My Favorites menu holds frequently accessed settings and can be reconfigured in a Setup Tree Map activated by a keyboard’s F3 function. We accessed deeper overclocking functions within the Ai Tweaker menu, where we pushed our Core i7-7900X to 4.40 GHz at the low 1.15V setting required to prevent throttling under the AVX-heavy Prime95 small-FFT’s stress test.

CPU voltage remained consistent under shifting loads using the Level 6 CPU Load-line Calibration setting from the External Digi+ Power Control menu. More surprising was that the DRAM voltage had a maximum variation of +5mV as measured at the DIMM, in an industry where many boards set 15mV or more beyond what the user desires. We have placed a hard cap of 1.355V on DRAM voltage in our overclocking tests to prevent cheating.

How many DRAM settings does the most advanced overclocker really need? The Strix X299-E Gaming has enough to fill 18 pages of screenshots. Our DDR4-3866 kit reached DDR4-3912 at its XMP-defined timings.

The Strix X299-E Gaming supports both PWM (pulse width modulation) and voltage-based fan control on all seven headers. You can also control three additional headers of an Asus Fan Extension card, should you choose to buy one. The Qfan Tuning option lets the board test every fan to find its minimum operating speed and adjust automatic curves accordingly, but users can also set their own curves mathematically or use the keyboard’s F6 function to access a menu with a three-point graphical mapping interface.

You can upgrade firmware from the firmware GUI by accessing Asus EZ Flash 3 from the Tools menu. Secure Erase didn’t function with our NVMe drive and we didn't test it on a SATA drive. You can save eight complete firmware settings as profiles and/or import/export to/from a USB drive from the Asus Overclocking Profile menu, and the SPD Information tab brings up a short page of memory module readings.

Users confused by too many options can enter EZ Mode using the keyboard’s F7 function, which reduces options to boot order, XMP activation, fan profiles and two levels of automatic overclocking. Entering Advanced Mode shows what those overclocking apps produced, with Fast Tuning pushing our CPU to 4.5 GHz non-AVX, 4.3GHz AVX and 4.2 GHz AVX 512 loads at 1.17 to 1.27V. The Extreme Tuning option increased AVX and AVX 512 loads to 4.4 and 4.3 GHz, respectively, at 1.20 to 1.27V. Both tuning algorithms selected DDR4-3200 at 1.35v for our memory.

MORE: Best Motherboards

MORE: All Motherboard Content

How We Test

Additional cores create additional thermal and power loads that can push other hardware to its limits, yet the parts we’re hoping to push to their maximum are all on the motherboard. Intel’s 10-core, 20-thread Core i9-7900X CPU is a great product for stressing the voltage regulators of various X299 motherboards, particularly under our full-AVX-load overclocking stability test.

Test Hardware

Sound	Integrated HD Audio
Network	Integrated Gigabit Networking
Graphics Driver	GeForce 382.53

The tremendous heat the Core i9-7900X produced required we use nothing less than the award-winning Fractal Design S24 liquid cooler system. Getting its fans to blow sufficiently over each motherboard’s voltage regulator required additional design consideration: Of the cases we had on hand, Cooler Master’s HAF-XB had the best layout.

Comparison Products

ASRock X299 Taichi

Gigabyte X299 Aorus Gaming 3

MSI X299 Gaming Pro Carbon AC

Similar pricing pits the ROG Strix X299-E Gaming against the AsRock X299 Taichi, Gigabyte X299 Aorus Gaming 3, and MSIs X299 Gaming Pro Carbon AC. Each of these is designed for overclocking, and the feature sets of all four are also designed to attract similar buyers.

Benchmark Settings

Synthetic Benchmarks and Settings
PCMark 8	Version 2.7.613Home, Creative, Work, Storage, Applications (Adobe & Microsoft)
3DMark 13	Version 4.47.597.0Skydiver, Firestrike, Firestrike Extreme Default Presets
SiSoftware Sandra	Version 2016.03.22.21CPU Arithmetic, Multimedia, Cryptography, Memory Bandwidth
DiskSPD	4k Random Read, 4k Random Write128k Sequential Read, 128k Sequential Write
Cinebench R15	Build RC83328DEMOOpenGL Benchmark
CompuBench	Version 1.5.8Face Detection, Optical Flow, Ocean Surface, Ray Tracing
Application Tests and Settings
LAME MP3	Version 3.98.3Mixed 271MB WAV to mp3: Command: -b 160 --nores (160 Kb/s)
HandBrake CLI	Version: 0.9.9Sintel Open Movie Project: 4.19 GB 4k mkv to x265 mp4
Blender	Version 2.68aBMW 27 CPU Render Benchmark, BMW 27 GPU Render Benchmark
7-Zip	Version 16.02THG-Workload (7.6 GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=9"
Adobe After Effects CC	Release 2015.3.0, Version 13.8.0.144PCMark-driven routine
Adobe Photoshop CC	Release 2015.5.0. 20160603.r.88 x64PCMark-driven routine (light and heavy)
Adobe InDesign CC	Release 2015.4, Build 11.4.0.90 x64PCMark-driven routine
Adobe Illustrator	Release 2015.3.0, Version 20.0.0 (64-bit)PCMark-driven routine
Game Tests and Settings
Ashes of Singularity	Version 1.31.21360High Preset - 1920x1080, Mid Shadow Quality, 1x MSAACrazy Preset - 1920x1080, High Shadow Quality, 2x MSAA
F1 2015	2015 Season, Abu Dhabi Track, RainMedium Preset, No AFUltra High Preset, 16x AF
Metro Last Light Redux	Version 3.00 x64High Quality, 1920x1080, High Tesselation, 16x AFVery High Quality, 1920x1080, Very High Tesselation, 16x AF
The Talos Principle	Version 267252Medium Preset, High Quality, High Tesselation, 4x AFUltra Preset, Very High Quality, Very High Tesselation, 16x AF

MORE: Best Motherboards

MORE: All Motherboard Content

Benchmark Results & Final Analysis

Synthetic Benchmarks

Boring synthetic benchmarks are the best indication that nobody is cheating, and we’re happy to say that the Strix X299-E Gaming’s edge in 3DMark is too slight to prompt an investigation.

Sandra Arithmetic is the best indication that none of the comparison motherboards has noticeably overclocked the CPU. One of the competing boards had an odd result in Sandra Multimedia, but that difference didn’t show up elsewhere.

Sandra Cryptography typically favors more memory bandwidth, and the tightening of advanced memory timings is one place where we haven’t really focused on vendor-specific optimizations even though the competing MSI motherboard’s bandwidth lead is somewhat compelling. The Strix X299-E Gaming performed slightly better than average in two Compubench GPU-accelerated metrics, but competitors edged it out in two others.

3D Games

The Strix X299-E Gaming hit a nearly perfect average in gaming performance, whereas a couple competitors were hit or miss. The X299 Gaming Pro Carbon AC, for example, has a premium audio software program that, when enabled, forces it to lose around 32 percent of its Talos performance. That’s because Talos is the only program compatible with that software, and enabling its features adds overhead (hollow bars show that board’s performance with the added software disabled).

Timed Applications

The ROG Strix X299-E Gaming stumbled in at least one application of each timed run chart, with noticeably longer completion times for Handbrake, Adobe Illustrator and PowerPoint. So, we changed some settings and retested. Several retests did not reveal the source of the minor setback. We discuss those retests in the Power, Heat & Efficiency section below.

Power, Heat & Efficiency

At default settings, the Strix X299-E Gaming throttled back to 3.40 GHz under the heavy AVX load of Prime95 small FFTs, saving 33W compared to the charted value. Changing its Long Duration and Short Duration firmware power settings to their maximums allowed it to run that test at 3.60 GHz, bringing it to par with the X299 Taichi and pushing its loaded wattage to 251W. After that, changing its CPU Integrated VR Current Limit firmware value to its maximum value allowed the board to run Prime95 small-FFTs at 4.00 GHz while consuming 326W. We ran our benchmarks three ways and determined that those applications weren’t stressful enough to produce noticeably different encoding times or frame rates. Since 11 of the 16 X299 boards we’ve reviewed consumed around 250W and ran at 3.60 GHz at full load, we decided to use this data.

The Strix X299-E Gaming runs a little cooler than its closest competitors, beating even the lower-powered X299 Taichi.

We looked at more than just the four boards presented in our small charts. We’re using all 16 previously reviewed boards in our overall performance and power calculations. It took a few small hits in our productivity benchmarks, but the Strix X299-E Gaming came out a mere 0.8 percent below the average of every X299 board we’ve tested in performance and produced a seven percent above-average efficiency number

Overclocking

Though a few of our four-module boards have pushed our memory past DDR4-4000, the Strix X299-E Gaming’s DDR4-3912 is a very respectable overclock compared to other eight-DIMM boards. Its 4.40 GHz CPU overclock is less noteworthy since 14 of the previous 16 boards have reached that limit, and it’s still better than the 4.5 to 4.8 GHz reported by people who don’t run 20 threads of Prime95 small-FFTs. When we say full load, we mean it!

The point of overclocking memory is to get more performance, and the Strix X299-E Gaming did that spectacularly as seen in the above chart. On the other hand, we’ve seen similar results from half of the last 16 boards we've tested. As usual, price similarity is the factor that limits the range shown in our comparison chart.

Final Analysis

The Asus ROG Strix X299-E Gaming is slightly cheaper than Gigabyte’s X299 Aorus Gaming 3 and slightly more expensive than the other two boards in our comparison chart. That price-to-performance comparison isn’t adjusted for features, however.

It’s impossible to run 3-way SLI on a 16-lane CPU, but if you’re a buyer hoping to do so on a 28-lane CPU, you can cross the Strix X299-E Gaming (and X299 Aorus Gaming 3) off your list. The same factors apply to the use of the third slot to host a PCIe x8 NVMe drive adapter to that third PCIe x16 slot, so it’s not just people running outdated graphics configuration who’ve been impacted by these design decisions. Ouch.

Builders who purchase a case with a 10Gb/s Type-C connector will want an interface for the Gen3.1 front-panel cable, and they’ll only get that from the Strix X299-E Gaming or the MSI X299 Gaming Pro Carbon AC. Those boards are also the only ones in the compared price range to include an 867Mb/s Wi-Fi solution.

A look at the complete features list shows that the X299 Gaming Pro Carbon AC does nearly everything for $20 less than the Strix X299-E Gaming, yet builders of show PCs will note that Asus’s ROG light box logo looks great and that its lighting software is universally recognized for ultimate stability and compatibility. The Strix X299-E Gaming also runs cooler and edged out the MSI board in our overclocking chart. Since buyers of 44-lane processors need not worry about the limitations of the Strix X299-E Gaming’s slot configuration, it’s at least worthy of their consideration.

MORE: Best Motherboards

MORE: All Motherboard Content

ASRock X299M Extreme4: Micro ATX Sweet Spot

Thomas Soderstrom — Fri, 29 Jun 2018 13:00:00 +0000

Features & Layout

ASRock surprised us by adding dual Gigabit Ethernet to the Micro ATX version of its Extreme4 series X299 board, ostensibly one-upping the feature set provided in its ATX version. That appears at first blush to put the X299M Extreme4 head-to-head against the X299M Gaming Pro Carbon AC from rival MSI, though the MSI board also includes a high-end Wi-Fi solution, a front-panel USB 3.1 Gen2 port, manually switched dual-BIOS, and a Port 80 diagnostics display. That feature difference is reflected in a $65 price delta that puts the value ball back in ASRock’s court.

Specifications

Socket	LGA 2066
Chipset	Intel X299
Form Factor	Micro ATX
Voltage Regulator	11 Phases
Video Ports	✗
USB Ports	10Gbps: (1) Type-C, (1) Type A 5Gb/s: (4) Type A; (4) USB 2.0
Network Jacks	(2) Gigabit Ethernet
Audio Jacks	(5) Analog, (1) Digital Out
Legacy Ports/Jacks	(1) PS/2
Other Ports/Jack	CLR_CMOS I/O-Panel Button
PCIe x16	(3) v3.0 (@44: x16/x16/x4) @28: (x16/x8/x4) @16: (x16/x0/x4, x8/x8/x4)
PCIe x8	✗
PCIe x4	✗
PCIe x1	✗
CrossFire/SLI	2x / 2x
DIMM slots	(4) DDR4
M.2 slots	(2) PCIe 3.0 x4 / SATA* (*Consumes SATA Port 0, 1)
U.2 Ports	✗
SATA Ports	(8) 6Gb/s (Ports 0, 1 shared w/M.2)
USB Headers	(1) v3.0, (2) v2.0
Fan Headers	(5) 4-Pin
Legacy Interfaces	✗
Other Interfaces	FP-Audio, TPM, (2) RGB-LED,VROC, Thunderbolt AIC
Diagnostics Panel	✗
Internal Button/Switch	✗ / ✗
SATA Controllers	Integrated (0/1/5/10)
Ethernet Controllers	WGI211AT PCIe, WGI219V PHY
Wi-Fi / Bluetooth	✗
USB Controllers	ASM3142 PCIe 3.0 x2
HD Audio Codec	ALC1220
DDL/DTS Connect	DTS Connect
Warranty	3 Years

The reason the Micro ATX Extreme4's second network port came as a surprise is that the Extreme4 series reflects ASRock’s reduced-price line of products for performance enthusiasts. The name implies that we’ll get SLI capability (even with 16-lane processors), a high-end audio codec, a fairly sturdy voltage regulator capable of pushing the air and water cooled limits of most processors, a high-end audio codec and, currently, a pair of USB 3.1 Gen2 ports on the I/O panel. As with the full-ATX version, all those features are found on the Micro ATX X299M Extreme4, in addition to its dual Gigabit Ethernet.

The X299M Extreme4’s I/O panel has a gap where you’d expect to find a Wi-Fi controller on higher motherboard models, with a CLR_CMOS button next to that gap. Competitor MSI justifies more of its price difference by including a Flashback button to enable updating firmware without a compatible CPU or RAM. This requires an additional ASIC (application-specific integrated circuit) that the X299M Extreme4 doesn’t have.

Panning out reveals that the X299M Extreme4’s enormous voltage regulator heat sinks are completely focused on cooling, lacking any of the integrated RGB lighting of its competitor. The board instead has only a few LED’s beneath the perimeter of its X299 PCH (platform controller hub) heat sink, and those all flash the same color at once--so no chase sequences or other flashy effects. We also find three x16-length PCIe slots, where the two fed by the CPU are metal-reinforced. The third slot, in black, gets only four PCH lanes.

Looking down from the top, we see that ASRock uses the same M.2 slot placement as its rival, with the first of the two slots located between the 24-pin primary power connector and forward DIMM slots. We also see two 8-pin EPS12V connectors to power the CPU, which seems a little redundant on a board this small. Yet things get a little strange from here, as there are ten two-port PCIe switches between the main PCIe slots, and no indication of what the other six switches do. We’re only told that the silver slots are configured with 16 lanes with our 44-lane CPU, 16 and 8 lanes with a 28-lane CPU, and switches from 16/0 to 8/8 lanes when a second card is added to a 16-lane CPU.

Speaking of the recently end-of-life 16-lane Kaby Lake-X processors, installing one of those dual-channel CPUs will also deactivate two of the board’s four RAM slots. Micro ATX has as much depth as ATX, yet it appears that no motherboard manufacturer has considered using that space to include the eight slots often found on the ATX versions of these boards.

The bottom edge includes a second HD-Audio front-panel header that faces downward to ease cable routing around the bottom edge of the board, a TPM connector, a jumper header that’s supposed to activate the Turbo Mode settings also available in firmware, one of the board’s two RGB LED strip headers, a VROC module connector (which adds RAID firmware to CPU-connected NVMe cards), two of the board’s five fan headers, a five-pin header for Thunderbolt add-in cards, an Intel configuration button/LED front-panel header, and an extra front-panel header with PC (beep code) speaker and three-pin-spaced power LED connection. Also found above that are six of the SATA ports, facing forward to help with tucking cables behind long graphics cards. The two additional SATA ports that point outward are located above the top expansion slot to avoid the card conflict.

The X299M Extreme4 includes cables for four of its eight SATA ports, an I/O shield, a high-bandwidth SLI bridge, a cable badge, a driver/application disc, and multilingual installation guides for both the hardware and included software.

MORE: Best Motherboards

MORE: All Motherboard Content

Software & Firmware

Offering most of the same software as its Z370 and even Z270 boards, ASRock’s X299 software suite hasn’t changed since we first covered it last year. Useful tools such as Reboot to UEFI and an ASRock-branded version of the cFos network prioritization utility remain, though software revisions have made new hardware compatible.

For example, we’re sure we wouldn’t have been able to see an illustration of the X299M Extreme4 on an older version of the software. The image shows the one onboard lighting zone, which sits under the PCH heat sink, in addition to the two RGB LED strip headers.

ASRock’s A-Tuning app still provides a software link to firmware overclocking profiles, which requires a reboot to save the firmware setting. The Auto Tuning algorithm doesn’t work with this motherboard model, likely because it uses a feature the X299M Extreme4 lacks: an external BCLK generator. Without the added part, we’re stuck at 100MHz.

Manual tuning works, though frequency is limited to CPU ratios. Voltage changes are appropriately reported on the System Info menu, though our meter shows DRAM voltage readings are slightly lower than those measured at the slots.

A-Tuning’s Fan-Tastic Tuning menu includes the ability to set your own fan profile, or allow the system to alter its baseline profile using the minimum detected fan speed.

Firmware

The X299M Extreme4's firmware opens to its Advanced Mode GUI, where the OC Tweaker menu allows users to choose between factory-programmed overclocking profiles, a default 100MHz overclock, and manual configuration. “Optimized” overclocks range from 4.20 GHz at 1.15V (using our Core i9-7900X) with a 200 MHz reduction for AVX loads, to 4.8 GHz at 1.35V.

We said there’s a default 100 MHz CPU overclock, and that’s because the firmware's default “Performance Mode” is “CPU Ratio +1”, which is plus 1 times the 100 MHz base clock. This is an alternative to the “Multi Core Enhancement” setting that most performance-oriented boards enable by default, which causes a CPU to run at its maximum frequency regardless of how many cores are loaded. By Intel spec, our Core i9-7900X is supposed to run at 4.50GHz when one or two cores are loaded, 4.1 GHz when three or four cores are loaded, and 4.0 GHz when more than four cores are loaded.

Our CPU reached 4.40 GHz at 1.15V under Prime95, a frequency that’s limited by a voltage that pushes our CPU’s thermal throttling threshold. We simply started off with the Turbo 4.40 GHz setting, disabled AVX offsets, and then set the fixed voltage ourselves.

The X299M Extreme4 pushed our DDR4-3866 memory to DDR4-4000 with a simple change in multiplier and voltage. Specifically, we had to turn the voltage down to reach the 1.35V slot-measured setting we picked to make overclocking comparisons fair.

The DRAM voltage setting, along with the CPU Load-Line Calibration setting needed to keep our core voltage stable under insane loads, are found within OC Tweaker’s Voltage Configuration menu. The X299M Extreme4 set our memory to its own 1.35V setting, even with XMP-mode disabled, and our voltmeter showed that the DIMM slots were in fact receiving 1.372V. Setting both controls to 1.330V got our voltage down to the 1.35 to 1.355V range we use for testing consistency.

OC Tweaker’s FIVR menu addresses voltage regulators that are built into the CPU, including core voltage. The 1.150V setting dropped only to 1.49V with level one CPU Load-Line Calibration set on the main voltage menu. That’s far better stability than most boards offer.

ASRock’s RGB LED firmware menu is found under its “Tool” main menu, and allows users to configure the PCH lighting and dual LED strip headers without loading a separate Windows application.

Four of the X299M Extreme4’s five fan headers support Voltage Control Mode in addition to PWM mode, and all four of those also have an increased 1.5A rating, compared the CPU fan header’s 1.0A baseline. The Fan Tuning menu enables a test to determine minimum fan speed and adjust automatic curves accordingly, while the Fan-Tastic Tuning menu adds the manual configuration option.

The keyboard’s F6 function key toggles between Advanced and EZ mode users interfaces, the later offering nothing more than a “TUBO” overclock which corresponds to the “Turbo 4.2GHz” profile of Advanced mode.

MORE: Best Motherboards

MORE: All Motherboard Content

How We Test

Additional cores create additional thermal and power loads that can push other hardware to its limits. But the parts we’re hoping to push to the limit are all on the motherboards here. Intel’s 10-core, 20-thread Core i9-7900X CPU is a great product for stressing the voltage regulators of various X299 motherboards, particularly under our full-AVX-load overclocking stability test.

Test Hardware

Sound	Integrated HD Audio
Network	Integrated Gigabit Networking
Graphics Driver	GeForce 382.53

The tremendous heat produced by the Core i9-7900X required us to use nothing less than our award-winning Fractal Design S24 liquid cooler system sample. Getting its fans to blow sufficiently over each motherboard’s voltage regulator required additional design consideration: Of our leftover case review samples, Cooler Master’s HAF-XB had the best layout.

Comparison Products

ASRock X299M Extreme4

MSI X299M Gaming Pro Carbon AC

ASRock X299E-ITX/ac

The less common Micro ATX form factor pits ASRock's X299M Extreme4 against MSI's X299M Gaming Pro Carbon AC. Our only other reduced-size board, ASRock's X299E-ITX/ac helps fill out the charts, while its full ATX X299 Extreme gives us additional perspective on the ATX/Micro ATX debate.

Benchmark Settings

Synthetic Benchmarks and Settings
PCMark 8	Version 2.7.613Home, Creative, Work, Storage, Applications (Adobe & Microsoft)
3DMark 13	Version 4.47.597.0Skydiver, Firestrike, Firestrike Extreme Default Presets
SiSoftware Sandra	Version 2016.03.22.21CPU Arithmetic, Multimedia, Cryptography, Memory Bandwidth
DiskSPD	4k Random Read, 4k Random Write128k Sequential Read, 128k Sequential Write
Cinebench R15	Build RC83328DEMOOpenGL Benchmark
CompuBench	Version 1.5.8Face Detection, Optical Flow, Ocean Surface, Ray Tracing
Application Tests and Settings
LAME MP3	Version 3.98.3Mixed 271MB WAV to mp3: Command: -b 160 --nores (160 Kb/s)
HandBrake CLI	Version: 0.9.9Sintel Open Movie Project: 4.19 GB 4k mkv to x265 mp4
Blender	Version 2.68aBMW 27 CPU Render Benchmark, BMW 27 GPU Render Benchmark
7-Zip	Version 16.02THG-Workload (7.6 GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=9"
Adobe After Effects CC	Release 2015.3.0, Version 13.8.0.144PCMark-driven routine
Adobe Photoshop CC	Release 2015.5.0. 20160603.r.88 x64PCMark-driven routine (light and heavy)
Adobe InDesign CC	Release 2015.4, Build 11.4.0.90 x64PCMark-driven routine
Adobe Illustrator	Release 2015.3.0, Version 20.0.0 (64-bit)PCMark-driven routine
Game Tests and Settings
Ashes of Singularity	Version 1.31.21360High Preset - 1920x1080, Mid Shadow Quality, 1x MSAACrazy Preset - 1920x1080, High Shadow Quality, 2x MSAA
F1 2015	2015 Season, Abu Dhabi Track, RainMedium Preset, No AFUltra High Preset, 16x AF
Metro Last Light Redux	Version 3.00 x64High Quality, 1920x1080, High Tesselation, 16x AFVery High Quality, 1920x1080, Very High Tesselation, 16x AF
The Talos Principle	Version 267252Medium Preset, High Quality, High Tesselation, 4x AFUltra Preset, Very High Quality, Very High Tesselation, 16x AF

MORE: Best Motherboards

MORE: All Motherboard Content

Benchmark Results & Final Analysis

The X299M is set to run 100 MHz higher than standard by default. We disabled that feature to make today’s comparison fair, since overclockers do their own overclocking. We also maxed the CPU current limit after finding that the board throttled the CPU under our power tests but not under the performance tests of our performance-to-power calculations.

Synthetic Benchmarks

The X299M Extreme4 appears to boost 3DMark performance out of the gate, but the impacted score is for the graphics card, and those scores fell back as we progressed to other tests. We’re considering the likelihood that this was simply luck-of-the-draw, since it’s not a CPU or DRAM metric.

Remaining synthetics show the X299M Extreme4 with average performance, which is great only because anything else would indicate a problem.

3D Games

The X299M Extreme4 produced slightly above-average performance in its first game test, the High Details setting on Ashe's of the Singularity. It falls back to average in remaining tests, with the X299M Gaming Pro Carbon AC getting a larger-than-expected benefit in F1 2015 from its improved memory performance.

The Talos Principle showed hollow bars where the X299M Gaming Pro Carbon AC’s performance would have been if we only measured it with its Nahimic Audio Solution disabled. The Talos-compatible software, which adds features such as virtual 3D and visual mapping of opponent noises, is considered an important-enough feature to MSI that we used the lower FPS score that shows its overhead. The X299M Extreme4 is unaffected.

Timed Applications

The X299M Extreme4 performed slightly worse in 7.Zip, and slightly better in Adobe InDesign, compared to other boards. In the average of all timed tests, it settles comfortably in the middle of the competition.

Power, Heat, & Efficiency

The X299M Extreme4 drew slightly more energy than its ATX sibling at full load, but both boards had remarkably low idle wattage.

Surprisingly, the X299M Extreme4 also ran a little cooler than its ATX sibling. More importantly, it beat the rival MSI board, though the MSI design likely compromises a little voltage regulator cooling thanks to its additional RGB-lit voltage regulator sinks.

The balance of performance and power yields to ASRock, even though the X299M Extreme4 falls behind other ASRock boards in efficiency.

Overclocking

The X299M Extreme4, X299 Extreme4, and X299M Gaming Pro Carbon AC all lack an external clock generator, yet only the MSI board has active BCLK multiplier control in firmware to allow users to set 100, 125, and 167MHz base-clock “straps”.

The X299M Extreme4, X299 Extreme4, and X299M had other things in common: All three pushed our CPU to a Prime95 stable 44x multiplier at 1.15V with all 20 threads running an AVX load, and all three kept our DDR4-3866 stable at a 20x multiplier (40x data rate) using 1.35V as measured at the DIMM slots.

The X299M Gaming Pro Carbon AC started off with better bandwidth at the DDR4-2133 defaults of our benchmark comparison, and that advantage scaled upward with the increased data rate. The X299M Extreme4 appears to beat its full-ATX sibling, though this could be due to its newer firmware.

Final Thoughts

Performance per dollar is exactly the formula implied by its name, taking no cues from advanced features. In other words, it’s a great chart that cheapskates can use as a shield against builders who brag about their advanced feature sets.

The above chart does leave us with a few questions, such as why the X299M Extreme4 costs $30 more than the bigger ATX X299 Exteme4, even though its circuit board is smaller and its second network controller is, at best, a $15 feature? We’ll count it as $10 and assume that ASRock is attempting to make up the other $20 on volume, since setting up a production line and paying a design team have relatively fixed costs regardless of the length of the production run.

The next question is, how does MSI justify its $65 price premium compared to the X299M Extreme4? We can start with its Intel 8265 Wi-Fi controller, which is worth around $15 bare, add its front-panel USB 3.1 Gen2 controller, which came out as an approximate $10 feature when comparing other boards to each other. And we consider the advanced overclocking features such as onboard buttons and a Port 80 diagnostics code display to get to around $35. We’ve heard that a flashback ASIC costs around $20 to implement, but we’re not sure that the thousands of buyers who don’t need it should pay that much to help the dozens of buyers who might. Our conservative estimate is that apart from its added onboard RGB lighting, the X299M Gaming Pro Carbon AC is worth $45 more than the X299M Extreme4. Maybe MSI is making a similar calculation concerning the added per-unit cost of reduced volume and assuming that it will sell even fewer of its costlier product?

The biggest value losers will obviously be buyers of the X299E-ITX/ac. And before we start counting features, we should start with the notion that Mini ITX builders are accustomed to paying a premium for convenience. The X299E-ITX/ac is, in fact, the best value in Mini ITX boards for LGA 2066, thanks to it being the only LGA 2066 board in that form factor. Sturdier voltage regulator components are required to deal with the heat of its tight confine. Riser boards are needed to deal with its lack of mounting space, and performance is the only truly fair way to compare it.

And that brings us back to the X299M Extreme4. It's a slightly lesser value than its full-ATX sibling, a slightly better value than its better-equipped Micro-ATX competitor, and a fair better value than its Mini ITX alternative. That’s a nice niche to own.

MORE: Best Motherboards

MORE: All Motherboard Content

Overclocking Intel's Core i9-7900X CPU Up to 5.8 GHz

Jean Michel "Wizerty" Tisserand — Fri, 08 Jun 2018 13:00:00 +0000

Preparing To Overclock: Hardware and Software

Intel's Core i9-7900X is an incredibly fast 10-core CPU. For the privilege of owning one, expect to pay a cool grand. However, Intel's decision to slather thermal paste between the processor's die and heat spreader really put a cap on the highest frequencies most overclockers can coax from it right out of the box.

But what if you're willing to get a little (or a lot) more adventurous? We de-lidded our sample, cleaned it up really well, replaced Intel's thermal paste with a couple of third-party alternatives, and collected plenty of data to show how Core i9-7900X's overclocking headroom improves as it's cooled more effectively. The culmination of our experiments was a -7900X under the effects of liquid nitrogen, running at a screaming 5.8 GHz, yet stable enough to complete a multi-core run of Cinebench.

Prepping Our Motherboard For Extreme Overclocking

By default, most motherboards aren't equipped to handle that kind of extreme testing. So, we had to do a bit of prep work and perform a couple of modifications first. Accomplished overclockers don't just make BIOS adjustments, after all. You also have to be handy with a soldering iron, drill, razor blade, and fingernail polish.

The MSI X299 Xpower Gaming AC we used is built for overclocking, so it served as the perfect platform to support our endeavor. Soldering a few wires to the board at various voltage detection points allowed us to monitor critical levels with a multimeter during our tests. While it's all well and good to dial in specific voltages through the motherboard firmware, operating at this level calls for precision only possible through verification.

BIOS: Expert Mode

We used MSI's BIOS version E7A91IMS.150, which was the latest one available when we started testing. It's the only firmware we used from start to finish, so feel free to track that one down if you want to retrace our steps. Of course, if you have an X299 board from a different manufacturer, the settings and terminology we use should still give you a good idea of what to adjust. The same principles apply from one platform to another, even if the nomenclature varies.

To access Expert mode on MSI's X299 Xpower Gaming AC, press F7 on the BIOS welcome screen.

The following settings come from the OC menu:

Extreme OC Setup: This toggle allows you to deactivate energy-saving options, certain protection features, and functionality that has no bearing on overclocking, such as the audio controller. This option isn't used for our ambient cooling tests, but it's mandatory for benchmarking under liquid nitrogen.
CPU Ratio Apply Mode: The [All Core] setting ensures that all cores have the same multiplier and operating frequency.
CPU Ratio: Set to 44 and multiplied by the 100 MHz CPU Base Clock, we get an operating frequency of 4400 MHz. The clock rate can then be adjusted via software running under Windows.
CPU Ratio Offset: In order to limit frequency variation during our most demanding tests, we applied an offset of [-4]. This constrains throttling to four 100 MHz bins during benchmarks that utilize AVX instructions, all the while improving stability.
Ring Ratio: A coefficient of [34] sets the ring bus to 3400 MHz.
EIST & Intel Turbo Boost: These settings default to [Disabled] when we modify the previous parameters. Since we are going after a static frequency, and don't want the clock rate to bounce around, we leave both features deactivated.
CPU Base Clock: We used a BCLK of 100 MHz to keep things simple for our frequency calculations. Later, we'll modify this from the operating system to fine-tune our overclock.

DRAM Setting: With fast G.Skill TridentZ modules on-hand, we're able to use the memory's XMP profile for reproducible results. These 8GB sticks operate at 4000 MT/s with 18-19-19-39 timings.
DigitALL Power: This is a very important sub-menu on the X299 platform, and the settings we used are shown below:

Buried among all of those other options, this is one of the BIOS' most important menus. As you may know already, Core i9-7900X's power consumption ramps up quickly when it's overclocked. Without any intervention on our part, the CPU's frequency throttles back in order to keep from exceeding defined power and thermal limits.

The worst part of this is that you may not even realize what is happening. You'll see the Adjusted CPU Frequency right where it should be, but performance won't correspond. What good is it to display high clock rates if they can't be sustained? We won't tease you like that. When we say the CPU hit 4800 MHz, rest assured that we took the time to validate its stability.

VR 12VIN OCP Expander: Increasing this value pushes the over-current limit higher. At too low of a level, maximum current is choked off and performance suffers. Moving it up to a maximum value of [20A] gives the processor more headroom when it's needed. Note that other manufacturers expose options as high as 180A. Following in our footsteps puts additional strain on your power supply, so make sure your PSU is up to the challenge first.
CPU Loadline Calibration Control: LLC is also an important parameter that we'll cover in more depth on the following pages. The ideal setting is [Mode 5].
CPU Over/Under Voltage Protection: OVP and UVP protect the system from over- and under-voltage conditions. Our primary interest is stability, and higher values offer more headroom.
CPU Over Current Protection: OCP protects the system from over-current conditions; the [Enhanced] option allows us to push the platform further before it forcibly intervenes.
CPU Switching Frequency: This is the frequency at which the voltage supply stages operate. Increasing it offers greater stability, but causes temperatures to rise. If your VRMs have a tendency to run hot, try setting this parameter to its minimum value and see if the system still runs stably.
CPU VRM Over Temperature Protection: The voltage supply stages on X299 motherboards have a tough job to perform, so ensure that you're using a well-ventilated case and plenty of airflow. We did not set this option to its maximum value. Instead, we went with a number that'd avoid damaging the motherboard. A temperature of [110°C] seemed high enough for our tastes.

If you put a fan on the supply stage, you can regulate its speed as a function of temperature. This also works for other fans and temperatures. Simply choose the temperature you're interested in (MOS, in the image below), select the name of the channel with a fan attached (CPU1, in our example), and place the colored points for each thermal range at the desired fan speed.

Press the [F] key to set all fans to 100%.

Of course, we'll also need to make adjustments to various voltages. Remember: while overclocking isn't particularly dangerous when you're careful, it's important to be reasonable. Abusing these settings won't get you closer to your objective.

VCCIN Voltage: This is the primary voltage responsible for supplying other parts of the processor. VCCIN feeds the Vcore, Vring, Vsa, Vio, and so on. Therefore, it's important to set this option higher than the others. Generally, we advise increasing it by 500 to 600mV. As you might imagine, a mismanaged VCCIN can have significant consequences.
CPU Core Voltage: This is the voltage applied to each core, and it's the main setting used for CPU overclocking.
CPU Ring Voltage: A higher Vring helps us hit higher ring bus frequencies without compromising stability.
CPU Uncore Voltage Offset: Adjusting this parameter helps us achieve stable memory overclocks. Without an offset, our RAM kit wouldn't run at 4000 MT/s. But it had no trouble with a setting of +300mV. The kit fired right up with this option set to [Auto], so it's safe to say an automatic configuration applies at least +300mV. Of course, we want to double-check using a multimeter since automatic settings have a tendency to get aggressive.
CPU SA and IO Voltage: Both options may also help stabilize the memory.
DRAM CH_X/X Voltage: These settings are also used to overclock the RAM, though they can be used to fine-tune each channel's voltage.

The following table should help guide your own overclocking endeavor, per MSI's own recommendations. It indicates the "safe" voltage operating ranges. But don't crank any of these up unless you determine that there's a useful reason to do so.

Voltage	General	LN2
VCCIN Voltage	Up to 2.1V	Up to 2.6V
CPU Core Voltage	Up to 1.4V	Up to 1.52V
CPU Ring Voltage	Up to 1.4V	Up to 1.4V
CPU Uncore Voltage Offset	Up to +400mV	Up to +400mV
CPU SA Voltage	Up to 1.35V	Up to 1.35V
CPU IO Voltage	Up to 1.35V	Up to 1.35V
CPU PLL Trim Offset	Up to +200mV	Up to +300mV
DRAM Voltage	Up to 1.9V	Up to 2V

MORE: Best CPUs

MORE: All CPUs Content

Intel Core i9-7900XView Deal

Delidding An Intel Core i9 CPU

Delidding a processor involves removing its integrated heat spreader, which sits on top of the delicate die and dissipates its thermal energy across a larger surface. We're looking to delid Intel's Core i9-7900X because the company's thermal compound isn't the highest-performance stuff out there, and we want to improve the chip's ability to transfer heat into a beefier cooler.

Back when Intel's Skylake-X-based CPUs launched, there was no tool available for delidding Core i9-7900X. We designed our own and introduced it in Tom's Hardware's Custom Core i9 De-Lidder. Check out that story for a step-by-step walk-through of the process.

To Clean, Or Not To Clean?

Before proceeding, we tested a couple of different preparations: there's the delidded processor on the left, and the delidded chip with its upper adhesive cleaned off on the right. We wanted to measure the temperature differences before and after cleaning. We used the same thermal paste in both cases, applying Thermal Grizzly Conductonaut between the die and IHS, and Thermal Grizzly Kryonaut between the IHS and heat sink.

Red finger nail polish covered the surface-mount components surrounding the die. Conductonaut is electrically conductive, so our goal was to prevent accidental short circuits.

Afterward, we prepared the processor for testing without its IHS. This is commonly referred to as direct-die cooling, and it naturally requires an even more delicate touch, since the die is completely exposed.

MORE: Best CPUs

MORE: All CPUs Content

Intel Core i9-7900XView Deal

Installing The Direct Die Frame

Delidding is already pretty advanced. With our Direct Die Frame, we take things to another level. Rather than temporarily removing the IHS to replace the thermal paste underneath, and then replacing it, the DDF concept calls for leaving it off. In theory, it is preferable to have less material between the die and heat sink to improve the transfer of thermal energy away from the CPU.

As mentioned, though, the die is more fragile in this unprotected configuration. The setup process isn't particularly difficult, but we still wouldn't recommend it to inexperienced overclockers who aren't accustomed to more extreme modifications like these.

Without its IHS, the processor can no longer be held securely in its motherboard interface. Extreme overclocker Der8auer developed a tool for addressing this conundrum: the direct die frame. In the photo below, MSI's locking system is on the left, while Der8auer’s DDF is on the right.

Step 1: Remove the four locking screws using the provided key.

Step 2 : Stick adhesive pads onto the back of the plate bundled with the DDF.

Step 3: On the motherboard's back side, replace the original plate with the one you just prepared.

The adhesive pads are very strong. Der8auer advises that you put all four of them in place, but that you only expose one. Leave the red part on the three others, allowing you to remove the plate later on, if needed.

Step 4: Place the processor in what is left of the socket.

When using the DDF, it is recommended that you remove all leftover adhesive from the IHS. To prevent damage to the processor’s PCB, do not use hard or sharp objects. Also, be careful not to tear off any of the surface-mounted components.

Step 5: Place the DDF on your board and secure it into place using four screws and the provided key. It only fits one way. If you don't get it right, the DDF won't sit flat.

Tighten the screws in an X pattern, switching diagonally from one to the next, to ensure a secure fit.

Step 6: Assemble your cooler, which should theoretically be compatible. Sadly, the screws from our EK-Supremacy stuck out slightly from the cooling plate. As a result, the block wouldn't make contact with the die. It turned out that the screws on the bottom of the water block caused it to push on the DDF instead of the die.

After reaching out to EK about this issue, the company let us know that compatibility problems were limited to products it manufactured more than a year ago. We did get our hands on a more modern version of the Supremacy and can confirm the problem was resolved. The DDF didn't even exist one year ago, so we can't fault EK here.

All the same, make sure that your water block is compatible before purchasing a DDF of your own. More than anything, it's best to check that nothing protrudes into the lower surface of your block.

Not wanting to wait for a fix (before we knew one was available), and not wanting to start over with another cooling solution, we decided to modify our water block and make it work with the DDF.

Conceptually, the operation was simple: use a counterbore to expand the four screw holes and completely countersink the screw heads into the base. Reality wasn't nearly that tidy, though.

As soon as the first shavings appeared, the cooling plate's protective nickel layer came up. Of course, water blocks aren’t intended to be modified like this. But the outcome was still surprising, causing us to question the protective layer's quality.

With deeper holes dug and the nickel coating removed, our water block was ready for action. With a bit of Thermal Grizzly Conductonaut between our block and the CPU's die, we were ready to start testing.

MORE: Best CPUs

MORE: All CPUs Content

Intel Core i9-7900XView Deal

Temperatures: Stock Vs. Delidded Vs. Direct Die Cooling

With our platform built and the components prepared, it was time to quantify the thermal differences between delidding a CPU and replacing its thermal paste, and cooling the die directly. In each of the following charts, Without Modification means that no changes were made to the CPU; it's a stock sample. Delidded indicates that the processor's integrated heat spreader was removed and then replaced after using higher-quality thermal paste. Delidded and Cleaned means that we also took the time to remove Intel's adhesive residue, in addition to replacing its thermal compound with better stuff from our lab. Of course, the Direct Die entry corresponds to removing the IHS and mounting our thermal solution on the die itself.

Establishing A Baseline

For our first pass, we stuck with default settings and tested the processor using three workloads. Cinebench R15 is one of the most common benchmarks for quantifying the effect of an overclock. It can utilize one core or all of them, and it's incredibly easy to run. We specified a static clock rate of 4000 MHz and dialed in a 1.06V supply voltage. Those parameters should be easily attainable by anyone.

As expected, the stock CPU with no modifications posted the highest temperatures. Delidding helped shave off 5 to 6°C on average, and cooling the die directly helps even more.

Prime95 v.26.6 does not utilize AVX instructions. But it's still one of the most taxing workloads you can hit a CPU with.

Our Core i9-7900X continued operating at 4000 MHz at 1.06V. Not surprisingly, the finishing order remains the same, though the impact of more effective cooling is amplified.

Switching over to Prime95 v.29.1 adds AVX instructions to the mix. The brutality of this workload automatically forced our CPU down to 3600 MHz at 1.0V (due to our CPU ratio offset of -4). But despite the more taxing benchmark, less aggressive clock rates and voltages lead to slightly lower temperatures than the preceding run without AVX support.

If we stopped benchmarking right there and had to draw a conclusion based on these results, we'd recommend leaving your Core i9 alone. The risk of damaging a $1000 CPU, voiding its warranty, and sinking money into extras like the delidding tool just aren't worth a slightly lower operating temperature. However, the gains become more significant once you start overclocking to higher frequencies.

The Impact of Overclocking

Next, we ran the same three workloads using higher clock rates and voltages. In our pursuit, we sought settings that put our chip's temperature as close to 100°C as possible.

In Cinebench R15, that meant a frequency of 4400 MHz at 1.25V. The overclock imposed significantly warmer temperatures, which increased from 46 to 75°C out of the box. And the delta grew as we improved the configuration's ability to dissipate heat quickly.

With or without adhesive, delidding facilitated gains of 11 to 12°C on average, but also reduced the hottest core's temperature by 15°C. Direct-die cooling helped even more, yielding an 18°C drop compared to the manufacturer’s stock heat sink and fan.

We used the same settings in Prime95 v.26.6 (without AVX), and were treated to significantly warmer temperatures. One core even got as hot as 96°C (never mind our high-performance water cooler and fairly modest 1.25V voltage setting). This is where we realized the benefit of delidding: gains of 15°C after replacing Intel's stock thermal paste and 22°C thanks to the Direct Die Frame are huge.

We had to lower the clock rate and voltage significantly under Prime95 v.29.1. At 4100 MHz and 1.15V, it was already getting worryingly close to our 100°C target. While executing AVX instructions is known to impose warmer temperatures, we've never seen anything like this. And the Core i9-7900X only has 10 cores. Just imagine the 18-core Core i9-7980XE!

MORE: Best CPUs

MORE: All CPUs Content

Intel Core i9-7900XView Deal

Overclocking: Frequency, Power, And Temperature Scaling

As you overclock, there's always a point when the CPU no longer runs stably and you have to stop pushing higher. The most well-known symptom of this is the blue screen of death (or BSOD). When you start seeing those, there are three possible remedies:

Dial in a lower clock rate to regain stability.
Modify other BIOS parameters and hope for a better outcome.
Use a more effective cooling solution.

Cores versus Vcore: IHS In Place

What does increasing the Vcore setting do to frequency scaling? To answer this question, we used two different Core i9-7900X CPUs at various supply voltages in order to determine the highest clock rate able to complete a run of Cinebench R15. The voltage rose in 0.1V increments up through 1.2V, at which point we added a 1.25V setting for more granularity. Frequency measurements are reported with a precision of 25 MHz.

At their base frequency, both processors performed identically, completing the test at 4250 MHz. Increasing the Vcore by 0.1V bought us a maximum clock rate of 4575 MHz on our second sample, and 4600 MHz on the first one. That's a nice speed-up for a relatively small voltage increase. The next increment was similar, maintaining a 25 MHz gap between the CPUs for a total increase of 250 MHz on both of them. With 1.3V applied, the progression slowed. One Core i9 jumped 100 MHz and the other overclocked 150 MHz higher, widening the gap between them.

Adding Vcore is the primary means by which enthusiasts coax more frequency from their CPUs. But this approach is not without its limitations, as processors run hotter and their scaling slows. When the chip gets too hot, or when it requires a lot of extra voltage to hit slightly higher clock rates, it's time to stop.

Cores versus Vcore: Direct Die

This test zeroes in on our second Core i9 sample, comparing its frequency gains before delidding to direct-die cooling.

Even at the lowest voltage setting, direct-die cooling proved advantageous. However, the gap widened beyond 1.2V. The explanation is simple: prior to delidding, the processor ran hot, keeping the cores from utilizing the extra voltage available to them. The DDF helped it run cooler, allowing the CPU to stabilize at higher frequencies.

Enabling a 100 MHz-higher overclock and a temperature drop of 20°C, direct-die cooling is looking a lot more attractive.

Cache versus Vring

Now we're interested in exploring how the cache frequency scales as we increase the Vring setting.

Increments of 100 MHz explain the shape of our curve. Smaller steps would have yielded a smoother line. But the important take-away is that, for each 0.1V interval, we can hope for roughly 100 MHz of additional headroom.

Power Consumption and Temperature

Next, we looked at the effects of overclocking on power consumption and processor temperature by fixing the Vcore at 1.3V and varying the clock rate.

The first thing we noticed was the almost perfect relationship between temperature and power. It's also remarkable that the progression was practically linear. By increasing clock rate by 15.8%, power consumption jumped 14.8% and the temperature rose 13.5%.

Next, we fixed the frequency at 4250 MHz and raised the Vcore from 1.0 to 1.3V. The temperature still seemed tied to power consumption, but its increase was no longer linear. According to our measurements, pushing Vcore up 30% caused power consumption to rise by 63% and temperature to increase 55%.

Our data illustrating the impact of Vcore on the processor's temperature and power consumption gives enthusiasts one more reason to take it easy with that parameter during an overclock attempt.

MORE: Best CPUs

MORE: All CPUs Content

Intel Core i9-7900XView Deal

Optimized Overclocking And Load-Line Calibration

As you overclock to progressively higher clock rates, the only time you increase Vcore is to obtain a more aggressive frequency setting. With this in mind, we revisited our two Core i9-7900Xes to suss out the optimal operating points for temperature and power consumption.

The temperature curves are similar to those obtained for power; one always correlates with the other.

Up until 4600 MHz, it was relatively easy to get big frequency increases from small Vcore adjustments. That meant we were in a range conducive to overclocking. An 8% overclock across 10 cores led to a rise in power consumption of "only" 22%. In this zone, our frequency gain to consumption ratio is close to 1/3.
Moving past 4600 MHz, a 5% overclock caused a 23%-higher power consumption, yielding a ratio close to 1/5.
Beyond 4800 MHz, a 2% overclock led to a 24% rise in power consumption, meaning a ratio of 1/12! At that point, there's little reason to chase even higher frequencies.

The Effects of Load-Line Calibration

When the CPU is idle, using very little energy, our 1.3V setting is easy to obtain. Once a demanding test launches, however, the voltage might drop to 1.26V. An even more taxing workload pushes it lower still. Obviously, that's not what you want to see from a stable system. Fortunately, many motherboards come with a function called Load-Line Calibration. Once activated, LLC artificially raises the supply voltage to compensate for this effect.

In the interest of thoroughness, we ran some tests to quantify the effects of LLC on Vcore. Notice that the measured voltage at idle was always lower than its BIOS setting, though not dramatically so since there was no risk of instability. In Prime95 (without AVX), LLC modes 1 though 5 delivered a Vcore slightly higher than what we asked for, while modes 6 through 8 were a little lower. Prime95 (with AVX) is an even more strenuous test, so modes 1 through 3 over-supplied the Vcore by a substantial amount. At a 1.2V setting, we saw a real voltage of more than 1.23V. Starting with mode 4, the Vcore more closely matched our set point.

Interestingly, we found that the effect of LLC wasn't identical from one sample to the next. Now, we know that Skylake-X-based processors are equipped with an internal voltage supply, so the motherboard doesn't directly supply Vcore. It instead supplies Vccin, which helps explain the small changes we just saw. With that bit of information in-hand, we restarted our tests and kept an eye on Vccin, rather than Vcore.

For a set point of 1.72V, we observed that mode 1 delivers voltages as high as 1.79V (+0.07V) under heavy load. At the opposite end of the spectrum, mode 8 is too weak; its voltage dropped as low as 1.69V (-0.03V). The ideal compromise seemed to be mode 5, which demonstrated balanced results.

MORE: Best CPUs

MORE: All CPUs Content

Intel Core i9-7900XView Deal

The Main Event: Overclocking With LN2

With all of that practical information out of the way, it's time to set aside our water-cooling loop and set up for liquid nitrogen.

Conductonaut versus Kryonaut

At ambient temperatures, Conductonaut performs better than classic thermal compounds. But experience tells us that it doesn't fare well under the influence of extreme cold. With this in mind, we maintained 2.1V Vccin, 1.4V Vcore, and 4800 MHz core clock rate settings.

With the cooling pot's base at 20°C, Core Temp indicated a power consumption of 330W and core temperatures ranging from 80 to 98°C.
At 0°C, power consumption reportedly dropped to 303W. That was naturally a surprising result, so we confirmed it with the monitoring capability of our Cooler Master MasterWatt Maker 1200. Lowering the cooling pot's temperature by 20°C also dropped the core temperatures by as much as 30°C.
At -20°C, power consumption fell as low as 285W. The cores cooled off by another 25°C.
At -40°C, our power consumption measurement was 275W (or 55W less than our starting point). The hottest core measured 17°C, while the coolest was at 1°C.
At -60°C, the system became unstable after we observed rapid and significant core temperature increases. This is our sign that the thermal paste stopped doing its job.

This test confirmed that Conductonaut, a high-performance paste at ambient temperatures, doesn't work below -60°C. Our experiment also gave us several interesting data points to show the direct correlation between CPU's operating temperature and power consumption. Although it's less effective than Conductonaut at ambient temperatures, Kryonaut is a better choice for extreme overclocking.

Direct Die Frame versus IHS

Next, we set out to see if the benefits of direct-die cooling mapped over to overclocking with liquid nitrogen.

Although we did our best to record accurate and precise results, bear in mind that it's very difficult to achieve repeatability with LN2.

With the cooling pot at -50°C, our DDF performs a lot like the Core i9 with its IHS intact. At the next temperature increment, direct-die cooling takes a small lead. But at -70°C, where we thought the gap would widen even further, both approaches converge once again. Tests at -80, -90, and -100°C resulted in similar observations. Whether you use direct-die cooling or a heat spreader between the die and cooling pot, the outcome is near-identical.

Maximum Frequency

The highest stable frequency we reached that'd still run Cinebench R15 was 5800 MHz. Not bad for a CPU that gave us so many problems at launch using more mainstream approaches to cooling.

MORE: Best CPUs

MORE: All CPUs Content

Intel Core i9-7900XView Deal

Conclusion

Core i9-7900X is a 140W CPU. But we already knew from our launch coverage that overclocking pushes it way beyond that thermal design power. In fact, the most remarkable result from today's testing was how quickly the CPU got hot when we overclocked it.

Although delidding isn't necessary if you use the Core i9-7900X at its default settings, replacing Intel's thermal compound does become an important step if you increase the supply voltage by even a little bit. And if you're looking to squeeze more performance from this 10-core behemoth, direct-die cooling can help decrease its temperature by more than 20°C. The Direct Die Frame is available now; just be sure it's compatible with your cooler.

As you work with your own Skylake-X-based processor, keep an eye on its critical parameters. These processors have a maddening tendency to throttle clock rates without much warning. Bad BIOS settings can severely limit performance, so take the time to adjust the Vccin and Load Line Calibration options carefully. As for Vcore, adjust it sparingly since small increases can seriously affect operating temperatures and power consumption.

Also, know when to stop. Avoid pushing for a 2%-higher overclock when it costs you 20% more power consumption. That's critically important if you want to maintain high performance without compromising stability.

The world of extreme overclocking is subject to a completely different set of rules. There, a few megahertz separate record-holders from everyone else, so it's important to start with the right equipment, and follow up with a delicate touch for finessing knobs and dials in the correct order. Conductonaut thermal paste, which performs very well in ambient environments, fails catastrophically once temperatures are lowered to -60°C. Moreover, although direct-die cooling worked brilliantly with our water-cooler, it wasn't particularly convincing when we broke out the liquid nitrogen.

As for our MSI X299 Xpower Gaming AC and Core i9-7900X combination, it continues running well in the Tom's Hardware FR lab, fast, stable, and cool. This goes to show that overclocking, performed carefully, won't damage your components. Take things slowly, test thoroughly, and don't be greedy.

MORE: Best CPUs

MORE: All CPUs Content

Intel Core i9-7900XView Deal

ASRock X299 Extreme4 ATX Motherboard Review: Making Core-X Affordable?

Thomas Soderstrom — Sat, 14 Apr 2018 01:00:00 +0000

Features & Layout

Similar prices between Intel’s Core i7-7800X and Core i7-8700K haven’t gone unnoticed by value-seeking performance PC builders. Both have six cores and 12 threads, but the i7-8700K is clocked higher and has more cache. The top processor in the mainstream segment looks like a clear value leader, until we consider the 7800X’s extra PCIe lanes. SLI can go more ways. CrossFire can be made faster. More drives can interface the CPU directly, rather than going through the chipset’s shared four-lane interface.

Yet just when you start to think that Core-X might be the way to go, you’re slapped with the reality of $300 motherboards. Maybe you can’t afford it? ASRock thinks it can help!

We can’t exactly say that the $200 X299 Extreme4 will fill all of your Core-X-pectations, however, as one of the ways ASRock kept cost down was to reduce the number of PCIe pathway switches, compared to these previously mentioned $300 rivals. Core i9s shed 12 onboard PCIe lanes, and the 78xx models four lanes, just to maintain slot compatibility with those low-end Kaby Lake-X processors we’d just as soon forget about. Rather than rant about Intel’s decision to produce Kaby Lake-X, let’s see what the board can do with the remaining lanes!

Rather than lose any storage pathways when connected to a Kaby Lake-X processor, the X299 Extreme4 ties both M.2 slots to the chipset. Users who would rather have faster storage than SLI can use an adapter such as ASRock’s Ultra Quad M.2 Card to add up to four NVMe drives to a single x16 slot. ASRock will tell you that you’ll need a 44-lane (Core i9) processor to get 16 lanes to its second PCIe x16 slot, but what it didn’t say is that users of 28-lane CPUs can alternatively put their graphics card in the second x16 slot and operate it with eight pathways, while using the first x16 slot for storage. ASRock even provides the settings to address the four storage cards independently in the top x16 slot...

The above settings worked perfectly for us, and that our graphics card lost only around 5% in the transition from x16 to x8 mode. We also found that PCIe 1.0 was the default mode for a single graphics card mounted in the second x16 slot, so that manual configuration was necessary to reach the card’s performance potential. Having shown that the first x16 slot can be used for storage (to enable 4x4 NVMe configurations on 28-lane processors), we’re ready to treat the X299 Extreme4 as a true High-End Desktop platform and get back to the pretty pictures!

A VROC interface at the top of the board’s front edge allows builders to essentially add RAID firmware to the CPU-based PCIe lanes, enabling RAID mode for multi-M.2 adapter cards such as the one mentioned above. Other high-speed storage options include eight SATA 6Gb/s ports at the front of the board, Type-A and Type-C USB 3.1 Gen2 ports on the I/O panel, and two M.2 interfaces above and below the top PCIe x16 slot. A third M.2 interface near the board’s back edge uses the PCIe/USB interface of Key-E for full compatibility with notebook-based Wi-Fi/Bluetooth combo cards.

We’ve been proponents for the retention of USB 2.0 as a keyboard/mouse interface ever since Intel introduced flexible HSIO interfaces for PCIe, SATA, and USB 3.0. Those resources are a little scarce, and there’s no need to waste any on a keyboard and mouse. ASRock takes resource conservation to the next level by having two USB 2.0 plus two PS/2 ports, none of which needs an HSIO pathway. Other I/O-panel items include five analog and one digital optical audio interface, four USB 3.0 and two USB 3.1 Gen2 interfaces, and a CLR_CMOS button.

While expensive boards typically preserve I/O resources by using a hub for some of their USB 3.0 ports, ASRock gets around that problem by not having a second USB 3.0 front-panel header. This also makes sense, since each header serves two ports, and most cases have only two USB 3.0 front-panel ports.

Layout oddities include a top PCIe x16 slot that aligns to the case’s third expansion slot, and a front-panel audio connector that’s pushed forward from the traditional corner by about an inch. While the audio-header placement is beneficial to those whose cables might otherwise come up a little short, pushing a graphics-card slot southward while maintaining three slots of separation (for improved graphics cooling) means that the second PCIe x16 slot aligns to the case’s sixth position. Due to the thickness of most graphics-card coolers, CrossFire and SLI users will typically find the third x16-length slot inaccessible. And remember what we said about 28-lane CPUs requiring the top x16 slot to host ASRock’s Ultra Quad M.2 Card? Moving a single graphics card to the second x16 slot will also effectively block the bottom x16-length slot. In that configuration, the only remaining expansion slots are the PCIe x1 and M.2 Key-E slot. And, since PCIe x1 and M.2 Key-E share a lane, using one of those interfaces excludes using the other.

In addition to its favorably positioned front-panel audio header, the X299 Extreme4’s bottom edge has a TPM header, one (of two) RGB headers, a CLR_CMOS jumper, a Thunderbolt add-in-card header, two (of five) fan headers, two USB 2.0 front-panel headers, a header for a Performance Mode/Easy OC switch set (not included), an Intel-style front panel LED/Switch group, a four-pin PC (Beep-code) speaker header, and a second power LED header for three-pin-spaced cables.

The X299 Extreme4 includes a driver disc, documentation, four SATA cables, a high-bandwidth SLI bridge, an antenna cable bracket, the I/O shield, and a case badge. The antenna bracket makes sense given the PCIe plus USB requirements of Wi-Fi/Bluetooth combo cards: While users could install a PCIe x1 adapter in the top slot and drape its USB cable all the way down to a header at the bottom of the board, they could alternatively remove the module from such a card, install it in the M.2 Key-E slot, and connect the card’s antenna cables to the X299 Extreme4’s bracket.

MORE: Best Motherboards

MORE: All Motherboard Content

Software & Firmware

ASRock’s RGB software interface is unchanged from recent reviews, but the image always reflects the motherboard model used. In this case, it shows that the X299 Extreme4 only has circumferential lighting under its PCH sink, plus RGB headers.

ASRock EZ OC mode for A-Tuning is based on manufacturer-programed firmware overclocking profiles, which we’ll get to within the firmware section in paragraphs below. Manual tuning worked within our ability to test it, which included changes to the CPU multiplier, CPU core voltage, and DRAM voltage.

A-Tuning’s System Info provides a realistic reading of temperatures and voltage, and its Fan-Tastic tuning allows users to pick cooling profiles or have the board use its own algorithm to determine an optimized ratio of temperature-to-noise.

Firmware

The X299 Extreme4 firmware opens to ASRock’s Advanced Mode GUI, where the OC Tweaker menu prominently displays factory-programmed overclocking profiles of 4.2GHz at stock core voltage, 1.90V CPU input voltage, and incredibly high -2/-9 AVX/AVX3 offsets for the CPU multiplier, 4.4GHz at 1.90V input and -4/-11 offsets, 4.60GHz at 1.26V CPU core / 2.00V input voltage and -6/-13 AVX offsets, and 4.80GHz at 1.32V core / 2.10V input and -8/-15 AVX offsets. No thanks!

The OC Tweaker menu also offers storage for five user-programmed overclocking configurations, a utility to export and import those using a USB flash drive, and several submenus from which users can make their manual overclocking adjustments.

We reached 4.40GHz with full AVX stability and no AVX offset, using only 1.15V CPU core and no increase in CPU I/O voltage. The CPU would go farther if not for the problem of extracting heat from all 10 cores when running 20 threads of Prime95 small FFTs: As with every other X299 board we’ve tested, the additional voltage required to reach higher clocks causes CPU thermal throttling under that load.

The X299 Extreme4 offers a complete set of primary, secondary, tertiary, and even quaternary timings, in addition to other memory settings intended to help users find the fastest stable data rates at the quickest stable timings. Using standard XMP voltage and timings, the board stably pushed our DDR4-3866 to DDR4-4000.

The DRAM Tweaker submenu simply displays timing tables taken from a module’s SPD IC.

ASRock divides its voltage menus by motherboard-based and CPU-integrated controllers. DRAM voltage is controlled by the motherboard, and our voltmeter showed a 25mV discrepancy between set voltage and actual voltage. Most manufacturers do similar things to achieve better stability, yet it has lead to a vicious cycle where the fastest memory often requires more voltage than it is rated at, simply to reach its rated settings. We like to compare our overclocks using the same voltage on every board tested, and 25mV is at the upper range of these cheats. On the other hand, ASRock is at least honest enough to report the actual voltage in its H/W Monitor.

The X299 Extreme4 did an excellent job of maintaining our desired 1.15V CPU core setting, using its “Level 2 Load-Line Calibration” setting.

The X299 Extreme4’s “Advanced” menu has a setting where users can change the default GUI from “Advanced Mode” to “Easy Mode.” This is also where users will find the important PCIe controls discussed on page 1 of this review.

ASRock’s “Tool” menu includes an RGB control interface identical to that of its software, an email-sending client for tech support, an “Easy Raid Installer” (it’s supposed to copy files from the motherboard’s support DVD to a USB flash drive, but it didn’t read the files on this support DVD), a firmware-flashing utility that reads files from a USB flash drive, and a utility that polls ASRock servers for new firmware and downloads it to a USB flash drive.

The X299 Extreme4’s H/W Monitor page includes fan settings, where we found that only two of the headers could be switched from PWM to voltage-based controls. Its “Fan Tuning” algorithm creates custom temperature-to-speed slopes supplemental to factory programming, and its FAN-Tastic Tuning menu allows users to create their own custom fan slopes.

If those settings seem a little elaborate to you, feel free to punch the keyboard’s F6 function key and go to Easy Mode, where simple things such as XMP-enable, boot order, and fan settings can still be accessed.

MORE: Best Motherboards

MORE: All Motherboard Content

How We Test

Additional cores create additional thermal and power loads that can push other hardware to its limits, yet the parts we’re hoping to push to the limit are all on the motherboards. Intel’s 10-core, 20-thread Core i9-7900X CPU is a great product for stressing the voltage regulators of various X299 motherboards, particularly under our full-AVX-load overclocking stability test.

Test Hardware

Sound	Integrated HD Audio
Network	Integrated Gigabit Networking
Graphics Driver	GeForce 382.53

Comparison Products

ASRock X299 Extreme4

Asus Prime X299-A

MSI X299 Gaming Pro Carbon AC

Close price matches to the X299M Gaming Pro Carbon AC include ASRock's X299 Taichi, Gigabyte's X299 Aorus Gaming 3, and Asus' Prime X299-A. We've added MSI's X299 Gaming Pro Carbon AC and ASRock's X299E-ITX/ac for a more-entertaining ATX-to-MicroATX-to-Mini-ITX comparison

Benchmark Settings

Synthetic Benchmarks and Settings
PCMark 8	Version 2.7.613Home, Creative, Work, Storage, Applications (Adobe & Microsoft)
3DMark 13	Version 4.47.597.0Skydiver, Firestrike, Firestrike Extreme Default Presets
SiSoftware Sandra	Version 2016.03.22.21CPU Arithmetic, Multimedia, Cryptography, Memory Bandwidth
DiskSPD	4k Random Read, 4k Random Write128k Sequential Read, 128k Sequential Write
Cinebench R15	Build RC83328DEMOOpenGL Benchmark
CompuBench	Version 1.5.8Face Detection, Optical Flow, Ocean Surface, Ray Tracing
Application Tests and Settings
LAME MP3	Version 3.98.3Mixed 271MB WAV to mp3: Command: -b 160 --nores (160 Kb/s)
HandBrake CLI	Version: 0.9.9Sintel Open Movie Project: 4.19 GB 4k mkv to x265 mp4
Blender	Version 2.68aBMW 27 CPU Render Benchmark, BMW 27 GPU Render Benchmark
7-Zip	Version 16.02THG-Workload (7.6 GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=9"
Adobe After Effects CC	Release 2015.3.0, Version 13.8.0.144PCMark-driven routine
Adobe Photoshop CC	Release 2015.5.0. 20160603.r.88 x64PCMark-driven routine (light and heavy)
Adobe InDesign CC	Release 2015.4, Build 11.4.0.90 x64PCMark-driven routine
Adobe Illustrator	Release 2015.3.0, Version 20.0.0 (64-bit)PCMark-driven routine
Game Tests and Settings
Ashes of Singularity	Version 1.31.21360High Preset - 1920x1080, Mid Shadow Quality, 1x MSAACrazy Preset - 1920x1080, High Shadow Quality, 2x MSAA
F1 2015	2015 Season, Abu Dhabi Track, RainMedium Preset, No AFUltra High Preset, 16x AF
Metro Last Light Redux	Version 3.00 x64High Quality, 1920x1080, High Tesselation, 16x AFVery High Quality, 1920x1080, Very High Tesselation, 16x AF
The Talos Principle	Version 267252Medium Preset, High Quality, High Tesselation, 4x AFUltra Preset, Very High Quality, Very High Tesselation, 16x AF

MORE: Best Motherboards

MORE: All Motherboard Content

Benchmark Results & Final Analysis

We test motherboards with all the power savings features enabled and any factory overclocks disabled, including the disabling of “enhanced turbo” modes that most manufacturers now use to force the highest Intel Turbo Boost ratios regardless of the number of cores loaded.

Synthetic Benchmarks

By focusing on specific system components, synthetic benchmarks are a great tool to determine if a manufacturer is cheating, or if a component is misconfigured. 3DMark and PCMark show the X299 Extreme4 to be within the normal performance range.

The X299 Extreme4 stays close to the high end of the previously defined performance range though Sandra, Cinebench, and Compubench.

3D Games

We would love to say that something stood out about the X299 Extreme4’s gaming performance, but the only real deviation we saw was a Talos performance deficit in the X299 Gaming Pro Carbon AC, where the board’s Nahimic 3D-positioning software imparted overhead when it encountered a game with which it was compatible. And MSI users can disable that software to boost Talos FPS, if they choose.

Timed Applications

Nothing is easier to explain than mundane numbers. Intel has done such a great job of integrating functions that there are few ways other than overclocking for a single board to get ahead. Oh, there is the X299 Extreme4’s slight miss in our PowerPoint bench, but a difference that small might have occurred by happenstance.

Power, Heat, & Efficiency

There’s really nothing to explain about the X299 Extreme4’s power numbers either, as it falls in line with an Asus and an ASRock board from previous reviews. The higher MSI and Gigabyte numbers are caused by poor power management and/or CPU core over-voltage.

The X299 Extreme4 has a large voltage regulator heat sink to keep those temperatures reasonable, and its loaded CPU temperature is close to the average of the two other boards of similar energy use.

Who would have thought that the three boards that consume similar wattage would have similar efficiency scores? But there could be an important point to be made about all these ordinary findings, since the X299 Extreme4 has around two-thirds of the price of the other products in this comparison.

Overclocking

The X299 Extreme4 reaches the same ultimate CPU frequency as other solid boards, with only one of its competitors falling behind. Moreover, it allowed our memory to run at DDR4-4000 settings. On the other hand, one of the ways ASRock reduced the cost of this board was to rely on integrated BCLK control, rather than a discrete clock generator, so that the BCLK couldn’t be altered whatsoever.

The extra DRAM clock came with extra DRAM bandwidth, dispelling the notion that ASRock might have gotten there by lengthening its timings. This could have been a great board for memory comparisons, if only it had an external clock generator to enable the fine adjustments required in those comparisons.

Final Thoughts

A few things that really stuck out about the value-priced X299 Extreme4 were that it performs as well as an expensive board, it overclocks our heat-sensitive CPU as well as an expensive board, and that its expansion slot configuration is far less flexible than any more-expensive board.

People love to talk about performance-per-dollar, a.k.a. bang for the buck, even though such charts don’t make concessions for added features. With none of its M.2 interfaces tied to the faster CPU-based PCIe controller, the X299 Extreme4 forces builders to choose between two graphics cards or a single card plus fast storage. Three-way CrossFire isn’t even a consideration when the third slot is only one space away from the second.

But the X299 Extreme4 is only $200, and anyone who wanted to run just a single graphics card and a PCIe x16-to-four-way-M.2 adapter will find it suitable. On the other hand, gamers choosing between the Core i7-7800X and Core i7-8700K will have a hard time justifying the use of a slower-clocked 7800X, since both platforms will restrict at least one card to x8 mode.

2018 brings us one award option, Editors' Choice, and if we had only $200 to spend on an LGA-2066 motherboard, the X299 Extreme4 is the only board we could choose.

MORE: Best Motherboards

MORE: All Motherboard Content

MSI X299M Gaming Pro Carbon AC Motherboard Review: MicroATX to the Max

Thomas Soderstrom — Thu, 29 Mar 2018 13:00:00 +0000

Features & Layout

Buyers trying to decide between ATX and Mini-ITX have our complete sympathy, as the larger form factor is beyond the needs of most enthusiasts. Hardcore gamers typically run one or two ultra-high-end graphics cards, and even power users typically have only four DIMMs, a single high-end graphics card, and no more than two other high-bandwidth devices. Going ATX means buying all the huge pieces that support it, which seems like a terrible waste of space to users who only wanted a couple more DIMM slots and a slightly larger CPU voltage regulator.

Conceptualized to support the low-energy processors of set-top boxes, digital signage, and office terminals, Mini-ITX was never intended to have those things. AMD tried to address the conceptual flaws of the Mini-ITX “godbox” in its 9.6”-deep DTX initiative, whereupon experienced builders mostly asked “why” and didn’t stick around to hear the answer.

Yet another compact form factor that applies all the fixes to satisfy most enthusiasts has been with us all along:

Within the X299M Gaming Pro Carbon lies the capability for gamers to install two full-size graphics cards in full PCIe 3.0 x16 mode. Power users can just as easily stuff a workstation card in one x16 slot, a 4x4 M.2 storage adapter in the other x16 slot, and the x8 card of their choice in a third slot. The X299M Gaming Pro Carbon AC offers most of what we liked about its full-ATX sibling, but in a smaller MicroATX package.

Specifications

We did say that the X299M Gaming Pro Carbon AC supports most of that features of its larger sibling, but a few key things are missing besides the two lower expansion slots. First, the board doesn’t support 16-lane (Kaby Lake-X) processors. (Then again, neither did most of us.) The U.2 connector is also missing, though users who need to move their NVMe storage off the board are free to use an M.2 slot adapter.

The MicroATX version also has one fewer USB port on the I/O panel than the full-size X299 Gaming Pro Carbon AC, yet the smaller board gains a second network port, along with the associated controller (Intel’s i211AT). Both use the same PCIe 3.0 x2-based ASM3142 to feed Type-C and Type-A USB 3.1 Gen2 ports, both have the same 867Mb/s Intel 8265 Wi-Fi/Bluetooth module, and both provide CLR_CMOS and USB BIOS Flasback buttons on the I/O panel. That last feature allows users to flash firmware from a USB flash drive using an onboard ASIC, regardless of whether a compatible processor and RAM have been installed.

The X299M Gaming Pro Carbon AC keeps all eight SATA ports, donating only one to each of the M.2 slots when SATA-based drivers are installed there. Yet in spite of each M.2 slot’s PCIe 3.0 x4 NVMe capability, the MicroATX board faces the same M.2 bandwidth-sharing limitation of the ATX version, due to the four-lane DMI of Intel’s X299 PCH. The quest for ultimate drive performance points to the 16 lanes of the second slot and the eight lanes of the third slot, since certain interface adapter cards can break those out to four-lane M.2 drives. Half of those lanes are disabled when using a 28-lane (Core i7-78xxX) processor, but x8/x4 without DMI restrictions is still a great bit better than what the chipset offers.

As far as we can tell, upper M.2 slot placement is the primary reason that MSI put four DIMMs on the X299M Gaming Pro Carbon AC, unlike its eight-DIMM big brother. Yet putting it there also limited space around the USB 3.1 Gen2 front-panel connector, so its dedicated ASM3142 controller was moved to the back of the board. The MicroATX board also loses the larger board’s second USB 3.0 header, though it keeps both USB 2.0 connections.

Software RAID from the CPU’s PCIe controller requires an add-in module, and the necessary VROC module header is located in the front lower corner, right between the USB 2.0 headers and the last pair of SATA ports. Power and Reset buttons are located above it. To the left of (i.e., behind) the USB 2.0 headers you'll find the front-panel switch/LED group, a dual-ROM BIOS selector switch, a TPM header, a button to enable “Demo Mode” for the onboard RGB controller, a power input for demonstrating LED modes in an uninstalled board, two (of the board’s five) fan headers, an RGB LED output, and an HD Audio front-panel header.

The X299M Gaming Pro Carbon AC does not include a high-bandwidth SLI bridge. The flexible bridge makes sense from the standpoint of enabling users to configure SLI using either the second or third slot, but leaving the low-cost OEM part out of the package when aftermarket pieces are quite pricey seems stingy. We still get a very nice pair of magnetic-base antennas, four SATA cables, an 80cm RGB LED splitter cable, two alternative top plates for the decorative I/O connector cover, a case badge, and some stickers, along with the required documentation, a driver and application disc, and the I/O shield.

MORE: Best Motherboards

MORE: All Motherboard Content

Software & Firmware

As indicated by the nearly identical screen shot, the X299M Gaming Pro Carbon AC includes the same software suite described in our X299 Gaming Pro Carbon AC review. That said, we can focus on the minor differences in how these applications work on the newer, smaller model.

For example, MSI’s normally solid Command Center software suite still has most of its features, but BCLK overclocking doesn’t work, and neither does Game Boost. The first problem comes down to a lack of BCLK fine tuning as of BIOS version 1.40. As for Game Boost, it was only able to set our CPU to a -3 offset for AVX loads.

The problem with Game Boost might have something to do with a crash that preceded our attempt to enable it in software. The setting is firmware-based, and a previous attempt to engage it from firmware caused the system to hang continuously.

Mystic LED worked on its home screen, but nothing we tried could bring up its advanced settings menu. As such, monochrome lighting patterns and sticky color changes were the only settings we could reach. Die-hard builders will note that the software isn’t completely necessary, since the onboard Demo LED button allows users to cycle through all lighting and color modes.

Firmware

MSI Click BIOS 5 opens to its EZ Mode GUI only upon first entry, after which it “remembers” the way it was exited the previous time. EZ Mode allows users to enable XMP memory and Game Boost factory-programmed overclocks, set boot order, access fan settings, and so on. Pressing the F7 keyboard function gets users to the more tuner-friendly Advanced GUI.

As with big boards, the X299M Gaming Pro Carbon AC easily pushed our 10-core CPU to 4.40GHz under 20-thread AVX loads using only 1.15V for the CPU core. That voltage is ultra-important, because Prime95 Small FFTs will push our CPU to its thermal threshold when attempting greater overclocks at higher voltage. Yet while the big MSI boards typically lack manual “boot strap” adjustment, the “CPU Base Clock” setting of this MicroATX model is nothing but boot strap. Users can only select 100MHz, 125MHz, or 166MHz.

Our DDR4-3866 easily reached DDR4-4000 without hesitation, using the board’s 1.340V setting. That again is important, because our voltmeter measured 1.3549V at that setting. We use the “1.355V rounds up” math rule to guide our voltage limit, because most manufacturers try to win overclocking comparisons by adding more voltage than the system reports.

Memory tweakers are welcome to chase down higher overclocks at looser timings, or better performance with tighter timings, as the X299M Gaming Pro Carbon AC supports a full range of primary through tertiary latency adjustments.

The DigitALL Power menu includes CPU Loadline Calibration, otherwise known as “vdroop compensation,” to hold up the CPU core voltage as it begins to sag under heavy data loads. Mode 3 kept our overclocked CPU just below its thermal throttle limit.

Other submenus of the OC Menu allow users to view CPU and memory specs, and enable or disable various performance-enhancement and power-saving features.

The firmware M-Flash button instructs the system to reboot to a special flashing mode for firmware updates.

There’s enough extra room on each firmware ROM to support six custom configurations. Users can also export and import these profiles using a USB flash drive.

All five fan headers can be switched between PWM and voltage-based control modes, using either factory-programed or user-defined temperature-to-RPM slopes.

MSI Board Explorer shows where things are plugged in, making it easy for a user to see whether a component is being detected at POST. Moving the mouse pointer over each connector prompts the GUI to display additional information.

MORE: Best Motherboards

MORE: All Motherboard Content

How We Test

Test Hardware

Sound	Integrated HD Audio
Network	Integrated Gigabit Networking
Graphics Driver	GeForce 382.53

Comparison Products

MSI X299M Gaming Pro Carbon AC

ASRock X299E-ITX/ac

MSI X299 Gaming Pro Carbon AC

Close price matches to the X299M Gaming Pro Carbon AC include ASRock's X299 Taichi, Gigabyte's X299 Aorus Gaming 3, and Asus' Prime X299-A. We've added MSI's own X299 Gaming Pro Carbon AC and ASRock's X299E-ITX/ac for a more entertaining ATX-to-MicroATX-to-Mini-ITX comparison.

Benchmark Settings

Synthetic Benchmarks and Settings
PCMark 8	Version 2.7.613Home, Creative, Work, Storage, Applications (Adobe & Microsoft)
3DMark 13	Version 4.47.597.0Skydiver, Firestrike, Firestrike Extreme Default Presets
SiSoftware Sandra	Version 2016.03.22.21CPU Arithmetic, Multimedia, Cryptography, Memory Bandwidth
DiskSPD	4k Random Read, 4k Random Write128k Sequential Read, 128k Sequential Write
Cinebench R15	Build RC83328DEMOOpenGL Benchmark
CompuBench	Version 1.5.8Face Detection, Optical Flow, Ocean Surface, Ray Tracing
Application Tests and Settings
LAME MP3	Version 3.98.3Mixed 271MB WAV to mp3: Command: -b 160 --nores (160 Kb/s)
HandBrake CLI	Version: 0.9.9Sintel Open Movie Project: 4.19 GB 4k mkv to x265 mp4
Blender	Version 2.68aBMW 27 CPU Render Benchmark, BMW 27 GPU Render Benchmark
7-Zip	Version 16.02THG-Workload (7.6GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=9"
Adobe After Effects CC	Release 2015.3.0, Version 13.8.0.144PCMark-driven routine
Adobe Photoshop CC	Release 2015.5.0. 20160603.r.88 x64PCMark-driven routine (light and heavy)
Adobe InDesign CC	Release 2015.4, Build 11.4.0.90 x64PCMark-driven routine
Adobe Illustrator	Release 2015.3.0, Version 20.0.0 (64-bit)PCMark-driven routine
Game Tests and Settings
Ashes of the Singularity	Version 1.31.21360High Preset - 1920x1080, Mid Shadow Quality, 1x MSAACrazy Preset - 1920x1080, High Shadow Quality, 2x MSAA
F1 2015	2015 Season, Abu Dhabi Track, RainMedium Preset, No AFUltra High Preset, 16x AF
Metro Last Light Redux	Version 3.00 x64High Quality, 1920x1080, High Tesselation, 16x AFVery High Quality, 1920x1080, Very High Tesselation, 16x AF
The Talos Principle	Version 267252Medium Preset, High Quality, High Tesselation, 4x AFUltra Preset, Very High Quality, Very High Tesselation, 16x AF

MORE: Best Motherboards

MORE: All Motherboard Content

Benchmark Results & Final Analysis

While a few X299 motherboards will attempt to hold the Core i9-7900X’s rated 4.00GHz Turbo Boost setting for 10-core loads and report a TDP of over 200W under the heft of Prime95 small FFTs, even fewer will attempt to keep the CPU within Intel’s original 140W TDP by clocking all the way down to around 3.4GHz. Most samples we’ve tested allow the CPU clock to drop to 3.60GHz, and report a TDP of somewhere between 160W and 190W, and the X299M Gaming Pro Carbon AC counts itself among these.

The power-management behavior of various boards and even firmware revisions is something to keep in mind when we get to the power and efficiency ratings at the bottom of this page, yet none of our regular benchmarks can hold so high a demand on the processor for so long that throttling becomes necessary.

Synthetic Benchmarks

Anyone hoping to see something special in 3DMark or PCMark might be disappointed, but we are happy to report that the MicroATX design has no significant impact on the results of these benchmarks.

The X299M Gaming Pro Carbon AC has a slight lead in Sandra Memory Bandwidth, and the improved memory performance also has a slight impact in a few other benchmarks, but there’s no differences large enough to set off any alarm bells.

3D Games

The slight memory-bandwidth advantage mentioned above pays dividends at our lowest setting of F1 2015, but it’s not like anyone who owns such otherwise-unrestricted hardware would need to use those low settings. Ashes is the game where some may claim they’ve no FPS to spare, and the X299M Gaming Pro Carbon AC produces on-par performance in that title.

Both the X299M Gaming Pro Carbon AC and its ATX sibling are delivered with Nahimic’s audio solution, for which Talos is the only compatible game we’re using. Using it imparts a moderate FPS impact. We retested it with the software disabled, and the performance gain is shown as the empty bars of each chart.

Timed Applications

The X299M Gaming Pro Carbon AC produces on-par overall performance in our mixed workloads, where less time means better performance.

Alas, today’s test subject had a small performance stumble in our PowerPoint test. But that’s only one test...no big deal, right?

Power, Heat, & Efficiency

The X299M falls right to the middle of our power chart, hampered slightly by what appears to be an inability to reach its lowest idle power state while still active (i.e., not suspended or sleeping).

Heat is another concern for the X299M Gaming Pro Carbon AC, as its voltage-regulator sink is too far from the fans of our test system and too low-profile to catch much of the remaining airflow. That’s not a MicroATX issue, though, and those who would claim otherwise should go back to Page 1 and look at the pictures so that they can see that MSI had plenty of room to increase heatsink size. Perhaps the designers had the top-mounted fans of liquid-cooled tower cases in mind with this design?

The X299M Gaming Pro Carbon AC ends our test 0.9% below average, even after the big hit it took in Talos with its Nahimic audio solution enabled. Those who don’t want the software are welcome not to install it, but some users really like it. Moreover, its efficiency is 3.7% above average.

Overclocking

The X299M Gaming Pro Carbon AC beats its larger sibling in CPU overclocking, but that’s probably because the larger board used early firmware. The ability to use a DDR4-4000 data rate setting was also a nice bonus, and we probably could have taken it even higher if BCLK fine adjustment had been available.

We continue to validate the performance gain of memory overclocking, since some manufacturers were using bad timings to boost O/C capability on past boards. The X299M Gaming Pro Carbon AC has none of those problems, and it tops our bandwidth chart with an excellent 74GB/s.

We'd even like to use this board for DRAM testing, if only it had BCLK fine adjustment capability. The only purpose we have for that setting is to determine winners in overclocking tests, as we’d be very happy with the simple 4.4GHz CPU clock and DDR4-4000 settings if we were using this in an ultimate-performance PC.

Final Analysis

Testing boards of vastly different price levels leaves nothing of credible value in our Performance Per Dollar chart, yet we were pleased to have three competing boards of similar price in the ASRock X299 Taichi, Aorus X299 Gaming 3, and Asus Prime X299-A for today’s charts. The other two comparison boards, the X299E-ITX/ac and the X299 Gaming Pro Carbon AC, show Mini-ITX and ATX alternatives to the MicroATX form factor.

From a features standpoint, the more expensive boards are the closest matches here to the X299M Gaming Pro Carbon AC. Both have the same premium Wi-Fi adapter as the MicroATX sample, the smaller board has the same dual Gigabit Ethernet configuration, and the ATX board has the same dual USB 3.1 Gen2 configuration. Boards in the X299M Gaming Pro Carbon AC’s price range come up even shorter, with the ASRock X299 Taichi offering only a low-end Wi-Fi controller and single USB 3.1 Gen2, the Aorus X299 Gaming 3 having only single Ethernet, a single USB 3.1 Gen2 controller, and no Wi-Fi, and the Prime X299-A having both USB 3.1 Gen2 controllers but no Wi-Fi. (Of course, all the ATX boards have eight DIMM slots and additional PCIe expansion.)

The X299M Gaming Pro Carbon AC thus meets all of our criteria for an award, but in 2018 the only award we issue now is an Editors' Choice. As this is the only X299 MicroATX model that any manufacturer has been willing to submit, we have to reserve a verdict on that score; but we’d love to put it in either a dual-card gaming rig or a productivity PC. BCLK fine-tuning is esoteric for most buyers, after all, and the onboard feature set is excellent. Unfortunately, the X299M Gaming Pro Carbon AC does not include one of those low-cost OEM HB-SLI bridges, and we really can’t bring ourselves to pay 10 times as much to get a flashy-looking one out of retail channels. Ah, for the want of a nail.

Like many of its high-end forbearers, the X299M Gaming Pro Carbon AC proves that MicroATX should be the ideal form factor for most enthusiasts, if not for a few elder advice-givers who associate the entire form factor with the cheapest boards of the 1990s. We’d love to remind our senior readers that relying on 20-year-old experiences can lead to opinions that are 15 years obsolete. The X299M Gaming Pro Carbon AC is your proof, if you’re open-minded enough to try it.

MORE: Best Motherboards

MORE: All Motherboard Content

Asus ROG Strix Z370-E Gaming Motherboard Review: Right-Priced RGB?

Thomas Soderstrom — Mon, 26 Mar 2018 13:00:00 +0000

Features & Layout

We’ve been breaking motherboards into our own defined market segments for a few years, at least since before the boards themselves quit doing some "breaking" of their own. By which we mean: Motherboard prices went up when the PC-enthusiast market broke away from traditional PCs, and since then anything marketed for less than $160 has typically been equipped with only relatively basic features. The middle market, designed to satisfy most performance enthusiasts, has crept up to a rough range of $160 to $220, while a complete set of high-end features generally costs a bit over $220. And the wild stuff typically happens past $280.

With that said, the Strix Z370-E Gaming is priced at around $210. So, is this a better value in a high-end board, or just a mainstream overclocking board with a few added features? Let's take a closer look.

The single light bar on the board hints that it’s not designed for serious show-offs, but that’s okay with us. Lots of people are more interested in what a board can do than with what it looks like. And besides, there’s nothing unsightly about brushed-aluminum heat sinks or a plastic I/O connector cover that’s been painted silver to match.

Specifications

A quick look over the specs shows primarily high-end intentions, with two ASM3142 controllers that separately provide Type-C and Type-A ports to the rear, and a newfangled USB 3.1 Gen2 header for the front, all at 10Gb/s. Sharing has even been minimized to allow all the SATA ports and M.2 slots to work simultaneously with certain configurations; the second M.2 port (the one under the heat sink) can be set to two lanes to enable the SATA ports it steals, and the upper M.2 only requires an SATA port if its filled with a SATA drive. Performance enthusiasts are all about the NVMe at this point, often reserving SATA for backup drives.

Even with those things in mind, it’s hard to ignore some of the port choices on the I/O panel. Do enthusiasts need three outputs for the CPU’s onboard graphics? Why are there only six total USB ports, when users who want to use the rear-panel USB 3.1 Gen2 will often reserve those two ports for portable devices? Why is there no CLR_CMOS button? And considering that this is an Asus-made enthusiast board, why isn’t there any USB BIOS Flashback?

That last-mentioned omission suggests “because this is an enthusiast-level mainstream board” rather than a true high-end offering. And that's fine, for what it is. In a search for things that would boost it to the top of the mainstream segment—where its price resides—we find an RTL8822BE 2x2 (867Mb/s) controller with Bluetooth 4.2, in Key-E format, mounted to a custom riser card that also holds its antenna connectors. That’s something. It might even be worth $20, to those who want it. And it’s certainly nice to have that feature on the I/O panel, where it won’t occupy valuable slot space.

That means the rest of the board, in theory, only needs to compete against products around $190. And that's a much easier hill to climb, particularly considering its dual USB 3.1 Gen2 controllers. We can also see that the two metal-sided PCIe x16 expansion slots, which automatically switch from x16/x0 to x8/x8 modes when a second card is installed, are reinforced with soldered-down through-pins. The third long slot is connected via four lanes to the PCH.

Another quick look back at the possible-configurations table shows that while the two-lane front-panel USB 3.1 Gen2 controller steals pathways from two PCIe slots, those two slots that get disabled are the ones directly under graphics cards. In other words, for most users of SLI or CrossFire, the two disabled slots were going to be covered up by graphics-card coolers anyway. You might think that’s not such a big deal, until you see that one of Asus’ biggest competitors hasn’t figured that out yet…on motherboards costing more than twice as much. Smart move, here, by Asus' designers.

The Strix Z370-E Gaming’s front-panel audio header is moved about an inch forward of its traditional corner to aid builders whose cables are a little short. There’s also a header for vintage serial-port breakout plates, which looks incongruous on a gaming-market motherboard, but we'll take it. Next to that are one of the board’s two RGB headers, a TPM header, an addressable LED, an Asus proprietary fan hub, one of the board’s two front-panel USB 3.0 headers, two USB 2.0 headers, and one of the board’s six fan headers. The front panel LED/button header contains a standard Intel pin layout, with PC Speaker and alternative (3-pin spaced) power LED segments added near the forward edge.

Five more fan headers surround the CPU, the second regular RGB LED header is located at the Strix Z370-E Gaming’s top edge, and all six SATA ports point forward, right next to the upper USB 3.0 front-panel header, to avoid mechanical interference with long expansion cards. Additional mounting points are placed in key locations for optional 3D-printable shrouds.

In the box, the Strix Z370-E Gaming includes a load of gear. You get four SATA cables, RGB LED and addressable-LED extension cables, a thermistor lead, a Wi-Fi antenna, an HB-SLI bridge, a CPU installation tool, a CPU fan holder, cable ties, documentation, a driver/application disc, the I/O shield, a doorknob hanger card, and stickers.

MORE: Best Motherboards

MORE: All Motherboard Content

Software & Firmware

Asus has done an excellent job of developing and promoting its Aura lighting-control software, which is compatible with various third-party devices (such as RGB DRAM lighting), in addition to the Strix Z370-E Gaming’s single light bar, RGB LED headers, and Addressable LED header.

3D Audio is designed to help gamers hear the direction and distance of the noises made by their opponents. Asus Sonic Radar 3 provides a video overlay to those who still need help.

Asus Sonic Studio 3 provides similar tools to a certain famous brand’s add-on software, such as Smart Volume to help maintain audio output at a more consistent level, Voice Clarity to boost dialog, and virtual surround sound.

Asus AI Suite 3 includes its Dual Intelligent Processors 5 system tuning app with TPU automatic tuning, Turbo App manual overclocking, EPU efficiency optimization, Fan Xpert 4 (to determine the lower limit and slope for use in fan speed temperature curves), a manual overclocking menu, Digi+ VRM additional controls for the CPU voltage regulator, and the PC Cleaner trash utility.

There’s lots of branding going on in AI Suite 3, and TPU runs its 5-Way Optimization program for automatic overclocking, fan tuning, and idle voltage tuning for improved efficiency. In our anecdotal testing, it picked a 1.35V core for our CPU to allow it to operate at 4.8GHz, with a -3x multiplier for AVX loads. It scaled down to 4x when fully loaded due to thermal throttling, before suddenly rebooting eight minutes into our Prime95 test.

Our own tests show that our own processor should be limited to 1.30V to avoid thermal throttling, even if that imposes tighter overclocking limits. Asus’ software could do that, but its firmware can do it outside the operating system. So you may want to do the manual-tuning thing to get the most out of this board if the auto feature doesn't delight.

Other Asus Applications include Asus EZ Update server polling, GameFirst IV network prioritization, Ramcache II hidden memory partition for caching frequently accessed files, and the Asus Clone Drive cloning and drive-imaging utility.

Firmware

The Strix Z370-E Gaming firmware opens to an EZ Mode interface, which includes the ability to enable XMP mode, set boot order, select fan profiles, and run its EZ System Tuning utility. We ran that utility, selecting “Gaming/Media Editing” as our primary task and “Liquid Cooling” as our cooling format, and were rewarded with a 5.15GHz overclock at 1.45V. Prime95 crashed our fourth CPU core instantly, however. Retrying it with “Tower” cooling selected, the program chose 1.425V CPU core at 5.05GHz, whereupon Prime95 crashed core 4. Our third try, choosing “Boxed” cooler, set a 4.95GHz clock at 1.40V and likewise saw Prime95 crashing core 4 instantly.

Pressing the keyboard’s F7 key switches the firmware GUI to Advanced Mode, where manual settings are only a few clicks away. Our CPU reached 4747MHz at 1.30V core, a bit shy of its previously attained 4848MHz limit. Dropping back to a 100MHz BCLK, we found the board capable of supporting our DDR4-3866 memory at DDR4-4000. Note that the DRAM is set to 1.3266V, since our voltmeter showed that the board has a +25mV offset at the DIMM slots (which the board only reports as an approximate 17mV offset).

The Strix Z370-E Gaming includes a full set of primary, secondary, and tertiary timing adjustments for memory, along with a few related settings to help tuners find their best memory performance at any given data rate.

The CPU overclocking limitation was found to be within the Strix Z370-E Gaming’s Load-Line Calibration limits, where Level 6 would allow CPU core voltage to slowly drop to 1.28V. The next setting, Level 7, caused it to spike to 1.350V before dropping, and that much core voltage causes our CPU to reach its thermal limits and throttle back.

We said on page 1 that the front PCIe 3.1 Gen2 output can be enabled by disabling the two PCIe x1 slots directly beneath the two main graphics card slots, and that SATA ports 5 and 6 can be enabled by reducing the second M.2 slot to four lanes. This is indeed the default configuration. If you want those two x1 slots, or you’d like to give up two SATA ports to boost bandwidth to the second M.2 slot, you can do it from the Advanced/Onboard Devices Configuration menu. Having said that, the entire chipset is fed by only four lanes, and the upper M.2 slot is x4 by default.

Since four of the Strix Z370-E Gaming’s fan headers are paired, it has only four control groups. These can still be switched between PWM and voltage control mode, and the fan tuning algorithm seen in software is also found in firmware.

Built-in tools include a firmware-flashing utility, secure erase for drives, storage for up to eight sets of firmware settings with the ability to import and export these profiles, and a utility for reading DRAM information from its SPD IC.

MORE: Best Motherboards

MORE: All Motherboard Content

How We Test

We initially upgraded our test bed during the X299 launch to handle the tremendous heat produced by the Core i9-7900X. Our award-winning Fractal Design S24 liquid cooler system sample serves the same purpose for the newer, lower-heat Core i7-8700K.

Sound	Integrated HD Audio
Network	Integrated Gigabit Networking
Graphics Driver	GeForce 382.53

Cooler Master’s HAF-XB provides an optimal layout to blow the Celsius S24’s fans sufficiently over each motherboard’s voltage regulator.

Comparison Products

Asus ROG Strix Z370-E Gaming

MSI Z370 Gaming Pro Carbon AC

EVGA Z370 FTW

Today, we compare our entire range of previously reviewed $200-to-$250 motherboards to the Strix Z370-E Gaming. As indicated by the table, all six boards have similar overclock settings.

Benchmark Settings

Synthetic Benchmarks and Settings
PCMark 8	Version 2.7.613Home, Creative, Work, Storage, Applications (Adobe & Microsoft)
3DMark 13	Version 4.47.597.0Skydiver, Firestrike, Firestrike Extreme Default Presets
SiSoftware Sandra	Version 2016.03.22.21CPU Arithmetic, Multimedia, Cryptography, Memory Bandwidth
DiskSPD	4k Random Read, 4k Random Write128k Sequential Read, 128k Sequential Write
Cinebench R15	Build RC83328DEMOOpenGL Benchmark
CompuBench	Version 1.5.8Face Detection, Optical Flow, Ocean Surface, Ray Tracing
Application Tests and Settings
LAME MP3	Version 3.98.3Mixed 271MB WAV to mp3: Command: -b 160 --nores (160Kb/s)
HandBrake CLI	Version: 0.9.9Sintel Open Movie Project: 4.19 GB 4k mkv to x265 mp4
Blender	Version 2.68aBMW 27 CPU Render Benchmark, BMW 27 GPU Render Benchmark
7-Zip	Version 16.02THG-Workload (7.6GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=9"
Adobe After Effects CC	Release 2015.3.0, Version 13.8.0.144PCMark-driven routine
Adobe Photoshop CC	Release 2015.5.0. 20160603.r.88 x64PCMark-driven routine (light and heavy)
Adobe InDesign CC	Release 2015.4, Build 11.4.0.90 x64PCMark-driven routine
Adobe Illustrator	Release 2015.3.0, Version 20.0.0 (64-bit)PCMark-driven routine
Game Tests and Settings
Ashes of Singularity	Version 1.31.21360High Preset - 1920x1080, Mid Shadow Quality, 1x MSAACrazy Preset - 1920x1090, High Shadow Quality, 2x MSAA
F1 2015	2015 Season, Abu Dhabi Track, RainMedium Preset, No AFUltra High Preset, 16x AF
Metro Last Light Redux	Version 3.00 x64High Quality, 1920x1080, High Tesselation, 16x AFVery High Quality, 1920x1080, Very High Tesselation, 16x AF
The Talos Principle	Version 267252Medium Preset, High Quality, High Tesselation, 4x AFUltra Preset, Very High Quality, Very High Tesselation, 16x AF

MORE: Best Motherboards

MORE: All Motherboard Content

Benchmark Result & Final Analysis

One note: Under Prime95 small FFTs, the board would spike to around 194W and hold momentarily due to current-limit throttling, before power-limit throttling would cause it to drop to 144W at 4.0GHz. Since some of the other boards were holding the CPU up at 4.30GHz by default, the only way to make this a direct comparison was to disable power-limit throttling.

Disabling both limits allowed system power to climb to 210W, with a reported CPU TDP of 160W. Since we were seeing the proper Turbo Boost frequencies for various numbers of loaded cores, we considered this “good enough” to get an accurate comparison.

Synthetic Benchmarks

The Strix Z370-E Gaming has a 100.3MHz BCLK, which may account for its small lead in 3DMark. Surprisingly, that doesn’t seem to have as much impact on Sandra’s Arithmetic and Multimedia scores.

The Strix Z370-E Gaming's gains in Sandra Cryptography are more closely related to its memory-bandwidth score than its minuscule CPU overclock. Advanced memory timings are the one place a manufacturer can attempt to seek performance gains.

3D Games

Memory performance really is a big deal in some benchmarks, such as our lower test settings of Ashes of the Singularity and, more acutely, F1 2015. We’re somewhat surprised to see such small changes in bandwidth show up at the lower settings of Metro and Talos, however.

Timed Applications

Less work time means better performance in most applications. The Strix Z370-E Gaming wasn’t consistently faster or slower than its competitors across our mixed workloads or Adobe Creative Cloud, but it did lag trivially behind in Office.

Power, Heat, & Efficiency

Our EVGA Z370 FTW review showed similar power behavior to the Strix Z370-E Gaming, yet increasing its limit to allow Prime95 small FFTs to run at a constant 4.30GHz CPU clock caused peak wattage to climb only from 129W to 148W. The Strix Z370-E Gaming went from 144W to 210W. Most of the competing boards fall somewhere in the middle.

More watts means more heat, generally speaking, but the Strix Z370-E Gaming stays reasonably cool even with its power limits removed.

Without CPU throttling, the Strix Z370-E Gaming’s efficiency is on par with that of the ASRock Z370 Taichi.

Overclocking

Most boards near the Strix Z370-E Gaming’s price will reach a CPU ratio of 48x at 101MHz BCLK with 100% stability (even Prime95 small FFTs) at 1.30V CPU core, a point that our CPU has been unable to cross due to its thermal limits. Two boards were slightly behind that high mark, and the Strix Z370-E Gaming was one of these, in our tests. At least it pushes DDR4-4000…

A few generations ago, several motherboard manufacturers were using quite loose advanced timings to reach DRAM overclocking records without regard for performance. Because the goal of most overclockers is to improve performance, we’ve added a chart to show how well the modules work at these high settings. This Asus board does well, as does the EVGA and the Gigabyte.

Final Analysis

Certain readers want to see some kind of value chart, even though such performance-per-dollar charts can’t accommodate changes in features. Having noted that, the Strix Z370-E Gaming’s feature set is fairly solid for the sub-$220 market. That makes sense since, at $210, it sits near the top of that market.

Priced around $10 lower than the Strix Z370-E Gaming, MSI’s Z370 Gaming Pro Carbon AC lacks the Asus board’s second USB 3.1 Gen2 controller. The value of that controller is probably a bit more than the price difference , yet those who don’t have the required case might not want to pay for the extra controller. Concerning value, your need for that interface should determine the winner.

For around $10 more than the Strix Z370-E Gaming, ASRock’s Z370 Taichi gives users dual Gigabit Ethernet, an added SATA controller, and a two-digit status code display. It also does x8/x4/x4 from the CPU across its x16-length PCIe slots, which would be a huge boon to storage geeks, but that configuration also prevents it from supporting SLI when a card is installed in the third slot. SLI requires at least eight lanes per card. But maybe you are a storage geek? Or maybe you aren’t going to use that slot? It’s a blessing or a curse, depending on your configuration.

The most feature-packed of the competitors, Gigabyte’s Z370 Aorus Gaming Seven has dual Gigabit Ethernet and dual USB 3.1 Gen2 controllers, but no onboard Wi-Fi module. This might look like a fine board for buyers who eschew Wi-Fi on desktops, but it’s also priced $40 more than the Strix Z370-E Gaming. Show builders will appreciate the extra onboard RGB lighting, overclockers will appreciate that we reached a slightly better CPU clock, and coupon cutters will be able to appreciate that it's frequently discounted to $220 or less.

That leaves the Strix Z370-E Gaming at roughly a tie with the major competitors outlined here, apart from a very, very minor overclocking deficit that won't matter to any but the most tweaking-minded. Once again, we’d make our choice based on interface preferences, but this is a solid board that

MORE: Best Motherboards

MORE: All Motherboard Content

How to Choose the Right CPU for 1080p Gaming: 14 Processors Compared

palcorn@outlook.com (Paul Alcorn) — Fri, 23 Mar 2018 18:30:00 +0000

Introduction

Last year, we saw innovation from AMD and Intel that we hadn't experienced in a long, long time. The Zen architecture made AMD competitive in segments of the CPU market it previously couldn't touch, and Intel moved as quickly as possible to defend its incumbent position. We thoroughly enjoyed the back-and-forth as both companies jockeyed for enthusiasts' adoration.

But even as new platforms were springing up with more PCIe connectivity than ever before, graphics-card availability dried up. Cryptocurrency miners bought up everything they could find to capitalize on rising valuations. Even today, you can't find modern models anywhere near their suggested retail pricing. We've resorted to buying pre-built systems and scouring the forums for previous-generation cards, trying to score a bargain.

The best spread of CPU technology in ages, paired with sky-high GPU prices, is a recipe for confusion for PC builders. For the same amount of money, enthusiasts can afford less graphics performance than they could not long ago. That makes it easy to overspend on host processing, since balance is thrown out of whack. But you can also get a lot more CPU for your dollar than this time a year ago. How do you make sure you're getting the most for your budget?

Well, we set out on a mission with 14 CPUs and three different GPUs to find the best combination in nine popular games.

Moving The Goalposts

For the last 11 years, Core i7 and Core i5 CPUs featured four cores. Coffee Lake changed this. Now, Core i3s sport four cores, Core i5s include six cores, and Core i7s boast six Hyper-Threaded cores. Intel also gave its low- and high-end models a makeover: Skylake-X stretches up to 18 cores/32 threads for high-end desktops, while Pentium processors have now gained Hyper-Threading technology.

Of course, AMD introduced its line-up of Ryzen 7, 5, and 3 models with copious core counts. Moreover, the Ryzen Threadripper series landed with up to 16 cores/32 threads and such friendly prices that Intel was forced to make its Skylake-X chips more affordable.

The Contenders

As you might imagine, the old rules of picking a CPU family to go with certain graphics cards changed as a result of these new processors. Thus, we decided to investigate using the best performers from each CPU class.

Representing AMD, we have the Ryzen 7, 5, and 3 models. We didn't bother testing last-generation Bulldozer-based CPUs, but we did throw in the Ryzen Threadripper 1950X for good measure.

For Intel, we have K-series Core i7, i5, and i3 CPUs from the Coffee Lake and Kaby Lake generations. We also added the Core i9-7900X and Core i9-7980XE to cover high-end desktops. Out of curiosity (or because we're gluttons for benchmarking punishment?), we couldn't help but include two of the latest Pentium processors, too.

That gives us 14 processors spread across five test platforms. We paired these with the GeForce GTX 1080, GTX 1070, and GTX 1060 (6GB) graphics cards. Although the GTX 1080 is considered an extravagance these days, we have to imagine it'll come down in price someday.

Finally, we selected nine games for testing. Some of the titles are new, while others are older. We did weigh the suite, though, more toward modern games. Some of them are CPU-dependent, others are decidedly graphics-bound, and a few are actually pretty well split down the middle. This allows us to explore bottlenecks from different angles.

Today's tests are all run at 1920x1080. (We have more data coming at 2560x1440 and 3840x2160, too, so expect a follow-up story or stories to present our findings there.) To best represent the experience we'd want to have, all benchmarks were run with the highest graphics settings possible.

Test Systems

To avoid variance from GPU Boost as our GeForce GTX graphics cards heat up, we use multiple runs from each benchmark in quick succession. We select the median value from the last recordings.

Test System & Configuration

Hardware

Intel LGA 1151 (Z370) @ 2666/2400AMD Socket AM4 @ 2667 Intel LGA 1151 (Z270) @ 2400AMD Socket SP3 (TR4) (In Game Mode) @ 2667 Intel LGA 2066 @ 2666 Common To AllGigabyte GeForce GTX 1060 G1 Gaming 6G 1TB 1500W (Creators Update Version 1703)

MORE: Best CPUs

MORE: All CPUs Content

Ashes Of The Singularity: Escalation

Nvidia GeForce GTX 1080

Ashes of the Singularity was one of the first titles to receive a Ryzen-optimized patch. And it appears that host processing is always going to be your bottleneck if you're using a GeForce GTX 1080 at 1920x1080. As we can see, the 18-core/36-thread Core i9-7980XE takes a healthy lead over the rest of the field, while the 10C/20T Core i9-7900X captures second place.

This benchmark scales well with increased core counts, highlighting some of the generational transitions in Intel's line-up. The 6-core/12-thread Core i7-8700K illustrates a big jump over Intel's previous-gen 4-core/8-thread Core i7-7700K, which basically ties the Core i5-8600K.

The Core i3-7350K (and every CPU below it in our chart) struggles mightily during the benchmark.

Nvidia GeForce GTX 1070

Stepping down a notch to the GeForce GTX 1070 finds Intel's Core i9-7980XE and Core i9-7900X providing nearly identical performance, while the rest of the field scales down as we'd expect.

Really, the Core i7-7700K and most processors under it offer nearly the same performance as they did paired up to a GeForce GTX 1080. We do measure a 2.3 FPS delta between the i5-8600K with both graphics cards, but that appears to be an exception. Such small variances hardly make it worth stepping up to the GeForce GTX 1080 in this title at this resolution.

Nvidia GeForce GTX 1060

The limitations of a GeForce GTX 1060 become clear, even in this CPU-bound test, though Intel's Core i9-7980XE still maintains a healthy lead over the rest of the field.

At this point, a graphics bottleneck is emerging, pushing the GTX 1060 to its limit. As a result, the Core i9-7900X and Core i7-8700K fall a few positions. It's possible that a lighter workload isn't pushing the execution cores hard enough for them to maintain their highest Turbo Boost bins. If true, the uncore frequency drops as well, slowing the cache and ring bus frequencies, reducing performance throughout. We'll see this phenomenon in other titles, too.

Compared to our GeForce GTX 1080 results, we see similar performance from every CPU beneath the Core i5-8600K, showing that lower-end processors are still the bottleneck.

MORE: Best CPUs

MORE: All CPUs Content

Civilization VI Graphics

Nvidia GeForce GTX 1080

Intel's Core i9-7900X leads this benchmark with an impressive 106.6 FPS. The remainder of the results are uneventful, generally following trends we've observed through past CPU reviews. The Core i7-7700K averages 102.8 FPS, and the nearest mainstream Ryzen competitor achieves 80.1 FPS.

We tested the Threadripper 1950X in Game Mode simply because there wasn't enough room in our charts for the many combinations that AMD enables. Head over to our Threadripper 1950X Game Mode, Benchmarked article if you're interested in the impact of the various modes on Threadripper's performance. Just be aware that the processor behaves differently in many games, depending on the settings you choose.

Nvidia GeForce GTX 1070

Stepping down to the GeForce GTX 1070 exposes a graphics bottleneck with some of our faster CPUs. Intel's Core i9 models push the limits at around 94 FPS, landing within the range we'd expect for a graphics-bound situation. The rest of the field falls into line based on host processing power, though some CPUs fare worse than others.

Nvidia GeForce GTX 1060

A more definite GPU limitation irons out some of the variance between CPUs. Worthwhile scaling is no longer observable. Case in point: There is only a 6.4% delta between the Ryzen 5 1600X and Threadripper 1950X, showing how a midrange graphics card can neutralize the benefit of a high-end processor if you aren't careful to maintain balance.

Fortunately, this also means you can snag an affordable Ryzen 3 or Core i3 and enjoy almost the same performance as you would with a much more expensive model.

MORE: Best CPUs

MORE: All CPUs Content

Dawn Of War III

Nvidia GeForce GTX 1080

Dawn of War III scales well with increased execution resources. As we can see from the fastest CPUs, though, it also favors high clock rates.

Intel's Core i7-8700K reaches up to 113.6 FPS. However, its 1.5 FPS advantage over the Core i7-7700K isn't very impressive considering the Core i7-8700K boasts 50% more Hyper-Threaded cores.

Meanwhile, the $1000 Core i9-7900X doesn't offer as much performance as a $175 Core i3-8350K. This is why we don't recommend HEDT processors for purpose-built gaming rigs.

Nvidia GeForce GTX 1070

The same general scaling trends persist with the GeForce GTX 1070, albeit at a lower performance level. Intel's Core i3-8350K continues behaving like older Core i5 CPUs, while the Core i5-8600K really hits a price/performance sweet spot with this pairing.

Nvidia GeForce GTX 1060

Again, some of the highest-end processors drop a few places because they aren't utilized fully. There's only a 3.3 FPS (4%) delta between the Core i9-7980XE and the first-place finisher...but that chart-topper is a Core i5!

Again: Striking the right balance is important. Not only is it a waste to buy too much CPU if you don't need the processing power for other workloads, but you can even handicap it with a graphics card incapable of keeping it busy.

MORE: Best CPUs

MORE: All CPUs Content

Far Cry Primal

Nvidia GeForce GTX 1080

Far Cry Primal's Dunia Engine 2 responds well to high clock rates and IPC throughput. However, it also benefits when we turn off simultaneous multi-threading. As a result, you'll notice that the Core i5s are very competitive.

An increasing number of games are being tuned for quad-core CPUs, and this one is no exception. We think it'll take some time before chips with more cores start distinguishing themselves, though Coffee Lake fares well enough.

Although AMD doesn't do as well in our average measurements, its Ryzen 5 1600X and Ryzen 7 1800X processors demonstrate the least amount of frame-time variance. Rest assured that they offer a smooth gaming experience.

Nvidia GeForce GTX 1070

We observe much less variance between CPUs as we step down to a less powerful GeForce GTX 1070. The 6-core/12-thread Ryzen 5 1600X continues outmaneuvering the 8-core/16-thread Ryzen 7 1800X, while the Threadripper 1950X inexplicably jumps higher in the rankings.

Nvidia GeForce GTX 1060

We enter graphics-bound territory with the GeForce GTX 1060, which mixes up the finishing order. Notably, even though they don't lead when it comes to average frame rate, Intel's Core i7-8700K and i5-8600K provide the smoothest performance with the least frame-time variance. The Pentium models also respond well to this pairing.

AMD's Threadripper 1950X is the only processor that lands below our expectations. Then again, we've proven that Game Mode isn't the best option for this title. A bit of tuning might rectify the issue.

MORE: Best CPUs

MORE: All CPUs Content

Grand Theft Auto V

Nvidia GeForce GTX 1080

Grand Theft Auto V favors Intel architectures and, more generally, multi-core designs with high clock rates. It may seem odd that the Core i5-8600K beats Intel's Core i7-8700K, but we've observed this tendency on other occasions.

Did you notice that Core i3-8350K out there in front of the quad-core Core i5-7600K? Clearly, Coffee Lake propels the Core i3 family to new performance heights.

Nvidia GeForce GTX 1070

The finishing order remains intact as we test with Nvidia's GeForce GTX 1070, though average frame rates fall a bit. Still, many of the slower models land within 1 FPS of what we saw from the GTX 1080, suggesting that we're not seeing a graphics-bound workload. Host processing appears to be the limiting factor.

Nvidia GeForce GTX 1060

Plain and simple: Nvidia's GeForce GTX 1060 restricts performance at this point.

MORE: Best CPUs

MORE: All CPUs Content

Hitman

Nvidia GeForce GTX 1080

We began our testing with the Pentium CPUs and worked our way up, benchmarking all nine games for each processor before moving on to the next. So, you can imagine how thrilled we were when we discovered that IO Interactive implemented a 90 FPS frame cap at some point through a patch. Soldiering on anyway, we observed little scaling above the Core i9-7900X.

Intel's Pentium processors average a solid 60+ FPS, though we could see quite a bit of hitching during the test sequence. The Ryzen 3 1300X and the Core i3-7350K fare better, but you want a Ryzen 5 or a Kaby Lake-based Core i5, at the least, for smooth performance in this title.

Nvidia GeForce GTX 1070

The frame cap continues to be a killjoy, but at least a GeForce GTX 1070 and a reasonably-priced Core i5-8600K push Hitman's limits at 1920x1080. There is a bit of unevenness to report from the Ryzen 3 1300X and Pentium models, which manifests in our frame-time variance chart.

Nvidia GeForce GTX 1060

We finally duck under the frame cap, but only because Nvidia's GeForce GTX 1060 isn't fast enough.

There's only a 3% delta between the quad-core Core i3-8350K and first-place finisher. This is definitely a graphics-bound workload that only breaks apart a bit when we drop to the lowest-end CPUs.

MORE: Best CPUs

MORE: All CPUs Content

Shadow Of War

Nvidia GeForce GTX 1080

Middle-earth: Shadow of War doesn't scale as dramatically as some of our other benchmarks, and it certainly isn't as sensitive to IPC throughput and frequency as Shadow of Mordor. CPU reviews tend to focus on games that scale well with certain host processing specifications, such as core count, clock rate, cache size, or memory bandwidth. But some games just can't get enough graphics performance.

This title exhibits a 12.6 FPS (~16%) gap between the fastest and slowest processors in our pool, but every system surpasses the 70 FPS mark easily. There is a difference in smoothness between the Pentium and Ryzen 3 1300X processors, but it isn't extreme.

Nvidia GeForce GTX 1070

Naturally, a GeForce GTX 1070 is even more tied up, tightening the delta from top to bottom to just 3.3 FPS.

The Pentium processors still struggle a bit with hitching.

Nvidia GeForce GTX 1060

A GeForce GTX 1060 puts all of these CPUs on an even footing. Subjectively, there is no difference between the processors, even though our data suggests that the Core i7-8700K lags behind.

MORE: Best CPUs

MORE: All CPUs Content

Project CARS 2

Nvidia GeForce GTX 1080

Project CARS 2 is purportedly optimized for threading, and it seems to scale well with increased host-processing resources. After all, we see a 41 FPS delta between the fastest and slowest processors in our test pool. The Core i5-8600K leads, and it is followed closely by Intel's Core i7-7700K.

Nvidia GeForce GTX 1070

The Core i3-8350K shows that it's a surprisingly capable quad-core CPU, besting AMD's entire line-up. With that said, the Ryzen 5 1600X specifically offers most of what higher-end models can do, but at a much lower price.

Nvidia GeForce GTX 1060

The GeForce GTX 1060 limits performance at 1920x1080. AMD's Ryzen 3 1300X is the only processor to exhibit a marked amount of hitching during our recording.

MORE: Best CPUs

MORE: All CPUs Content

PlayerUnknown’s Battlegrounds (PUBG)

Nvidia GeForce GTX 1080

PlayerUnknown's Battlegrounds can be a frustrating title to benchmark. Its constant cadence of updates seems to introduce more problems than are solved, and we've seen visible hitching from even the fastest CPUs and GPUs.

Aside from the Pentiums, scaling in PUBG is fairly meager. AMD's Ryzen 3 1300X suffers the most jarring stuttering during playback of our recorded game.

Nvidia GeForce GTX 1070

Stepping down to a GeForce GTX 1070 confirms that we're decidedly graphics-bound. Not that this game is particularly detailed; it may just be poorly optimized. Either way, there's no real meaningful conclusion we can draw from such an outcome.

Nvidia GeForce GTX 1060

Our final test leaves little to explain: There are no meaningful differences between CPUs when you're constrained by a GeForce GTX 1060.

MORE: Best CPUs

MORE: All CPUs Content

The Best Combinations of CPU and GPU at 1080p

Boiling our testing down into one general recommendation is almost impossible, especially given the pricing of graphics cards right now. But we can identify some of the more obvious trends.

Fast-paced first-person shooters tend to prefer high clock rates and fast graphics cards. Nimble CPUs like the Intel Core i5 and i7, and AMD's Ryzen 5 and 7, fare best in our performance measurements. Some titles are easily graphics-bound, though that could be due to poor optimizations or minimal effort to parallelize the game engine. PlayerUnknown’s Battlegrounds is the perfect example. It doesn't respond at all to faster processors. It's far better to buy a higher-end graphics card if that game is your main squeeze.

Real-time strategy titles seem to respond more favorably to CPUs with lots of cores; so do many turn-based games. Of course, that genre isn't as sensitive to graphics performance because it doesn't require quick response times. But a fluid experience in RTS games is a must. Often, host processing is the bottleneck, so dialing back your GPU investment and buying a better CPU yields the best performance possible.

Charting Out The Trends: The Best Pairings

We typically plot our test results on a scatter chart with price and performance on two axes, but we aren't even going to take a stab at graphics pricing during these tumultuous times. Instead, we're providing a geometric mean of the 99th percentile frame times, representing smoothness, converted into an FPS measurement for each class of graphics card. Each game factors into this calculation except for the artificially hamstrung Hitman.

Balance is the goal here. Intel's Core i7-8700K is the first chip we think of when it comes to top-of-the-line for gaming PCs. But it doesn't always lead in our benchmarks. It would likely fare better with more heavily threaded titles tested. Otherwise, the Core i5-8600K seems to be a better value pairing for Nvidia's GeForce GTX 1080.

The GeForce GTX 1070 pairs well with either a Core i5-8600K or Ryzen 7 1800X, though the latter notably offers additional performance in other enthusiast-oriented desktop apps. We've also found that the Ryzen 7 1800X is a great chip for game streaming, so give it serious consideration for more taxing workloads. If you're looking to go Intel, aren't interested in overclocking, and want to save a few dollars, the Core i5-8400 complements a GeForce GTX 1070 well, too.

On the value scale, Core i3-8350K and Ryzen 5 1600X both pair well with a GeForce GTX 1060. If you're willing to do a bit of overclocking, though, a Ryzen 5 1600 looks even better.

Speaking of overclocking, most of the CPUs we tested can operate at even higher frequencies. AMD has a much friendlier policy here; all of its chips are unlocked. Meanwhile, Intel charges a premium for its unlocked K-series SKUs. If you want to know more about available headroom, we have plenty of test results with every processor overclocked in their individual reviews, so head to those for an idea of potential gains.

Of course, our findings would be all the more interesting if we could run out and build a well-balanced machine with affordable hardware. Here's hoping the availability of graphics cards improves sometime soon!

MORE: Best CPUs

MORE: All CPUs Content

Gigabyte X299 Designare EX ATX Motherboard Review: Solid OC, Plus Thunderbolt 3

Thomas Soderstrom — Thu, 22 Mar 2018 13:00:00 +0000

Features & Layout

Intel refers to the X299 platform as “High-End Desktop” (HEDT), but the term "high-end" means different things to different people. A gamer might think the top mainstream parts are better because they support higher overclocks, while a power user would scoff at that platform’s lack of connectivity. The overlap continues as the cheapest HEDT components don’t even have any of the features called “high end” when placed on a mainstream platform.

Conversely, some boards far exceed "high-end" expectations. And the X299 Designare EX’s feature set places it among these elite parts.

Since its metal parts are painted the same aluminum color as its plastic parts, we should probably start off by mentioning that all of the sinks and M.2 covers are aluminum. Likewise striking is an I/O shield mounted to the board and designed to press up against the edges of the hole in the back of a case, rather than snap into that hole, a feature normally reserved for OEM system components.

Specifications

The best news is that Gigabyte took nearly every criticism we leveled against its Aorus Gaming 7 and used it as an opportunity to build a board, basically, for us. That should make it a shoo-in for our highest award. Yet the X299 Designare EX is not a variation of the Aorus Gaming 7 despite design similarities, and this model must still stand on its own merits.

The headline upgrade is on the I/O panel, where we see not one but two Type-C connectors. And just when you were about to say, “So what, the Gaming 7 had 10Gb/s Type-A and Type-C ports,” you should probably notice the inputs next to those ports. Gigabyte equipped the X299 Designare EX with Intel’s latest JHL6540 Thunderbolt 3 controller, providing up to 36Gb/s of PCIe bandwidth, plus DisplayPort pass-through. Users with the right equipment get to choose whether to pass all 40Gb/s through a single cable or through two. Or you can just connect a couple of USB 3.1 Gen2 devices, with each at its full 10Gb/s bandwidth. Choices, choices…

The X299 chipset still has the same HSIO restrictions as the Z370, so the extra lanes required for the Thunderbolt 3 controller must come from somewhere. Gigabyte substitutes two USB 2.0 ports for the previously reviewed model’s USB 3.0, and that’s fine with us, since it’s the standard for keyboards and mice. More concerning to those who need a bunch of USB 3.0 is that all four ports are connected to a single chipset port through a hub. The chipset’s remaining 25 HSIO connections go to six SATA ports, a PCIe-based 2-port SATA controller, two front-panel USB 3.0 headers (four ports), a PCIe 3.0 x2 controller for front-panel USB 3.1 Gen2, an M.2 Key-E based Wi-Fi controller, the second Gigabit Eithernet controller, an M.2 NVMe slot, and two four-lane PCIe expansion card slots. But wait, doesn’t this board have three NVMe interfaces?

The first, third, and fifth expansion slots are served by CPU lanes, allowing a 44-lane CPU to support three graphics cards in x8/x16/x8 mode, a 28-lane CPU to support three graphics cards in x8/x8/x8 mode, and a 16-lane CPU to support two cards in x8/x8 mode with the third slot disabled. The second PCIe slot shares four lanes with the middle M.2 interface, so that builders can choose whether to install an NVMe drive on either interface (but not both). And the third M.2 slot takes all four HSIO resources from SATA, shutting off ports 4 through 7, so that we can’t count one without excluding the other. The fourth PCIe slot is permanently connected to the X299 PCH, but most SLI and CrossFire users will lose access to that slot when it’s covered by the graphics cooler of the card above. We’re still not sure why Gigabyte would choose to keep that slot active at all times, rather than the second slot, but trace-routing optimization probably had much to do with its decision. And we’re not sure why Gigabyte couldn’t have shared that slot with the third M.2 rather than a bunch of SATA, but ditto on what we said a sentence ago.

Apart from a usually-exposed slot being disabled for the sake of M.2 while an often-covered slot remains always-on, the X299 Designare EX has a great set of features. The bottom edge of the board is lined with front-panel audio, S/PDIF-out, digital LED, one (of two) RGBW, TPM, dual USB 2.0, two (of eight) PWM/voltage switchable fan, one (of two) thermistor, one (of two) USB 3.0, and VROC module headers. We also find a button to enable the board’s programmed 4.7GHz O/C, a reset button, a power switch, a two-digit status code display, the front panel LED/switch group header, and a CLR_CMOS button. Our greatest concern, that the USB 3.0 header will be nullified by a bottom-mounted graphics card, is small since most cases don’t have four USB 3.0 headers. And our least concern is that the front-panel audio cables of some cases won’t reach all the way to the bottom-rear corner header: Such poorly specified cables typically come up less than an inch short, and most case companies have solved that issue.

Appealing to users of open platforms and frame-style cases, Gigabyte covers the back of its X299 Designare in a steel plate that’s covered in the same aluminum-silver paint as the aluminum heat sinks and plastic trim.

On one hand, it was very thoughtful of Gigabyte to include an M.2-to-U.2 adapter in the X299 Designare EX installation kit. On the other hand, it reminds us that putting the card in the lower M.2 slot will disable four of the SATA ports. Other pieces include a cable bundler for front-panel switch and LED connections, six SATA cables (if you can use them), two RGBW splitters, two Velcro cable ties, two thermistor leads, a Wi-Fi antenna, and a high-bandwidth SLI bridge.

MORE: Best Motherboards

MORE: All Motherboard Content

Software & Firmware

Gigabyte APP Center remains the launching point for most Gigabyte apps, which we described back in our X299 Aorus Gaming 7 Review. Differences in appearance are limited to the new default color scheme, and the board even has the same Creative SoundBlasterX 720° audio suite.

The layout hasn’t changed for Gigabyte’s RGB Fusion app, but we took a few screenshots anyway, since each board has its own set of lighting zones. Lacking any LEDs for the expansion or memory slots, the X299 Designare EX has only a single zone for onboard RGBs, and a second zone for its RGBW outputs. We didn’t have any compatible peripherals to test.

EasyTun remains familiar, and the good news is that the 4.7GHz O/C, which is programed into the board’s firmware, works well. The reason we won’t be reporting a 4.7GHz “max stable O/C” in our charts is that it uses a 1.25V setting, causing our CPU to reach its thermal limit and throttle back to 4.3GHz at 1.20V when running 12 threads of Prime95 small FFTs. Since most workloads aren’t so demanding as Prime95, most users can get a mostly consistent 4.7GHz setting using Gigabyte’s method.

That’s not to say there weren’t any hiccups. Switching to ECO mode when running Prime95 caused the PC to crash. And while ECO mode did work if set before staring the Prime95 workload, switching back to Default mode caused our CPU to “stick” at 4.5GHz and 1.25V CPU core. Once again it would throttle back under Prime95 small FFTs, reaching a full 472W power consumption (from the wall) before doing so! Rebooting corrected the issue.

The Advanced CPU OC menu provides Windows access to the firmware’s full range of voltage and ratio settings. It works perfectly, at least within our ability to monitor changes (i.e., BCLK, CPU multiplier, CPU core voltage).

The DRAM ratio and XMP settings work too, but they require a reboot. DRAM timing menus are inaccessible.

Advanced Power settings are limited to CPU VCORE Loadline Calibration, better known as “vdroop compensation” in overclocking circles.

Users can also save Easytune overclocking configurations and assign hotkeys to enable these.

Clicking the little gear icon on the main EasyTune interface brings up the Hardware Monitor menu on the right edge of the screen, the image of which I split and placed side-by-side for easier viewing within this article’s format. Clicking the leftward arrow on the Hardware Monitor menu takes users to Gigabyte’s System Information Viewer fan-tuning app.

Firmware

X299 Designare EX firmware opens to the M.I.T. menu of its Classic mode, providing submenu links for overclocking, system monitoring, and fan control.

Our super-tough validation test combines with the mediocre thermal interface of Intel’s Core i9-7900X to thermally constrain our overclock to 4.4GHz at 1.15V. Higher voltage is tolerated at lower clocks, but the additional voltage needed to sustain a higher clock causes our CPU to reach its thermal limit and throttle back. The X299 Designare EX is one of several boards we’ve tested to provide a stable-enough voltage to reach the CPU’s limit.

None of the LGA 2066 processor range supports the half-ratios needed to get our CPU to its rated DDR4-3866. The X299 Designare EX treats it as DDR4-3800, and a little extra tuning proved these settings stable up to a 102MHz BCLK (DDR4-3876).

Gigabyte’s “Memory Timing Modes” include a Manual mode that adjusts all channels together, and an Advanced Manual mode for per-channel configuration. Primary through tertiary timing adjustments are available, along with channel-to-channel settings.

The Advanced Voltage Settings submenu is a launching point for another set of submenus to control CPU Loadline Calibration, CPU core voltage, Chipset, and DRAM voltages. Our CPU operated best using the firmware’s “Turbo” mode “Vcore Loadline Calibration” in conjunction with its 1.15V CPU core voltage setting.

There was an approximate 5mV bias between the front and rear DRAM voltage regulators. Two channels operated at approximately 25mV above our setting, and the other two at around 20mV above our setting. Most motherboards push more voltage than a user manually sets, to ensure stability when used with the most-inflated RAM specs. Some memory manufacturers have responded to the inflated voltage by producing RAM with even more-inflated specs, thereby requiring the use of boards with over-spec voltage. Not willing to play those games, we check the slots with a voltmeter.

The X299 Designare EX’s PC Health Status menu shows our memory operating at 1.356V when using the 1.325V and 1.330V settings. There’s a slight error in that report, as the monitor appears to read DIMM voltage only in 12mV increments.

All eight fan headers can be switched between voltage and PWM speed control. Gigabyte says that the board supports one 3-amp pump, but it doesn’t say which of the two pump headers supports it.

Found within the Peripherals menu, the firmware version of RGB Fusion allows only a single mode for all devices.

A popup menu at the bottom of the Classic Mode screen allows users to jump to Easy Mode, where basic settings such as XMP profile, boot order, and fans are still adjustable.

MORE: Best Motherboards

MORE: All Motherboard Content

How We Test

Test Hardware

Sound	Integrated HD Audio
Network	Integrated Gigabit Networking
Graphics Driver	GeForce 382.53

Comparison Products

Gigabyte X299 Designare EX

Asus Prime X299-Deluxe

MSI X299 XPower Gaming AC

Four comparison boards were chosen based on market proximity, with the $390 ASRock board treated as a discounted member of the $400+ price class. A quick look at firmware settings shows that all five boards have similar overclocking adjustments, leveling the field for that portion of today's comparison.

Benchmark Settings

Synthetic Benchmarks & Settings
PCMark 8	Version 2.7.613Home, Creative, Work, Storage, Applications (Adobe & Microsoft)
3DMark 13	Version 4.47.597.0Skydiver, Firestrike, Firestrike Extreme Default Presets
SiSoftware Sandra	Version 2016.03.22.21CPU Arithmetic, Multimedia, Cryptography, Memory Bandwidth
DiskSPD	4k Random Read, 4k Random Write128k Sequential Read, 128k Sequential Write
Cinebench R15	Build RC83328DEMOOpenGL Benchmark
CompuBench	Version 1.5.8Face Detection, Optical Flow, Ocean Surface, Ray Tracing
Application Tests & Settings
LAME MP3	Version 3.98.3Mixed 271MB WAV to mp3: Command: -b 160 --nores (160 Kb/s)
HandBrake CLI	Version: 0.9.9Sintel Open Movie Project: 4.19GB 4k mkv to x265 mp4
Blender	Version 2.68aBMW 27 CPU Render Benchmark, BMW 27 GPU Render Benchmark
7-Zip	Version 16.02THG-Workload (7.6GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=9"
Adobe After Effects CC	Release 2015.3.0, Version 13.8.0.144PCMark-driven routine
Adobe Photoshop CC	Release 2015.5.0. 20160603.r.88 x64PCMark-driven routine (light and heavy)
Adobe InDesign CC	Release 2015.4, Build 11.4.0.90 x64PCMark-driven routine
Adobe Illustrator	Release 2015.3.0, Version 20.0.0 (64-bit)PCMark-driven routine
Game Tests & Settings
Ashes of the Singularity	Version 1.31.21360High Preset - 1920x1080, Mid Shadow Quality, 1x MSAACrazy Preset - 1920x1090, High Shadow Quality, 2x MSAA
F1 2015	2015 Season, Abu Dhabi Track, RainMedium Preset, No AFUltra High Preset, 16x AF
Metro Last Light Redux	Version 3.00 x64High Quality, 1920x1080, High Tesselation, 16x AFVery High Quality, 1920x1080, Very High Tesselation, 16x AF
The Talos Principle	Version 267252Medium Preset, High Quality, High Tesselation, 4x AFUltra Preset, Very High Quality, Very High Tesselation, 16x AF

MORE: Best Motherboards

MORE: All Motherboard Content

Benchmark Results & Final Analysis

We start with synthetic benchmarks, which are typically used to diagnose performance problems or the benefits of specific hardware changes. Intel has standardized its platform to the extent that the fairest comparison will yield unnoticeable performance differences.

Synthetic Benchmarks

The X299 Designare EX slides through our synthetic tests without missing a beat, coming out completely average in a suite where standing out is usually a bad thing.

3D Games

The F1 2015 test revealed one stand-out result, which isn’t unusual except that on the X299 Designare EX, that high result wasn’t offset in the averages by a low result. At this point, it appears the game’s Ultra setting occasionally produces outliers, and the newer Gigabyte board is simply the surprise beneficiary.

The X299 Designare EX is otherwise unexceptional, which again is usually a good thing when all the underlying hardware (CPU, DRAM, GPU) is unchanged. The X299 XPower Gaming AC fell behind in Talos due to the overhead of an included program that’s Talos-compatible, and its review produced a chart that shows its performance with the same program disabled.

Timed Applications

The X299 Designare EX looks as average in our timed applications as it has in the other benchmarks. So, how does an honest board look in power, heat, and overclocking?

Power, Heat & Efficiency

The X299 Designare EX allowed our complete system to consume 246W fully loaded by default, appearing to follow the standards of voltage-regulated efficiency set by the Asus X299 Deluxe and the Fatal1ty X299 Gaming i9. Yet Intel XTU showed current limit throttling to 3.6GHz, where the competing boards were hitting 4GHz, and changing the firmware’s “Package Power Limits” from the default 140W to anything higher didn’t appear to have any effect. (Core Current Limit is already maxed by default.) Overclocking disabled this feature, but to get “stock” readings at 4GHz, we had to manually key the stock Turbo Boost multipliers for our CPU into the firmware.

While manual configuration of “CPU default multipliers” got us back to the same full-load frequency as the other boards, it also caused power use to increase from 246W to 324W. The CPU core temperature also climbed from 39 to 57°C and the voltage regulator from 49 to 59°C, over ambient. No wonder these companies don’t want us to use Prime95 small-FFTs!

Thanks in part to the presence of a similar board, the X299 Designare EX falls mid-pack in efficiency (our overall performance and power numbers are an average of all X299 motherboards tested, including those not used in today’s charts).

Overclocking

When we started testing X299 motherboards, we didn’t have enough products to compare additional boot straps. Those results were attained but not used until today. I personally didn’t know that they’d ever be relevant to the majority of overclockers. Without those numbers, the X299 Designare EX looks like a completely competent overclocker. So if the higher boot straps aren’t important, why bring them up now?

The X299 Designare EX became unresponsive while trying to overclock using the 1.66x boot strap, and “CLR_CMOS” didn’t fix it completely. It booted into Windows but hung. I tried everything including the slow drain and reflashing to different firmware revisions to get the board stable. The next time I was able to enter Windows, Gigabyte’s firmware flashing utility popped up. Since the system still wasn’t completely stable, I gave that a go. It hung at the point where it said “Rebooting in 5 seconds”. Windows wouldn’t load after that, and even Windows PE would crash. Fortunately, I found the solution.

You’re not going to believe this. Thinking that a part of the firmware that wasn’t being cleared was probably retaining some of the wrong settings for the 1.66x boot strap, I keyed a 101MHz BCLK into the board’s M.I.T. menu. The board made all the necessary changes for a 101MHz BCLK, and everything was stable once again. Remember, the crash had caused the system to hang even at firmware defaults, and only a manual overclock would force it to choose the right complimentary settings on its own. FYI for anyone who encounters the same bug: You're welcome.

Performance is the point of overclocking, and the X299 Designare EX takes second place out of five high-end boards in overclocked DRAM performance.

Final Thoughts

Gigabyte’s X299 Designare EX is the only comparison board to include Intel Thunderbolt 3, so a price-to-performance comparison might not even be justified within its own price class. On the other hand, the last time I checked, an onboard Thunderbolt 3 controller was roughly a $40 feature, and the Designare EX costs only $50 more than the Gaming 7.

Alternatively, the Aorus Gaming 7 includes two more of ASMedia’s high-end USB 3.1 Gen2 controllers. And the last time I checked, one of those PCIe 3.0 x2 to dual-port USB 3.1 Gen1 controllers onboard was approximately a $20 feature. That’s probably why I questioned the lack of Thunderbolt 3 on the Gaming 7.

So, the X299 Designare EX is a fine board from a features standpoint and a solid overclocker worthy at least of an informal stamp of approval. Would I choose this over MSI’s X299 XPower Gaming AC? That would probably depend on whether or not someone handed me a bunch of Thunderbolt 3 devices...

MORE: Best Motherboards

MORE: All Motherboard Content

ASRock Z370 Extreme4 ATX Review: A Solid Coffee Lake Board for Value-Minded Enthusiasts

Thomas Soderstrom — Fri, 16 Mar 2018 13:00:00 +0000

Features & Layout

Before the idea of an enthusiast motherboard broke completely away from the general consumer market and shot into the price stratosphere, $180 would get you a “high end” board that could be appreciated by gamers and power users alike.

Buyers of such boards would usually get an extra local storage controller, an extra high-bandwidth external interface, and a big voltage regulator that wouldn’t be likely to burn out when the CPU was overclocked. In more recent times, though, “high end” has since come to mean something completely different to enthusiasts. Welcome, ASRock’s Z370 Extreme4, to the enthusiast mainstream.

PCI Express storage has mostly outmoded add-on storage controllers, and sheer dollars-and-cents inflation has bumped prices upward. Yet the Z370 Extreme4 still provides a high-bandwidth add-on controller for modern external devices…but at old-school prices. What other surprises did we see on this board given the price? Let's take a look.

You also get RGB, for what it’s worth, both onboard and via a single lighting-cable header. But beyond those things, we’re basically looking at Z370 PCH functions. We still get an oversize voltage regulator for overclocking, and ASRock is thoughtful enough to provide the latest high-mainstream ALC1220 audio codec. The Z370 Extreme4 is even able to support two M.2 and one slot-mounted NVMe drive simultaneously without dipping into the CPU’s 16 PCIe lanes: It’s always nice to see a company make full use of the chipset’s limited capabilities.

The high-bandwidth interface is USB 3.1 Gen2, courtesy of ASMedia’s PCIe 3.0 two-lane controller, which feeds a Type-C and a Type-A port. ASRock wasn’t so kind as to also include a pair of USB 2.0 ports for a keyboard and mouse, which would have made sense given the Z370’s HSIO limitations, but there is a PS/2 port for your vintage keyboard or mouse. There’s even a VGA port for customers old enough to have Y2K flashbacks.

The only annoyance around the lack of additional USB ports, even if they must be USB 2.0, is that some of us like to leave the 10Gb/s ports open for portable devices. Doing so leaves only four USB 3.0 ports (a.k.a. USB 3.1 Gen1). Buyers also get DVI-D and HDMI, six analog audio jacks, and a digital optical audio output that is feature-enhanced with a DTS Connect license.

The antenna bracket resides directly behind a Key-E M.2 port, which can hold a notebook-style Wi-Fi module. We often criticize ASRock for not including antenna cables with these brackets, but the proximity should allow builders to remove the module and antenna wires from an adapter card that’s been assembled in the same manner. The benefit of using this Key-E interface isn’t just that you’ll spare a PCIe slot, but also that you won’t need to run that annoying USB internal link cable that’s required by adapter cards.

You may have noticed the Gen2 front-panel USB 3.1 header and questioned why it hasn’t been mentioned yet? That’s because it doesn’t use a Gen2 interface. Instead, it receives one USB 3.0 lane. And since USB 3.0 lanes are typically paired, the other lane feeds one of the USB 3.0 headers via a two-port hub. The other USB 3.0 header gets two lanes directly from the PCH. (For those counting HSIO resources, that’s four.)

Eight of the remaining 26 HSIO ports are allocated to two M.2 drive slots, four to the bottom x16-length slot, four to SATA, four to the USB 3.0 ports of the I/O panel, three to PCIe x1 slots, two to the USB 3.1 Gen2 controller, and one to the Key-E slot. If you’re wondering why only four are required for the SATA ports, that’s because the other four ports steal lanes from M.2 slots. Or vice versa. The answer depends on whether you’d rather use eight SATA drives or four SATA and two M.2 drives.

Fed directly by the CPU, two metal-shrouded PCIe x16 slots transition from x16/x0 to x8/x8 modes when a card is detected in the second slot. We already mentioned the four-lane bottom slot, but should add that the three x1 slots are open-ended so that longer cards can be installed. For two of those slots, the length of the card interface that fits will be determined by the length of any M.2 device mounted before it.

Along the bottom edge are a front-panel audio header, a CLR_CMOS jumper, an RGB cable header, a Thunderbolt add-in-card header, two four-pin fan headers, a Trusted Platform Module interface, three USB 2.0 headers, a serial COM port breakout header, and a front-panel button/activity LED header cluster. Two more fan headers are on the top edge, and another is slightly forward of the I/O panel audio jacks for easy reach to rear exhaust fans.

In addition to documents and software, the Z370 Extreme4 includes a case badge, an I/O shield, four SATA cables, and an HB-style SLI bridge.

MORE: Best Motherboards

MORE: All Motherboard Content

Software & Firmware

ASRock Live Update & App Shop includes a bunch of utilities, plus freeware/shareware/trial apps that we cropped out here so that you wouldn’t have to expand the image to see the important stuff. A separate tab provides easy access to ASRock’s servers for driver and software updates.

ASRock's RGB LED software hasn’t changed from previous reviews, but the board image this time shows the Z370 Extreme4’s four lighting zones. These can be synced or addressed separately, with multiple color and pulsation options, plus a color-cycle pattern that other companies often label as a "rainbow pattern."

The EZ OC page of ASRock A-Tuning pulls factory-programmed overclocking profiles from firmware, including a 4.5GHz fixed-clock setting at “Auto” CPU core, a 4.6GHz fixed-clock at 1.32V CPU core, a 4.7GHz fixed-clock at 1.36V CPU core, a 4.8GHz fixed clock at 1.39V CPU core, and a 5.0GHz fixed-clock at 1.50V CPU core. Since our CPU reaches its thermal limits under extreme loads at 1.30V and 48.48GHz, none of these settings is appropriate.

A-Tuning’s manual overclock settings worked very well through our evaluation, and they were even well-reported by its monitoring software. That’s right, by default the board sets 1.35V DIMM, and by default the result of that setting is 1.376V.

Firmware

The Z370 Extreme4 firmware opens to “EZ Mode” by default, which still allows users to pick boot order, set fan profiles, and enable or disable XMP memory settings.

Tapping the keyboard’s “F6” key loads the board’s Advanced Mode, where ASRock’s OC Tweaker menu provides the above-mentioned factory-programmed overclocking profiles in addition to a link to the CPU, DRAM, and Voltage configuration submenus, storage for up to five custom firmware settings profiles, and a shortcut to import or export one of those overclocking profiles to a USB flash drive.

Often capable of reaching 4848MHz, our CPU was slightly limited (to 4800MHz) on the Z370 Extreme4. We also encountered an issue in Firmware 1.30 where setting XMP memory mode would cause the board to force a 3.7GHz fixed clock rather than the intended overclock. Manual memory settings above DDR4-2800 caused the same issue. Experimentation revealed that these memory setting changes caused an error in the board’s Intel Turbo Boost configuration, which was correctable by disabling Intel Turbo Boost to cancel the error. That, in turn, would disable the custom multiplier. Re-entering the custom multiplier would re-enable Intel Turbo Boost, without the CPU ratio error.

The Z370 Extreme4 perfectly handled the XMP value of our DDR4-3866, but overclocking beyond DDR4-3866 caused the board to configure more-relaxed advanced settings to remain stable. Since those settings caused a net loss in performance, our memory’s XMP was the board’s practical limit.

If we had infinite time, we suspect we could have found a proper set of timings to configure a perfect-performing DRAM overclock, beyond our XMP values. The Z370 Extreme4 provides a full set of secondary and tertiary timings to help with that, along with a “DRAM Tweaker” page to help us get back to the memory’s programmed settings without resetting our manual changes.

Voltage was the most likely reason we couldn’t reach the full 4848MHz capability of our CPU: The Z370 Extreme4’s “Level 1” CPU Load-Line Calibration is a little too aggressive, so that we had to use a lower baseline voltage to prevent our CPU from overheating. On the other hand, the “Level 2” setting was far too reserved, allowing CPU voltage to sag far below what was needed to run it above 4.8GHz.

Also note that the DRAM is set to 1.325V, as the Z370 Extreme4 adds ~26mV to whatever the user selects. That’s equally true of its automatic values and its XMP settings for our DRAM. We use a voltmeter to check the voltage at the memory slot, and were happy to at least see that the board’s monitoring software showed the same voltage as our meter: Some other brands go to great lengths to hide their DRAM over-voltage.

Overclockers who don’t like hitting the F6 key to enter Advanced mode will find a setting within the Advanced table of Advanced mode, from which they can change UEFI Setup Style from Easy Mode to Advanced Mode. While a simple enough thing to do, the setting gets erased when using CLR_CMOS or loading Optimized Defaults.

The Tool menu offers an RGB interface identical to that of ASRock’s control software, plus a few niceties such as an e-mail sender for tech-support questions, a utility to copy RAID drivers from the installation disc to a thumb drive, a firmware-flashing utility, a network utility that polls the server for new firmware and downloads it to your USB flash drive, a utility to copy the active firmware to backup firmware, and a network configuration utility.

Even the hardware monitor of the Z370 Extreme4’s firmware is showing realistic results for our voltage settings, and we’re really happy to see a company at least attempt to be honest about these things. Scrolling down, we find the Fan Tuning program that tests system fans to determine the appropriate lower limit and adjust fan curves appropriately, and the Fan-Tastic Tuning link for manual fan configuration.

Only the two headers marked as pump-compatible can be switched from PWM to voltage-based control mode, and both of those headers have a 1.5A (18W) rating. The other three headers are each rated to support up to 1A (12W).

MORE: Best Motherboards

MORE: All Motherboard Content

How We Test

Sound	Integrated HD Audio
Network	Integrated Gigabit Networking
Graphics Driver	GeForce 382.53

Cooler Master’s HAF-XB provides an optimal layout to blow the Celsius S24’s fans sufficiently over each motherboard’s voltage regulator.

Comparison Products

ASRock Z370 Extreme4

MSI Z370 Tomahawk

MSI Z370 Gaming Pro Carbon AC

Overclock settings for the $175 Z370 Extreme4 are very similar to those of the $220 Z370 Taichi, and approximately equaled by both MSI and EVGA boards to produce solid competition in our overclocking comparison.

Benchmark Settings

Synthetic Benchmarks & Settings
PCMark 8	Version 2.7.613Home, Creative, Work, Storage, Applications (Adobe & Microsoft)
3DMark 13	Version 4.47.597.0Skydiver, Firestrike, Firestrike Extreme Default Presets
SiSoftware Sandra	Version 2016.03.22.21CPU Arithmetic, Multimedia, Cryptography, Memory Bandwidth
DiskSPD	4k Random Read, 4k Random Write128k Sequential Read, 128k Sequential Write
Cinebench R15	Build RC83328DEMOOpenGL Benchmark
CompuBench	Version 1.5.8Face Detection, Optical Flow, Ocean Surface, Ray Tracing
Application Tests & Settings
LAME MP3	Version 3.98.3Mixed 271MB WAV to mp3: Command: -b 160 --nores (160 Kb/s)
HandBrake CLI	Version: 0.9.9Sintel Open Movie Project: 4.19GB 4K mkv to x265 mp4
Blender	Version 2.68aBMW 27 CPU Render Benchmark, BMW 27 GPU Render Benchmark
7-Zip	Version 16.02THG-Workload (7.6GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=9"
Adobe After Effects CC	Release 2015.3.0, Version 13.8.0.144PCMark-driven routine
Adobe Photoshop CC	Release 2015.5.0. 20160603.r.88 x64PCMark-driven routine (light and heavy)
Adobe InDesign CC	Release 2015.4, Build 11.4.0.90 x64PCMark-driven routine
Adobe Illustrator	Release 2015.3.0, Version 20.0.0 (64-bit)PCMark-driven routine
Game Tests & Settings
Ashes of the Singularity	Version 1.31.21360High Preset - 1920x1080, Mid Shadow Quality, 1x MSAACrazy Preset - 1920x1090, High Shadow Quality, 2x MSAA
F1 2015	2015 Season, Abu Dhabi Track, RainMedium Preset, No AFxUltra High Preset, 16x AF
Metro Last Light Redux	Version 3.00 x64High Quality, 1920x1080, High Tesselation, 16x AFVery High Quality, 1920x1080, Very High Tesselation, 16x AF
The Talos Principle	Version 267252Medium Preset, High Quality, High Tesselation, 4x AFUltra Preset, Very High Quality, Very High Tesselation, 16x AF

MORE: Best Motherboards

MORE: All Motherboard Content

Benchmark Results & Final Analysis

Synthetic Benchmarks

After testing two boards that returned some slightly lower-than-average graphics scores, we’re glad to see the Z370 Extreme4 confirm our graphics card in good order. Synthetics are great for finding bugs, after all.

The only benchmarks that really stand out for the Z370 Extreme4 are Sandra’s Cryptography and Memory Bandwidth test. It’s tied with the other ASRock board for last place, which is at least consistent: A slight deficit in Sandra Cryptography is usually caused by that test’s dependence on high memory bandwidth.

3D Games

The Z370 Extreme4 produces slightly average frames in Ashes, but performs surprisingly well at our most memory-constrained settings of F1 2015...as did its pricier sibling, the Z370 Taichi. These are the same two boards that finished dead last in Sandra Memory Bandwidth, and the contradiction is a little startling.

The Z370 Extreme4 finishes our gaming suite about average in overall performance. The odd Talos results for one of the competing boards is attributable to an included third-party program, as explained in its review.

Timed Applications

Both ASRock boards show slight deficits in our 7.Zip benchmark, but appear average across our timed workloads as a whole.

Power, Heat & Efficiency

Our use of Prime95 small-FFTs hasn’t always met the approval of manufacturers, as the load level on AVX registers far exceeds most real-world applications. It’s a good way to make sure a product is 100% stable under any stress level, though, and a good way to find the maximum power consumption.

Intel specifies that Turbo Boost drops from 4.7GHz under single-core loads to 4.3GHz under six-core loads, and all five boards are configured to do that in our tests. On the other hand, the lower clock is supposed to allow the CPU to run at lower core voltage, and ASRock’s competitors appears to scale back their voltage a little more. Remember, we checked to make sure they were all running at 4.3GHz, and all power-saving features were enabled for this test.

The Z370 Extreme4 runs a little warm and…who are we kidding? That voltage regulator is hot. We added a fan over it prior to overclocking assessment.

Our percent charts are shown against the average of every Z370 board we’ve ever reviewed, which is a little more than twice as many as the ones shown. That average includes a couple of terribly inefficient boards that push the Z370 Extreme4’s rating up, but it also pushes up the ratings of the other boards. At full load, it slurps watts.

Overclocking

The amazing thing in our overclocking test is that the low-cost Z370 Tomahawk holds its own against our top boards. The Z370 Extreme4 is close enough for horseshoes, but it hits only the outer bull in darts.

The Z370 Extreme4 performed best at our memory’s DDR4-3866 XMP value. MSI and EVGA scaled a little higher, and EVGA also had some memory-timing advancements that put it far ahead in the overclocked memory-bandwidth test.

Final Thoughts

The Z370 Extreme4 falls exactly between the Z370 Tomahawk and Z370 Gaming Pro Carbon AC in price, and that’s also where it ends up in our Performance Per Dollar chart. Is it also halfway between these boards in worth?

All three boards have a single USB 3.1 Gen2 controller. ASRock adds a front-panel “Type-C” cable header to the Z370 Extreme4, but it’s not connected to any Gen2 controller. ASRock also adds a VGA output, while the two MSI boards offer DisplayPort. But don’t jump to any conclusions yet: The Z370 Tomahawk’s three x16-length slots are equally deceiving, having 16/x4/x1 bandwidth. At least the Z370 Extreme4 can flip its top two slots from x16/x0 to x8/x8 for SLI compatibility and full dual-card graphics performance, and at least its bottom slot provides an always-active x4 interface so that users can match an NVMe drive placed there with two others placed in the M.2 slots.

For more comparison, the Z370 Gaming Pro Carbon AC adds a high-end Intel Wi-Fi module on a PCIe adapter. That alone is worth the price difference, even though the layout of the board makes us reticent to install it. And because that module is removable from the card (complete with antenna wires), the best deal would probably be if an MSI fan teamed up with an ASRock fan to sell him that card for half the price difference. The Z370 Extreme4 has a perfect place to install that module and its antenna wires. The MSI fan could brag that he got an extra 48MHz out of his overclock, the ASRock fan could brag that he got an extra feature for the same money, and both participants in that transaction would win.

In sum, the Z370 Extreme4 is at least an average value compared to the two MSI boards that already received our stamp of approval. Accordingly, the Z370 Extreme4 gets the same.

MORE: Best Motherboards

MORE: All Motherboard Content

Asus Prime X299-A Motherboard Review: Primo Value for Core X?

Thomas Soderstrom — Mon, 12 Mar 2018 13:00:00 +0000

Specifications & Layout

December 2017 saw emerge what was the one of the most remarkable values in high-end desktop (HEDT) motherboards: The original $260 discount price for Asus’ Prime X299-A dropped to a mere $200. That put the board in Z370 territory, and with added features that would have done a $200 Z370 board proud.

Now, those add-ons are in addition to the benefits X299 offers high-end desktop (HEDT) buyers, including a range of high-core-count processors that boast up to 2.75x the PCI Express (PCIe) lane count. Nobody who is truly serious about building out a high-end PC wants to live within the confines of the Z370 platform’s 16 CPU lanes and four-lane chipset link, and that discount created an opportunity for many buyers to step up to the big leagues of X299.

The short-lived $200 price is history, and we’re not even sure if the $260 price will ever make a comeback. (At this late February 2018 writing, it was hovering in the $275 to $300 range from a variety of notable resellers.) Does the Prime X299-A remain our X299 "value" choice at current pricing? Let's dig in.

Front-panel USB 3.1 Gen 2 is the primary standout on the Prime X299-A’s specifications list. Driven by a second high-end ASM3142 controller, the Prime X299-A is the least expensive X299 board we’ve tested to include it.

Motherboard companies tend to rely on a single circuit board to address lower market segments by leaving off specific controllers and connectors, so we’re surprised to see no solder points within the gaps of the Prime X299-A I/O panel. Since USB 2.0 doesn’t require any of the PCH’s limited HSIO resources, yet is still adequate for keyboards and mice, we're happy to see two of those ports for a keyboard and mouse, in addition to the four USB 3.0 ports, the two USB 3.1 Gen 2 ports with Type-A and Type-C connectors, a USB BIOS Flashback button, the five analog audio jacks, and the single optical S/PDIF output. The digital audio output is bolstered by Asus’ inclusion of DTS Connect, which allows the ALC1220 to combine live multichannel streams to a single output. Without that technology, users of digital receivers would be forced to choose either stereo output for games and multichannel for pre-encoded recordings (such as movies), or resort to a bunch of analog cables.

We need to throw up a caution flag to anyone excited by the Prime X299-A’s slot arrangement: Examine the details in the manual before making any assumptions. For example, the x4 slots: According to Asus, the first x4 slot supports only one lane, and is available with either a 44-lane CPU or if the bottom x16 slot is disabled. (That’s partly because the third x16 slot is x8 only when paired with a 44-lane CPU, whereas both the first x4 slot and the third x16 slot share a lane from the PCH when paired with a 28-lane or 16-lane processor.)

The second x4 slot gets four lanes from a 44-lane CPU. Lacking that, it converts HSIO resources that were previously reserved for SATA to PCIe lanes. That means that users of 28-lane or 16-lane CPUs must choose between using it, or using SATA ports 5 through 8. We’re not sure why the two x4 slots are “swapped” in this manner, but the practical implication is that even users of 44-lane processors can’t use all 44 lanes to feed three double-slot graphics cards and a slot-mounted PCIe x4 SSD, as the slot with four lanes will be covered up, and the uncovered x4 slot will have only one lane. We urge readers likely to push the limits to pore over the motherboard’s documentation.

A portion of the PCH heat sink comes off to reveal the primary M.2 slot, which, unlike the vertical slot at the top of the board, supports SATA mode in addition to the expected PCIe x4 mode. Also shown in this closeup are “3D Mounts,” meant for printable covers for which Asus hosts the models, two thermistor headers, the VROC interface (which enables software RAID for PCIe drives connected to the CPU), a header for a Thunderbolt upgrade card, and an AC97-style front-panel LED and switch section. The latter has four additional pins set to the right for PC Speaker and AT-style power LEDs. Also heer: three of the board’s seven four-pin fan headers, the solitary USB 2.0 header, one of the two USB 3.0 headers, a five-pin header for an Asus fan extension card (sold separately), an onboard power button, a Port 80 diagnostics display, a legacy serial COM port, and one of the two RGB headers.

The Prime X299-A comes with four SATA cables, a support for a vertical M.2 drive mount next to its primary power connection, an HB-style SLI bridge, and a front-panel extension block that makes it easy to unplug and re-plug the switches and activity LEDs simultaneously. Also in the box: an I/O shield, a driver disc, the manual, and a coupon code for cablemod.com.

MORE: Best Motherboards

MORE: All Motherboard Content

Software & Firmware

The Asus software suite is carried over from our Prime X299-Deluxe review, though this less-expensive model doesn’t have the “Live Dash” display for which the Deluxe version had additional software. Even the “Dual Intelligent Processors 5” tuning suite is unchanged, though the Aura interface appears slightly different when paired with the connectors of this model.

Both models featured DTS connect, DTS Headphone X, and DTS Audio for External Speakers, but we’ve added a few screenshots missing from the earlier review.

Firmware

The reason we’re so happy that the software suite was already detailed in the earlier X299-Deluxe review? The Prime X299-A’s firmware requires a large load of screenshots. It opens to EZ Mode for amateurs. It doesn’t show the actual clock settings here, but you do get XMP and boot-priority selection, along with popup menus for fan control and monitoring.

Pressing the keyboard’s F7 key brings users to the Prime X299-A’s Advanced Mode menus. Here, things like overclock and power settings are accessible. Users will also find a setting to change to Advanced Mode as an initial entry point during consecutive boots.

Our test CPU is thermally constrained to overclocking at no more than 1.15V CPU core, and it typically reaches 4.4GHz at that voltage. The Prime X299-A got us there without any additional fuss. DRAM voltage was a smidge less consistent, as we noted a 5mV bias between banks where one ran around 14mV and the other around 19mV higher-than-set, as revealed by a voltmeter rather than the internal reading.

The Prime X299-A supports a full set of manual adjustments for primary, secondary, and tertiary memory timings. We test motherboards at XMP-3866 default timings, and the board reached only DDR4-3696 without throwing up memory errors under extended testing.

The Prime X299-A includes memory submenus for ODT Control, Skew Control, IO Control, and TRL IOL Control. Testing these settings could have added weeks to the review time, but we’re sure someone with the right head and heart for the task will find some of these menus useful.

CPU Load-Line Calibration allows users to offset “vdroop” by telling the board to dial up CPU core voltage as it sags under load. The “Auto” setting was perfect for our CPU sample and overclocking voltage.

We didn’t have to manually set maximum power to achieve our overclock, but did it anyway out of habit.

Still within the Ai Tweaker overclocking menu, the Asus Tweaker’s Paradise submenu provides additional clock and voltage controls.

Though a fan-menu popup is also available via hotkey, the settings are included at the bottom of the Prime X299-A’s Monitor menu. All seven fan headers can be switched between PWM and voltage-based fan control as desired, and the board also has automatic mode detection.

The “Tool” menu includes an integrated firmware update utility, eight spaces to store custom firmware settings as user profiles, and a handy tool to display information stored on the SPD IC of memory modules. EZ Flash 3 now includes a working function for polling Asus servers for new updates, and Asus Overclocking Profile adds the ability to transfer profiles to a USB flash drive.

MORE: Best Motherboards

MORE: All Motherboard Content

How We Test

Test Hardware

Sound	Integrated HD Audio
Network	Integrated Gigabit Networking
Graphics Driver	GeForce 382.53

Comparison Products

Asus Prime X299-A

MSI X299 Gaming Pro Carbon AC

ASRock X299 Taichi

A quick look at firmware settings shows that all four boards in today's charts have similar overclocking adjustments, leveling the field for that portion of today's comparison.

Benchmark Settings

Synthetic Benchmarks & Settings
PCMark 8	Version 2.7.613Home, Creative, Work, Storage, Applications (Adobe & Microsoft)
3DMark 13	Version 4.47.597.0Skydiver, Firestrike, Firestrike Extreme Default Presets
SiSoftware Sandra	Version 2016.03.22.21CPU Arithmetic, Multimedia, Cryptography, Memory Bandwidth
DiskSPD	4k Random Read, 4k Random Write128k Sequential Read, 128k Sequential Write
Cinebench R15	Build RC83328DEMOOpenGL Benchmark
CompuBench	Version 1.5.8Face Detection, Optical Flow, Ocean Surface, Ray Tracing
Application Tests & Settings
LAME MP3	Version 3.98.3Mixed 271MB WAV to mp3: Command: -b 160 --nores (160 Kb/s)
HandBrake CLI	Version: 0.9.9Sintel Open Movie Project: 4.19GB 4k mkv to x265 mp4
Blender	Version 2.68aBMW 27 CPU Render Benchmark, BMW 27 GPU Render Benchmark
7-Zip	Version 16.02THG-Workload (7.6 GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=9"
Adobe After Effects CC	Release 2015.3.0, Version 13.8.0.144PCMark driven routine
Adobe Photoshop CC	Release 2015.5.0. 20160603.r.88 x64PCMark driven routine (light and heavy)
Adobe InDesign CC	Release 2015.4, Build 11.4.0.90 x64PCMark driven routine
Adobe Illustrator	Release 2015.3.0, Version 20.0.0 (64-bit)PCMark driven routine
Game Tests & Settings
Ashes of the Singularity	Version 1.31.21360High Preset - 1920x1080, Mid Shadow Quality, 1x MSAACrazy Preset - 1920x1090, High Shadow Quality, 2x MSAA
F1 2015	2015 Season, Abu Dhabi Track, RainMedium Preset, No AFxUltra High Preset, 16x AF
Metro Last Light Redux	Version 3.00 x64High Quality, 1920x1080, High Tesselation, 16x AFVery High Quality, 1920x1080, Very High Tesselation, 16x AF
The Talos Principle	Version 267252Medium Preset, High Quality, High Tesselation, 4x AFUltra Preset, Very High Quality, Very High Tesselation, 16x AF

MORE: Best Motherboards

MORE: All Motherboard Content

Benchmark Results & Conclusion

The Asus Prime X299-A, like its “Deluxe” sibling, exhibited some low power-use numbers under Prime95 with small-FFTs. The reason it appears so much lower than typical boards is that Asus adheres to Intel’s stated 140W TDP for our Core i7-7900X test CPU. The problem that leaves us with is that this causes the CPU to throttle back under Prime95 more than any other program, down to 3.4GHz, complicating any efficiency calculations.

Disabling that limit, though, is as easy as changing “Overclock Mode” from “Auto” to “Manual”. Doing that without actually picking an overclock resulted in the CPU reaching 3.6GHz under full load, which is similar to other boards we've tested. Intel XTU is no longer able to detect the TDP in this configuration, and some performance benchmarks are slightly boosted.

Here, we'll present the X299-A in both configurations, affixing the label “W/O 140W TDP” to the tests that have “Manual OC Mode” enabled without a set overclock. As with the other boards, both settings have all of the CPU’s other power-savings features enabled and turbo-boost “enhancement” disabled.

Synthetics Benchmarks

The Prime X299-A is roughly on par with its competitors across 3DMark, though the difference between its stock 140W TDP configuration (at the top) and its competition-matched power setting (at the bottom) is more pronounced in PCMark.

The Sandra Memory Bandwidth benchmark, meanwhile, drops a little with the stock TDP limit disabled, in spite of the benefit seen in Cinebench.

3D Games

There’s scant benefit to disabling the 140W stock TDP setting in games, though the difference is at least measurable in Ashes and F1 2015.

The odd Talos results for the competing X299 Gaming Pro Carbon AC are due to the game's compatibility with its Nahimic software, which imparts some overhead. Disabling the software would have put that board in first place.

Timed Applications

Our Blender CPU rendering and 7.Zip tests both benefitted from disabling the Prime X299-A’s default 140W TDP, which was also supposed to be the default for our CPU. Since the other boards don’t have such strict TDP limits, the alteration is necessary to put the board on equal performance footing.

Though the altered power setting did not make the Prime X299-A consistently faster, it helped noticeably in Adobe Illustrator.

Power, Heat & Efficiency

Now it’s time to come clean: Power measurements are taken after the system has “settled into” its workload, as spikes are otherwise too inconsistent to provide meaningful results. The reason that’s more important in today’s discussion is that the Prime X299-A would spike to 246W at its default settings, before its 140W TDP would pull it back. Intel says that’s supposed to happen. Without that restriction, it climbs only to a steady 247W, matching the X299 Taichi.

The 140W TDP is supposed to prevent the CPU from overheating, but our cooling configuration is so large that the difference was only 4°. The voltage regulator ran 6° warmer. Gigabyte’s sample had a cooler voltage regulator, ASRock’s voltage regulator ran hot, and MSI’s pushed CPU temperature through the roof. So this board keeps things modest.

The Prime X299-A defaults cause the CPU to throttle down primarily during our power test, so the numbers shown at the bottom are closer to being fair. Even then, it’s far more efficient than the MSI and Gigabyte samples, and further edges out the ASRock board.

Overclocking

The Prime X299-A does what it’s supposed to with CPU overclocking, surpassing here the X299 Gaming Pro Carbon AC. Conversely, the Asus board had the lightest memory overclock in this test set.

Performance, though, is the whole reason for overclocking, and the X299-A has the best overclocked memory performance of the four boards shown. Perhaps Asus was aiming for best performance rather than best data rate? That’s completely achievable using nothing more than secondary and tertiary timings, but those tighter timings usually limit overclocking.

Value

At its current price, the Prime X299-A competes against the X299 Taichi from ASRock. That board has the advantages of a second Gigabit Ethernet controller and a cheap Wi-Fi module, but it lacks the Prime X299-A’s front-panel USB 3.1 Gen2 connection.

On the whole, the board lacks any key flaws, and value parity makes the Prime X299-A eligible for a stamp of approval high-five. Had it remained at $200, it would have vied for our top award, Editors' Choice. Like any other shopper, we would always love to spend less money, and even back when the board was priced at $260 (or even better, at $200 in late December), it was a fine value.

It would still be our choice had the $260 price stuck, but choosing it has become a more of a coin-toss at current pricing, given that other X299 boards fluctuate in pricing all the time, too. Still, we approve of this board, and given the vagaries of X299 board prices on any given day, it ranges from a good to a potentially excellent deal. Just watch the comparison pricing at your favorite reseller.

MORE: Best Motherboards

MORE: All Motherboard Content

Corsair Vengeance DDR4-4000 SODIMM (4x 8GB) Kit Review

Thomas Soderstrom — Tue, 27 Feb 2018 02:00:00 +0000

The Modules & The Math

Every so often a new memory kit comes out that makes us go "hmmmm," yet we can always find at least one benchmark to justify its existence. Such was the case with our recent DDR4-3800 review, and Corsair would like to show us a better path with its DDR4-4000 SODIMMs.

We should qualify that these short modules will only work with K and X series desktop processors, and that while we’re sure someone has released a “mobile workstation” or “mobile server” with the right specs, we’re not certain that any of those boards will ever have the firmware functionality to support anything nearly this fast at full speed. Instead, we’re left to consider the one desktop board that is known to support DDR4 SODIMMs and overclocking: the ASRock X299E-ITX/ac.

Unlike DDR4-3800 that runs at 19x 100MHz, DDR4-4000 runs at 15x 133MHz. Lower multipliers tend to be easier to use on some processors, so there’s at least a little hope of expanded compatibility. Unfortunately, that hope fades when one realizes how few boards and CPUs can run memory at DDR4-4000 without imparting errors (signal noise or degradation) that causes such high-speed memory to crash. Once again, we’re left with Skylake-X and, thus far, the X299E-ITX/ac.

In fact, the DDR4-4000 XMP setting isn’t even shown properly in the DRAM Tweaker menu of the X299E-ITX/ac. The strange thing is that the board does show it under basic XMP settings menus. Oh, and if you can’t run DDR4-4000, the DDR4-2666 SPD value is still a good starting point.

And now for a little more math: Because latency is measured in cycles, and cycle time is the inverse of frequency, DDR4-4000 CAS 20 would only be as “quick” as DDR4-3200 CAS 16 or DDR4-2400 CAS 12. That means the rated CAS 19 should be acceptable at DDR4-4000, but the tRCD and tRP timings of 23 cycles are fairly relaxed by enthusiast standards. We’re going to try to run this stuff at CAS 19 straight timings just to see if it’s still stable and performs better.

Specifications

We’re comparing Corsair’s Vengeance DDR4-4000 to G.Skill’s DDR4-3800, using Super Talent’s DDR4-2400 as a baseline. The Super Talent modules are relabeled Crucial DIMMs, but both manufacturers offer similar prices and similar lifetime warranties to Corsair and G.Skill.

Corsair Vengeance DDR4-4000 SODIMM (4x 8GB)

G.Skill DDR4-3800 32GB (4x 8GB) SODIMM

Super Talent 8GB (32GB)

We used our standard X299 motherboard testbed to evaluate today’s modules, adding the ASRock X299E-ITX/ac to make the comparison possible. Since ASRock’s current firmware uses “stability mode” settings when pushed past XMP settings, benchmarking the G.Skill modules at DDR4-4000 required the use of ASRock’s A-Tuning utility for Windows.

Benchmark Configuration
Blender	Version 2.68a: BMW 27 CPU Render Benchmark
7-Zip	Version 16.0:, THG-Workload (7.6 GB) to .7z, cmd line "a -t7z -r -m0=LZMA2 -mx=9"
F1 2015	2015 Season, Abu Dhabi Track, Rain: Ultra High Preset, 16x AF
Metro Last Light Redux	Version 3.00 x64: Very High Quality, 1920x1080, Very High Tesselation, 16x AF
SiSoftware Sandra	Version 2016.03.22.21: Memory Bandwidth, Cache & Memory Latency Benchmarks
Sound	Integrated HD Audio
Network	Integrated Gigabit Networking
OS	Windows 10 64-bit
Graphics Driver	GeForce 382.53

MORE: Best Memory

MORE: All Memory Content

Benchmark Results & Final Analysis

Tom’s Hardware recognizes that certain XMP settings, such as nonstandard voltage, are merely old-school overclocking techniques that many of our readers are perfectly capable of doing themselves. In that spirit, we’ve attempted to push each of today’s comparison kits to its best timings at various data rates and a CPU-safe 1.35V DIMM setting.

Lowest Stable Timings at 1.35V (Max) on ASRock X299E-ITX/ac (BIOS 1.03)
	DDR4-4000	DDR4-3200	DDR4-2400
Corsair Vengeance 32GB CMSX32GX4M4X4000C19	19-19-19-38 (2T)	14-14-14-28 (2T)	11-11-11-28 (1T)
G.Skill Ripjaws 32GB F4-3800C18Q-32GRS	19-19-19-38 (2T)	14-14-14-28 (1T)	11-11-11-28 (1T)
(4) Super Talent 8GB F24SA8GM (32GB)	X	17-19-19-38 (1T)	13-14-14-28 (1T)

Super Talent’s surprisingly overclockable DDR4-2400 doesn’t support enthusiast-class timings, nor does it reach DDR4-4000. The Ripjaws and Vengeance modules both reach DDR4-4000 at the same timings, and even have the same minimum timings at lower test frequencies. Lower timings allow quicker response via fewer cycles of delay.

Both the Vengeance and Ripjaws modules also have similar overclocking capability, though the Ripjaws gets a greater percent-overclock thanks to its lower starting point. But if we were going by percent-over-stock, the Super Talent DDR4-2400’s overclock looks enormous.

Value seekers will see the DDR4-2400’s memory bandwidth at custom DDR4-3200 settings and perhaps think that they should replicate this overclock. Unfortunately, the only way you’ll get a factory-verified overclock is to buy overclocked RAM such as the Vengeance or Ripjaws. Cheapskates can’t know or even guess that they’ll be as lucky as we’ve been.

Slow stock timings (tRCD, tRP, tRAS, and tRFC) put the Vengeance DDR4-4000 behind the Ripjaws DDR4-3800 in Sandra Latency, when both are set to XMP values. They come out roughly equal in this metric when using manually optimized timings.

The Vengeance and Ripjaws kits look very similar in F1 2015, a title that’s known to be constrained by memory performance (once CPU frequency and GPU capability have been optimized). On the other hand, the cheap kit’s DDR4-3200 settings also produce strong results.

Metro: Last Light is closer to the typical gaming scenario, in that it has shown us noticeable differences between memory configurations, but only when one of those configurations is deficient (such as yesteryear’s dual-channel DDR3). Nearly any DDR4 quad-channel configuration is more than enough to satisfy its needs.

The thing we just said about Metro also applies to Blender. The funny thing is that this benchmark is too short to show real differences, which tend to be around half a second, from the best to the worst DDR4 quad-channel configuration.

Ripjaws DDR4-3800 wins the 7-Zip test at all settings, likely due to shorter secondary and tertiary timings. Vengeance is close, and it carries the DDR4-4000 certification.

Final Analysis

Fortune favors the bold, and those seeking the best value in performance memory can often find it by overclocking. On the other hand, those who’ve already made their fortune can pay someone else to do their overclocking, and that’s where premium RAM becomes valuable.

Despite the Vengeance DDR4-4000 having a slightly higher list price than Ripjaws DDR4-3800, the Vengeance kit is available at a lower price through Amazon, compared to the Ripjaws pricing we saw a few weeks back at Newegg. As we write this, the Ripjaws kit is now out of stock at Newegg, and seemingly everywhere. Still, a G.Skill rep told us that it should be back in stock in short order, and--who knows?--it's possible G.Skill may tweak the pricing in response. And then, everybody will win. (That is, everyone with more than $500 to spend on RAM.)

So, which would we choose, assuming both were on hand to choose between? Since the two kits are so similar, the bucks-up buyer would probably be best just getting the cheaper of the two and spending the savings on his or her favorite beverage.

MORE: Best Memory

MORE: All Memory Content

Gigabyte X299 Aorus Gaming 7 Pro Motherboard Review

Thomas Soderstrom — Thu, 15 Feb 2018 14:00:00 +0000

Features & Layout

Manufacturers like Gigabyte have made great strides in the X299 market in the six months following the Skylake X launch, but a lingering problem has been what they should do about the old reviews that used under-developed firmware to produce atypical performance and power data. Retesting previous samples can easily snowball into continuous retests with no further new product reviews as we attempt to be fair to everyone, but manufacturers do have a workaround: Release revised hardware under a new model name. Welcome the X299 Aorus Gaming 7 Pro.

Adding the word “Pro” to the model name gets buyers an amped-up voltage regulator (60A x 12 phases), a new set of heatsinks to cool it, two additional M.2 heat spreaders, and all six months of firmware advancement out of the box. In other words, the new model has everything that many potential buyers felt was missing during the original model’s review.

Specifications

Buyers still get five x16-length PCIe expansions slots and are still limited to 3-way SLI at most when using a 28-lane CPU at least (ie, Core i7-7800X and above). That configuration typically requires an eight-slot case, since the third 8-lane slot is at the bottom of the board. Crossfire can go more ways, but only if single-slot cards are used. Most of the SLI restrictions won’t matter at this market level since the top cards don’t support 3-way, but the second slot’s sharing of PCIe lanes with the middle M.2 connector could be an issue for any buyer who was hoping to use it. Though we’d have preferred the shared pathways be directed to a less-useful slot position (such as the fourth slot, where it’s more likely to be blocked by a graphics cooler), Gigabyte likely designed the layout for optimized trace routing.

The X299 Aorus Gaming 7 Pro also retains the previous version’s excellent I/O panel port selection, with ASMedia’s high-bandwidth ASM3142 two-port controller driving a Type-C connection from one port, and a four-port hub from its other port. While that means all four of its connected Type A ports share bandwidth, any one of these could still operate at or near the peak rate of a single port.

Four USB 3.0 ports (aka USB 3.1 Gen1) are also present, one marked in white for Gigabyte’s Q-Flash Plus updater, which can automatically update firmware from a flash drive if both on-board firmware images are corrupted or incompatible with a newer model CPU (when it didn’t automatically initiate on a previous model, we forced initiation by removing the CPU). It seems a little wasteful to us to exclude USB 2.0 entirely in favor of USB 3.0, since the former supports all keyboards and mice, while the later does so while also requiring additional resources at one HSIO pathway per port. HSIO pathways are also able to feed PCIe lanes or SATA ports, and evidence of this shortage has already been discussed in the PCIe slot vs M.2 pathway sharing mentioned above.

One of the two I/O panel gigabit Ethernet ports is controlled by Intel’s PHY, the other by a Killer E2500 PCIe-based network IC. The latter is paired with the Killer 1535 Wi-Fi controller to enable Killer DoubleShot Pro bandwidth and packet optimization.

Even the analog audio jacks are enhanced with gold-plated connectors, an ESS SABRE 9018 DAC, and Creative’s Sound BlasterX 720° enhancement software.

You may have heard most of this before, but it’s worth repeating for those who haven’t that the Aorus Gaming 7 Pro’s front-panel USB 3.1 Gen 2 header is located next to its USB 3.0 header, a fan connector, the 24-pin power connector, and another fan header within the upper half of the board’s front edge. The lower half of the same edge features eight SATA ports, which all face forward so that attached cables will fit underneath the leading edge of long expansion cards.

The bottom edge has a front-panel audio header and two-pin S/PDIF output, an RGBW header, a digital LED strip connector with 5V/12V selector, buttons for ECO and OC modes, Reset and Power, two USB 2.0 headers, two four-pin fan and a two-pin thermistor headers, a debug code port, a second USB 3.0 header, the front-panel LED/switch group, a CLR_CMOS button, and a VROC module connector for adding RAID to NVMe drives that interface the CPU’s PCIe controller.

The top edge features three fan headers, a second RGBW header, two EPS12V connectors, and an I/O connector for the I/O panel shield’s lighting. An eighth fan connector is positioned above the rear edge of the top slot for easier cabling to the rear fan of most cases, and all eight fan connectors support both automatic detection and manual selection of PWM/voltage fan control modes.

The X299 AORUS Gaming 7 Pro motherboard installation kit includes a driver disc, user’s manual, quick installation guide, four SATA cables, two RGBW LED strip extension cables, two thermistor cables, an HB-SLI bridge, an I/O Shield, a dual-channel Wi-Fi antenna, two Velcro cable ties, a G-Connector front-panel cable end bundler, a Wi-Fi antenna retention cover, and an M.2 screw kit.

MORE: Best Motherboards

MORE: All Motherboard Content

Software & Firmware

The X299 Aorus Gaming 7 Pro’s software suite hasn’t changed significantly from its non-Pro predecessor, and still includes that board’s Killer network suite with DoubleShot Pro functionality and Creative’s Sound BlasterX 720° in addition to Gigabyte utilities launched from its App Center menu. Without new features to discuss, we’ll look for changes in how the overclocking software works with the new hardware.

EasyTune still offers Core i7-7900X users a 4.7 GHz factory-programmed overclock, and that overclock crashes instantly under Prime95. The PC didn’t even stay up long enough to measure idle and load voltage levels.

Gigabyte’s AutoTuning app pushed our CPU even higher, to 4.80 GHz at 1.27V idle and 1.29V load. This could be an attempt for 1.30V, but again the system crashed before further analysis. Part of this is caused by our CPU’s inability to expel heat quickly enough to run full-AVX loads at CPU core voltages over 1.15V. Shouldn’t the load testing of Gigabyte’s AutoTuning algorithm be harsh enough to make that determination?

EasyTune’s Advanced menu provides CPU frequencies and voltage levels consistent with the board’s limitations, which can also be found within its firmware settings. Aggressive overclockers will sometimes use tuning software to make on-the-fly changes that exceed the board’s cold-boot capabilities.

Timing adjustments are disabled in EasyTune’s Advanced DDR OC menu, and frequency or XMP-mode changes require a reboot. Clicking “apply” will then initiate a persistent reboot popup.

The Advanced Power menu provides CPU VCore Loadline Calibration, which reduces the amount of core voltage “droop” that occurs under heavy loads.

EasyTune allows users to save custom overclock settings as profiles, and its “Hotkey” menu allows users to configure shortcuts to apply these.

Users are also welcome to download a smartphone app to control their overclocks externally.

Firmware

X299 Aorus Gaming 7 Pro firmware opens to Advanced mode, where its M.I.T. menu is the launch point for various overclocking and fan control menus.

Since our core can’t expel heat fast enough to use voltage settings beyond 1.15V, its limited to 4.40 GHz at full AVX load. The X299 Aorus Gaming 7 Pro gets us there, but so did the non-pro. We’d probably need to de-lid our CPU before we could push enough current through its core to take advantage of this newer motherboard’s beefier voltage regulator.

We only reached DDR4-3636 at 101 MHz BCLK on our DDR4-3866 at its XMP timings. Additional menus allow users to choose either per-channel or combined latency adjustments for primary, secondary, and tertiary timings.

The Advanced Voltage Settings submenu is yet another launching point for a group of submenus.

Our CPU stayed at our chosen 1.150V core despite changing loads after setting “CPU Vcore Loadline Calibration” to “Turbo” mode.

The X299 Aorus Gaming 7 Pro exhibited a 10mV bias between the rear and front banks, measuring between 1.350 and 1.355 volts at 1.325 and 1.335 volt respective settings.

Fans can be set to numerous manufacturer-defined RPM-to-temperature curves or custom adjusted, using either PWM or voltage control mode. Users can even select which temperature sensor the board should use to control a specific header.

Gigabyte’s RGB Fusion firmware control is found within its “Peripherals” menu, and provides a simplified version of the lighting controls also available in software.

Anyone scared off by the X299 Aorus Gaming 7 Pro’s numerous advanced-menu settings can tap the F2 keyboard key to enter Easy Mode, where boot order, XMP, and energy savings features are still accessible.

MORE: Best Motherboards

MORE: All Motherboard Content

How We Test

Test Hardware

Sound	Integrated HD Audio
Network	Integrated Gigabit Networking
Graphics Driver	GeForce 382.53

Comparison Products

Gigabyte X299 Aorus Gaming 7 Pro

Asus Prime X299-Deluxe

MSI X299 XPower Gaming AC

A quick look at firmware settings shows that all five boards in today's charts have similar overclocking adjustments, leveling the field for that portion of today's comparison.

Benchmark Settings

Synthetic Benchmarks and Settings
PCMark 8	Version 2.7.613Home, Creative, Work, Storage, Applications (Adobe & Microsoft)
3DMark 13	Version 4.47.597.0Skydiver, Firestrike, Firestrike Extreme Default Presets
SiSoftware Sandra	Version 2016.03.22.21CPU Arithmetic, Multimedia, Cryptography, Memory Bandwidth
DiskSPD	4k Random Read, 4k Random Write128k Sequential Read, 128k Sequential Write
Cinebench R15	Build RC83328DEMOOpenGL Benchmark
CompuBench	Version 1.5.8Face Detection, Optical Flow, Ocean Surface, Ray Tracing
Application Tests and Settings
LAME MP3	Version 3.98.3Mixed 271MB WAV to mp3: Command: -b 160 --nores (160 Kb/s)
HandBrake CLI	Version: 0.9.9Sintel Open Movie Project: 4.19 GB 4k mkv to x265 mp4
Blender	Version 2.68aBMW 27 CPU Render Benchmark, BMW 27 GPU Render Benchmark
7-Zip	Version 16.02THG-Workload (7.6 GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=9"
Adobe After Effects CC	Release 2015.3.0, Version 13.8.0.144PCMark driven routine
Adobe Photoshop CC	Release 2015.5.0. 20160603.r.88 x64PCMark driven routine (light and heavy)
Adobe InDesign CC	Release 2015.4, Build 11.4.0.90 x64PCMark driven routine
Adobe Illustrator	Release 2015.3.0, Version 20.0.0 (64-bit)PCMark driven routine
Game Tests and Settings
Ashes of Singularity	Version 1.31.21360High Preset - 1920x1080, Mid Shadow Quality, 1x MSAACrazy Preset - 1920x1090, High Shadow Quality, 2x MSAA
F1 2015	2015 Season, Abu Dhabi Track, RainMedium Preset, No AFxUltra High Preset, 16x AF
Metro Last Light Redux	Version 3.00 x64High Quality, 1920x1080, High Tesselation, 16x AFVery High Quality, 1920x1080, Very High Tesselation, 16x AF
The Talos Principle	Version 267252Medium Preset, High Quality, High Tesselation, 4x AFUltra Preset, VeryHigh Quality, VeryHigh Tesselation, 16x AF

MORE: Best Motherboards

MORE: All Motherboard Content

Benchmark Results & Conclusion

Manufacturers scrambled to deal with the unrealistic 140W TDP of Intel’s Core i7-7900X for a brief period before and after the X299 platform launch, with the early Asus Prime X299-Deluxe firmware getting closest to the target and the early X299 Aorus Gaming 7 farthest away. As the market starts to normalize, the X299 Aorus Gaming 7 Pro’s current firmware likely represents the new normal.

These power numbers are important because they regulate the amount of performance available under extreme CPU loads. Lower TDP settings frequently cause the CPU to throttle back in applications as harsh as Prime95’s small-FFTs, and large differences in allowed power can even affect real-world benchmarks.

Synthetic Benchmark Results

Synthetic benchmarks can identify problems, but we don’t see anything that really stands out until we get all the way down to Sandra Multimedia. Differences there appear to point to a variability in the benchmark itself, since no other set of results is so widely varied.

The X299 Aorus Gaming 7 Pro takes middle position in Sandra Memory Bandwidth. Memory timings are one of the few places where manufacturers can play with performance without overclocking the CPU.

3D Games

The X299 Aorus Gaming 7 Pro outpaces its predecessor in F1 2015, but other benchmarks are close enough to fall within the normal variation between consecutive runs.

The one standout in Talos, MSI’s X299 XPower Gaming AC includes Nahamic software that, because of its compatibility with this game, adds overhead. Disabling the software would have put it in first place.

Timed Applications

Less time means more performance in timed applications, where the X299 Aorus Gaming 7 Pro again falls to the middle of the pack.

The X299 Aorus Gaming 7 Pro edges out its predecessor in MS Office, but the difference is still rather small.

Power, Heat, & Efficiency

The X299 Aorus Gaming 7 Pro is certainly more miserly than its predecessor, but that still puts it only mid-pack in power consumption.

Fortunately, that lower power use also allows significantly lower CPU temperature.

Combining a miniscule performance gain with noticeable power savings compared to its predecessor, the X299 Aorus Gaming 7 Pro climbs to second place in efficiency.

Overclocking

Thanks to our CPU’s hard thermal barrier at relatively low voltage levels, nearly every board can push it to 4.40 GHz. That’s unfortunate because it doesn’t allows the X299 Aorus Gaming 7 Pro’s improved voltage regulator to shine.

Despite its meager memory frequency, the X299 Aorus Gaming 7 Pro is roughly tied for second in overclocked memory performance. This is where the X299 XPower Gaming AC really shines.

Final Thoughts

The X299 Aorus Gaming 7 Pro is a great looking board packed with features, but so was the X299 Aorus Gaming 7. Improvements include an upsized voltage regulator, two additional heat spreaders for M.2 drives, and firmware updates that eliminate the poor efficiency of the X299 Aorus Gaming 7’s early review. Builders can still get firmware updates for the earlier board though, so we’re really looking at a price difference of around $50 to purchase a couple chunks of aluminum and a few voltage regulation components.

The X299 Aorus Gaming 7 earned our stamp of approval in spite of a couple flaws, which were its inability to set a 0x AVX offset and its poor efficiency. Since firmware updates have addressed both issues, we’re still left to consider whether the new voltage regulator is still worth the extra money. If your CPU is thermally constrained such as ours, it probably isn’t. Yet if you’re de-lidding your CPU to eliminate such constraints, it probably is. The X299 Aorus Gaming 7 Pro is either a better or lesser value depending on what you’re planning to do with your CPU, so it gets our same nod of approval.

MORE: Best Motherboards

MORE: All Motherboard Content

G.Skill DDR4-3800 32GB SODIMM Review

Thomas Soderstrom — Fri, 19 Jan 2018 15:00:00 +0000

Features & Specifications

Enthusiasts are accustomed to exceeding expectations and blowing past common perceptions, and a few of the myths surrounding memory are among these. SODIMMs are assumed to be “notebook RAM,” but we’ve seen them in everything from compact desktops to servers. And just to flip things, we tested DDR3L long DIMMs two years before the Z170 platform launch brought that 1.35V limit to “regular” desktops. So let’s flip this again: Why can’t we have super-fast SODIMMs? When ASRock launched its X299E-ITX/ac in October, G.Skill was first to say “Yes you can!”

That we’ve had this kit for three months is no secret, as it was even used in the eventual review of that board, yet a pesky thought remained even as we moved on to other projects: Exactly how far could this RAM go? And what performance could be gained by using it?

The name “Ripjaws 4” may have had the desktop gurus expecting heat spreaders, but notebook guys know that when it comes to SODIMMs, such additions are usually just 3D-looking stickers. That’s because the modules are often spaced too closely to allow the extra thickness. G.Skill’s stickers take on a more minimal approach in that they simply appear to be exactly what they are.

We still get everything else we’d expect from a set of Ripjaws long-DIMMs in a smaller package, right down to the DDR4-3800 CAS 18 XMP profile. If your SODIMM-equipped system isn’t designed with XMP support, perhaps you should ask yourself what else it’s missing?

Comparison Products

G.Skill DDR4-3800 32GB (4x 8GB) SODIMM

Super Talent 8GB (32GB)

The baseline for this comparison will be a set of four Super Talent F24SA8GM DDR4-2400 SODIMMs. Each DDR4-2400 SODIMM is 8GB, and each uses JEDEC-standard CAS 17 timings. Even though Super Talent validates its SODIMMs four-at-once, buyers must still purchase four individual modules (at 4x the above DDR4-2400 price) to get the tested 32GB configuration.

Buyers who don’t have access to Super-Talent’s sales team will find that the firm is re-distributing unaltered Crucial DDR4-2400, right down to the 12 timing sets that range from DDR4-1866 CAS 13 to DDR4-2400 CAS 20. Lower is better when it comes to latency, since it’s measured in cycles. Cycle time is the inverse of frequency, so that DDR4-1866 CAS 13, DDR4-2133 CAS 15, and DDR4-2400 CAS 17 have approximately the same quickness.

With one 4x 8GB kit in one hand and four 8GB single modules in the other, both configurations are 32GB quad-channel, and both manufacturers provide a lifetime warranty.

Test System

We used our standard X299 motherboard testbed to evaluate today’s modules, adding the ASRock X299E-ITX/ac to make the comparison possible. Since ASRock’s current firmware uses “stability mode” settings when pushed past XMP settings, performance benchmarking at settings above DDR4-3800 required the use of ASRock’s A-Tuning utility for Windows.

Sound	Integrated HD Audio
Network	Integrated Gigabit Networking
OS	Windows 10 64-bit
Graphics Driver	GeForce 382.53

MORE: Best Memory

MORE: All Memory Content

Benchmark Results & Final Analysis

We shouldn’t be surprised that the Ripjaws DDR4-3800 reached DDR4-4000 simply by loosening the timings, but we’re a little surprised that it also reached DDR4-2400 at a mere CAS 11. Also, is anyone noticing the 33% overclock achieved on Super Talent’s CAS 17-rated DDR4-2400? Amazing.

Lowest Stable Timings at 1.35V (Max) on ASRock X299E-ITX/ac (BIOS 1.03)
	DDR4-4000	DDR4-3200	DDR4-2400
G.Skill Ripjaws 32GBF4-3800C18Q-32GRS	19-19-19-38 (2T)	14-14-14-28 (1T)	11-11-11-28 (1T)
(4) Super Talent 8GB F24SA8GM (32GB)	X	17-19-19-38 (1T)	13-14-14-28 (1T)

So, how far will these modules go? We first had to determine what the maximum settings should be. Our CPU has to last through the entire X299 review series, so excessive voltage wasn’t even on the table. Yet DDR4-4800 is on the horizon, and following the trend of DDR2-800 CAS 4, DDR3-1600 CAS 8, and DDR4-3200 CAS 16, we picked CAS 24. Unfortunately, most boards don’t work well past CAS 19, so the compromise of 19-24-24-48 timings was chosen as the limit.

We tried for DDR4-4200, but the setting wouldn’t even boot. Perhaps that’s due to the high 21x multiplier? The next higher setting of DDR4-4266 uses a lower 16x multiplier thanks to the CPU’s available 4:3 memory controller BCLK to core BCLK ratio, yet it also wouldn’t boot. Without spending days trying to configure advanced settings, DDR4-4160 was still a reasonable limit. And not to be outdone as a value-priced overclocking option, the DDR4-2400 samples reached DDR4-3328.

Ripjaws DDR4-3800 tears though Sandra’s Bandwidth bench, annihilating the rated settings of the DDR4-2400. For value-seeking overclockers, custom tweaks bring the lower cost memory’s loss up to acceptable levels.

Lower is better regarding latency, and the Ripjaws rips through Sandra’s test there as well. On the other hand, custom timings again take the embarrassment out of the lower cost Super Talent memory’s results.

F1 2015 is famed for its bandwidth dependence, or its CPU dependence, or its timing dependence, depending on which weakness is the greatest. With a 4.40 GHz CPU clock, the game shows big advantages for performance memory settings compared to the stock DDR4-2400 results. Once again though, we still see that custom tuning can push the lower cost configuration up to nearly competitive performance.

Metro Last Light’s reaction to changes in memory settings is closer to the typical reaction we see in most games, and we’ve included it just to remind users that fast DRAM isn’t a panacea. Slow RAM has previously pulled the bench down, but we generally don’t touch anything below mainstream.

Blender is also a type of application that’s harmed by slow RAM, but reaches a plateau somewhere around mainstream level. The chart doesn’t show that the actual performance range was 78.6 to 79.4 seconds, but we saw it in the results text, which only showed us that the performance differences were electronically measurable but imperceptible.

7-Zip loves fast RAM. The Ripjaws DDR4-3800 started out strong and got stronger with a bit of tuning. Tuning helped the cheap set more because it started out so poorly at stock settings.

Ripjaws DDR4-3800 started out around 8% stronger than the baseline DDR4-2400 across the average of real-world benchmarks, but it also costs around 50% more. This would have looked much better had we compared it against the price of an entire PC, since it accelerates applications for the entire PC. Adding $2000 to the price of each kit would have made the black bars for both kits reach 100%, allowing one to honestly say that if the rest of the PC costs more than $2000, the Ripjaws DDR4-3800 could be the better deal at stock rated settings. Memory tuners, on the other hand, could point out that the blue bars don’t reach the same value parity unless the rest of the machine costs $4800.

Beyond the classic calculations of value are the concepts of risk and bestness. One doesn’t need to spend $4800 on their machine to want the “best” RAM, just as many users buy a top-model processor when the next one down would have sufficed. And there’s always the risk of not reaching performance goals for those who decide to start with a lower cost product and overclock. We give the product a nod for those who refuse to settle for less.

MORE: Best Memory

MORE: All Memory Content

A New Player in Motherboards: Our NZXT N7-Z37XT Review

Thomas Soderstrom — Tue, 09 Jan 2018 15:00:00 +0000

Specifications & Layout

It’s not every year that we see a new motherboard brand. Heck, it’s not even every 10 years. In the world of enthusiast-level PCs, certain things are relatively easy, while others present larger challenges. Slapping a label on an Asetek-made CLC? Relatively easy. Hiring a designer to work with an established production facility to make a variation of an existing case that looks slightly different than everything else? Not too difficult. Getting that facility to produce an entirely new case design that may involve new tooling? Slightly challenging.

But producing an entirely new motherboard from a clean sheet? Only the most experienced motherboard brands would do that. Until now?

Labeled as having been designed in California and manufactured in China, the N7-Z37XT is NZXT’s first foray into the motherboard market. We aren’t given details about how much control the ODM was given concerning its circuit design, although it’s likely that NZXT was at least responsible for specifying which connectors to use and where to put them.

Designed to support NZXT’s Grid+ (fan) and Hue+ (lighting) controls, the N7-Z37XT is available with either -W1 white or -B1 black shrouds.

Specifications

Conspicuously absent from the N7-Z37XT spec sheet are any USB 3.1 Gen2 controllers, in an enthusiast market where nearly every traditional brand has recently upgraded to the latest PCIe 3.0 x2 version. Some boards in its price class even have two of those, the second dedicated to a recently-introduced front-panel header. We’ve only tested one case with the newer Gen2 connector, but it will be interesting to see if NZXT introduces one that isn’t supported by this board.

Boards in the N7-Z37XT’s price class typically have dual gigabit Ethernet as well, and the absence of that extra feature leaves a noticeable gap in the I/O panel. On the other hand, a few absent features have allowed NZXT to avoid resource sharing, which is something its competitors aren’t even attempting to do. All of the N7-Z37XT’s features can be used simultaneously.

Intel’s Z370 PCH is still limited to 30 HSIO resources, and NZXT employs 29 of those as two NVMe M.2 slots, two additional PCIe x4 slots, a PCIe x1 slot, eight USB 3.0 ports, and four SATA ports. NZXT fills some of the I/O panel with USB 2.0, which doesn’t require HSIO resources. We also find enough analog audio ports to support 7.1 surround, Line In and Line Out simultaneously, along with an optical S/PDIF output, HDMI, DisplayPort, and a CLR_CMOS button.

While it’s possible to configure up to four NVMe drives using a combination of four-lane M.2 and four-lane PCIe x4 slots, the chipset’s link to the CPU is still limited to Intel’s four-lane DMI. And while there isn’t any HSIO sharing that would disable slots or devices, the CPU’s separate x16-lane PCIe controller still serves the two x16-length slots in either x16/x0 or x8/x8 modes, using automatic detection of a second card to make that transition. There are no magic tricks or latency-inducing hubs; just a design that’s well-planned to make full use of available resources.

Available in blue, red, purple, white, and black, heat sink shrouds are made of the same painted steel as the larger motherboard shroud. Removing the upper heat sink shroud also eases access to connectors along the upper edge.

Removing two plastic covers exposes the two M.2 slots and a Port 80 diagnostics code display. A third plastic piece of similar style covers nothing of import, and was likely added to match the style of the M.2 covers.The primary USB 3.0 connector and four SATA ports can also be seen from this angle.

The bottom edge has a front panel HD-Audio header, power and reset buttons, three USB 2.0 headers, a second USB 3.0 header, four fan headers, an AC-97-style front-panel button and LED group, a BIOS IC selector switch, and a button that, when deployed, copies the active firmware ROM over the inactive ROM.

The N7-Z37XT’s upper edge has five fan headers and an RGB LED connector. A second RGB LED connector is seen in the rectangular shroud hole near the top of the front edge.

While the black plastic I/O hood is screwed on, the painted steel motherboard shroud is secured with metal snaps. Underneath these are a traditional-looking upper-range board with matte-black mask.

The original N7-Z37XT installation kit includes four SATA cables, two magnetic/adhesive RGB lighting strips, two RGB adapter cables, two RGB extension cables, a single-link SLI bridge, an I/O shield, and a nondescript set of threaded hardware. NZXT is revising the contents of its next motherboard shipment to remove the RGB strips and cables, and has reduced its MSRP by $50 to match this hardware reduction. A few buyers will likely "luck out" by receiving the original kit at the new, lower price.

MORE: Best Motherboards

MORE: All Motherboard Content

Software & Firmware

The biggest news for NZXT’s N7-Z37XT is that it supports the firm’s CAM cooling management and RGB software. Readings are adequate, but the CPU temperature appears to be an average rather than a maximum, and it’s a few degrees cooler than the “Package Temp” we use from Intel XTU.

Night mode is an optional color scheme to make the menu emit less light from your monitor. There’s also a toggle for the two main readings menus to switch from gauge style to line-graph logging mode.

The CPU overclocking menu functions as expected, but one thing you might not have expected was to see the CPU core at 1.294V when fully loaded using default settings. That’s certain to hurt its efficiency score.

The fan menu also works well. Users should click the “Edit” icon to reach the settings menu, where standard profiles are available and custom slopes can be configured.

RGB controls address both outputs either in tandem or separately, and the latter created some unusual light shows during our tests.

Shown in “night mode,” a miniature display can be used to monitor changes with less intrusion on desktop space.

Firmware

N7-Z37XT firmware opens to a simplified interface by default, but only when using certain types of graphics output. The firmware doesn’t include screenshot saving capability, and outputs that worked with our antique capture device did not support the simplified interface. Advanced mode is where the real action happens, anyway.

Once inside Advanced Mode, the CPU & Ram menu provides links to overclocking submenus and clock readings.

Our CPU reached its expected 4848 MHz final clock by simply increasing the CPU ratio to 48x and the BCLK to 101.00 MHz. This allowed Turbo Boost settings to be ignored.

Primary and secondary timings can be manually configured, and the settings of the chosen mode remain when switching to “Custom Profile.” In our case, selecting XMP Profile 1 first, then switching to Custom Profile, allowed us to pick a different multiplier while maintaining XMP timings. Still, our highest DRAM data rate came at the XMP Profile’s 38.66 setting.

The board reports that its 1.34V DIMM setting produced up to 1.364V, but our meter only showed 1.354V. Most companies cheat in the other direction, as some competitors have been known to bolster their overclocking credentials by tricking builders into using more voltage than they intend.

The N7-Z37XT can store up to eight complete firmware settings as user profiles and export those profiles to a USB flash drive.

The N7-Z37XT doesn’t have PWM to voltage-control fan mode switching, and its firmware-based fan utility is limited to four modes (the "Disabled" setting is full-speed mode).

One thing that users of NVMe drives must do to make their drives bootable is disable RST support for Port 21. Manually. Because the firmware isn’t intelligent enough to automatically configure the only installed drive to its only functional mode.

MORE: Best Motherboards

MORE: All Motherboard Content

How We Test

Sound	Integrated HD Audio
Network	Integrated Gigabit Networking
Graphics Driver	GeForce 382.53

Cooler Master’s HAF-XB provides an optimal layout to blow the Celsius S24’s fans sufficiently over each motherboard’s voltage regulator.

Comparison Products

NZXT N7-Z37XT

EVGA Z370 FTW

ASRock Z370 Taichi

The N7-Z37XT provides overclock settings competitive with the similarly priced products used in today's comparison, apart from the lacking PCH voltage adjustment and the broader 50mV steps of CPU VCore. We don't overclock the PCH, and the larger VCore steps still got us to our 1.30V goal, so we should expect similar achievable frequencies from all models.

Benchmark Settings

Synthetic Benchmarks and Settings
PCMark 8	Version 2.7.613Home, Creative, Work, Storage, Applications (Adobe & Microsoft)
3DMark 13	Version 4.47.597.0Skydiver, Firestrike, Firestrike Extreme Default Presets
SiSoftware Sandra	Version 2016.03.22.21CPU Arithmetic, Multimedia, Cryptography, Memory Bandwidth
DiskSPD	4k Random Read, 4k Random Write128k Sequential Read, 128k Sequential Write
Cinebench R15	Build RC83328DEMOOpenGL Benchmark
CompuBench	Version 1.5.8Face Detection, Optical Flow, Ocean Surface, Ray Tracing
Application Tests and Settings
LAME MP3	Version 3.98.3Mixed 271MB WAV to mp3: Command: -b 160 --nores (160 Kb/s)
HandBrake CLI	Version: 0.9.9Sintel Open Movie Project: 4.19 GB 4k mkv to x265 mp4
Blender	Version 2.68aBMW 27 CPU Render Benchmark, BMW 27 GPU Render Benchmark
7-Zip	Version 16.02THG-Workload (7.6 GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=9"
Adobe After Effects CC	Release 2015.3.0, Version 13.8.0.144PCMark driven routine
Adobe Photoshop CC	Release 2015.5.0. 20160603.r.88 x64PCMark driven routine (light and heavy)
Adobe InDesign CC	Release 2015.4, Build 11.4.0.90 x64PCMark driven routine
Adobe Illustrator	Release 2015.3.0, Version 20.0.0 (64-bit)PCMark driven routine
Game Tests and Settings
Ashes of Singularity	Version 1.31.21360High Preset - 1920x1080, Mid Shadow Quality, 1x MSAACrazy Preset - 1920x1090, High Shadow Quality, 2x MSAA
F1 2015	2015 Season, Abu Dhabi Track, RainMedium Preset, No AFxUltra High Preset, 16x AF
Metro Last Light Redux	Version 3.00 x64High Quality, 1920x1080, High Tesselation, 16x AFVery High Quality, 1920x1080, Very High Tesselation, 16x AF
The Talos Principle	Version 267252Medium Preset, High Quality, High Tesselation, 4x AFUltra Preset, VeryHigh Quality, VeryHigh Tesselation, 16x AF

MORE: Best Motherboards

MORE: All Motherboard Content

Benchmark Results & Final Analysis

Synthetic Benchmarks

The N7-Z37XT shows slightly-off graphics scores in 3DMark, but its overall scores are comparable to other boards. That small deficit is repeated in Compubench, but nowhere else.

Sandra’s CPU and memory test show perfect results, so it’s time to move on to some real-world tests!

3D Games

The N7-Z37XT performed about average in Ashes, but came up a hair short in Metro and Talos. Speaking of short, the MSI board’s Talos-compatible Nahimic software suite put that otherwise class-leading board behind the pack.

Timed Applications

There’s nothing odd, either good or bad, to discuss about the N7-Z37XT’s performance in our mixture of timed workloads. Even its GPU-accelerated task performance is on-par.

Power, Heat, & Efficiency

Power use is the N7-Z37XT’s Achilles heel, as the board uses the CPU’s highest voltage setting even as Intel Turbo Boost defaults instruct it to clock down to 4.3 GHz (from a maximum of 4.7 GHz) under six-core loads. Lower clocks are supposed to have a lower voltage requirement.

Although Intel’s XTU utility shows a supposed 164W TDP, actual power consumption far exceeds the 190W or so that would be required of the system to maintain that TDP.

Of the four competing boards, the Z370 Taichi shows the power we’d expect our system to draw to maintain that high TDP, and the Z370 FTW shows the power we’d expect our system to draw if voltage were fully optimized for the 4.3 GHz six-core-loaded frequency.

The high power draw imparts high temperatures for both the CPU and voltage regulator, although the N7-Z37XT’s large voltage regulator sink does a good job of minimizing that second heat penalty.

The best news might be for future reviews, since these terrible power numbers will go into the averages that those other boards must compete against.

Overclocking

One might say that the firmware interface for N7-Z37XT overclocking is a bit under-developed, but others might just call it sparse or simplified. It certainly has all the controls our processor needed to reach its 4848 MHz goal under 12 threads of AVX-heavy Prime95, and that target is based on the thermal limits of the CPU’s core-to-heat-spreader interface. We couldn’t go higher without modifying the CPU package.

In fact, the only thing under-developed from a review standpoint is that the board didn’t appear to automatically compensate for higher BCLK. You’re free to make some of those adjustments manually, but we prefer to start our overclocks with the CPU multiplier anyway.

At a 3906 MHz data rate, the DRAM barely exceeded its XMP-3866 rating before it became unstable on this board. Performance is the point of overclocking, and we can clearly see a division between boards that optimize secondary and tertiary timings for performance, and those that don’t. While the N7-Z37XT is among that second group, it’s in good company with the Z370 Taichi and Z370 Godlike Gaming.

Final Analysis

As with the other percentage benchmarks, the N7-Z37XT is compared to all previously-reviewed boards in the performance-per-dollar chart. Perhaps we should have narrowed it down to include only today’s contenders? It’s $300 price would still have put it in fourth place, simply because three of the comparison boards were cheaper.

What we’re really looking at is the price of features, where the EVGA Z370 FTW is the cheapest of the five boards compared and also the one with the fewest added features. It doesn’t even have any RGB headers. It does, however, have a Gen2 USB 3.1 controller.

Moving up to something closer to the N7-Z37XT’s price, the Z370 Taichi has two Gen2 USB 3.1 controllers, plus dual Ethernet, plus a low-model Intel Wi-Fi controller, plus a couple of added SATA controllers, plus a front-panel header for Gen2 Type-C ports. On the other hand, it only has one RGB LED header. Oh, and it’s only $220.

The closest we can get to the N7-Z37XT’s price is with the $250 (MSRP) Z370 Aorus Gaming 7. It swaps out the PCIe 3.0-x1/PCIe 2.0-x2 interface of the older Gen2 USB 3.1 controller for a pair of the newer controllers, which use PCIe 3.0-x2. It has two RGBW LED headers plus two D-LED headers for strips that have their own controller IC.

Those competing boards have something else in common: Due to the 30 HSIO limit of Intel’s Z370 PCH, they’re all forced to disable some on-board device to enable some other on-board device or interface. At least the N7-Z37XT is an honest board.

The major value-added feature that’s supposed to set the N7-Z37XT above its competitors is integration of NZXT’s famed fan and lighting controller functionality. While those are indeed strong selling points, they’re also the reason why I’ve been recommending other boards that have similar integrated features as an alternative to external controllers. Asus’ Aura software, for example, is famed for its functionality, even supporting third-party devices such as RGB RAM lighting. The RGB Fusion software of the compared Z370 Aorus Gaming 7 isn’t far behind that of Asus Aura. Even the pared-down RGB LED interface of the Z370 Taichi is functional, though that particular model has but one RGB output. Only the EVGA board gets cut from that race, and only because it’s designed for builders who eschew RGB.

The N7-Z37XT’s steel shroud is the one feature that all the competing models lack, and whether that feature is worth $100 on its own is up to readers to decide. Competitive builders will certainly blow big cash to get a specific look, and it would be terribly difficult to replicate such a piece by hand in less than a day. What’s a day’s work worth to you? And then there’s the cost of design and development: The economies of scale require that buyers of low-volume products pay a larger portion of the development cost for those products. If a large-scale manufacturer later offers us a review sample with similar features, the math for that product will be completely different.

I can’t recommend for or against the N7-Z37XT from a value perspective: If you’re an interested buyer within this lower-volume market, “can’t recommend against” should be considered adequate praise.

MORE: Best Motherboards

MORE: All Motherboard Content

Gaming And Streaming: Which CPU Is Best For Both?

palcorn@outlook.com (Paul Alcorn) — Wed, 03 Jan 2018 14:00:00 +0000

Introduction

Game streaming is immensely popular right now. And with more than 2 million broadcasters serving approximately 100 million viewers a month, Twitch is unquestionably the leading platform for gamers looking to show off their skills.

Streaming is pretty easy to set up once you decide what you'll use to encode your content. There are several options, each with its own strengths and weaknesses. Hardware-accelerated encoders, powered by Nvidia's NVEnc, AMD's Video Coding Engine, and Intel's QuickSync, all promise to wrap your gaming experience up in a good-looking and easy-to-stream package without hammering your CPU. But it's generally agreed that these encoders sacrifice quality and flexibility compared to software-based alternatives.

Conversely, software encoding with the x264 library is easy alongside utilities like Open Broadcaster Software (OBS) and XSplit. It offers better stream quality than GPU-driven approaches, but at similar bit rates. That's important for services with defined bit rate limitations (like Twitch). Streaming can also chip away at bandwidth caps. Just one hour of streaming at 10 Mb/s equals 4.5GB of data, so you want the most efficient encoder possible.

A host processing-based approach offers more flexible configuration options. Better quality comes at a price, though. Software encoding is a CPU-intensive process, which obviously tugs the other way against our usual aim of running games at the highest frame rates possible.

Hitching, stuttering, lag, and generally bad performance curse gamers who try doing too much with limited software encoding resources. Thus, it's often recommended to use a secondary system for high-quality streaming. In fact, that's the best way to achieve an unencumbered gaming experience while also providing your audience a crisp video stream. That requires attaching your gaming rig to another computer via capture card or local area network to offload the encoding workload, though. Many casual gamers simply don't have the money for such a configuration.

Software encoding can have a huge impact on gaming performance. Until recently, if you wanted to stream and game using the highest-quality solutions on one PC, you almost assuredly had to buy a pricey CPU with lots of threads to throw at the problem. But now those high-end desktop processors are more accessible than ever. AMD's Ryzen 7 chips lowered the bar for getting lots of cores into affordable platforms, and most games can't fully utilize them. That leaves lots of horsepower leftover for streaming at a reasonable price.

Intel's Coffee Lake processors recently debuted with more cores as well, giving those Ryzen CPUs a run for their money in multi-tasked, multi-threaded workloads. Good thing, too. As we'll see, the previous-gen Kaby Lake architecture simply cannot compete against like-priced AMD alternatives. Today's story compares a number of different host processors in combined gaming/streaming workloads. Along the way, we developed test methodology you'll see used in our future CPU reviews.

MORE: Best CPUs

MORE: All CPUs Content

Test Setup

How We Tested

Repeatability is one of the most important components of any useful benchmark methodology. All tests have some degree of uncertainty, but we're looking for a minimal and consistent amount of variability. Results plagued by wild swings in performance from one run to the next aren't usable as accurate benchmarks.

As an example, we've yet to develop any reliable multi-tasking benchmarks. In response to reader requests, we have worked diligently to create a series of tests that measure gaming performance with background applications like Web browsers, email clients, media players, Discord, and Skype open. Windows' prioritization appears to be based on fickle and unexplained factors. The operating system suspends various background processes unpredictably during one scripted sequence, then leaves them fully active during the next (even when the test environment hasn't changed). This unpredictability becomes more, well, unpredictable, as the number of open applications increases. Switching Windows into Game Mode only complicates matters further. So far, we have no solution. Our multi-tasking experiments yield deltas from 5 to 15 FPS between successive runs, which means they land nowhere near our expectations for a reliable benchmark.

Luckily, game streaming is much easier to control. Encoding is a CPU-intensive task that chews up plenty of cycles, so Windows doesn't suspend or otherwise interfere with it. This allows us to create repeatable benchmarks without extreme outliers.

What We're Measuring

Evaluating game streaming performance works across two axes: game quality and stream quality. Of course, we'll measure average, minimum, and 99th percentile frame rates with and without streaming in the background. We'll also include our usual frame time and variance results, which become more important once we start streaming.

We also need to account for stream quality. That means recording the percentage of frames encoded. Each processor pushes different frame rates, so each run correspondingly generates a different number of frames. As such, we measure the percentage of frames successfully encoded as "% of Frames Delivered." In the test below, a Threadripper 1950X CPU encoded 98.9% of the frames generated by our gaming session, meaning it skipped 1.1% of the frames due to encoding lag.

We're streaming at 60 FPS, so we also measure stream quality by listing the percentage of frames encoded within the desirable 16.667ms (60 FPS) threshold. We also include the percentage of frames that land above and below the 60 FPS threshold, which helps quantify the hitching and stuttering a viewer would see on the stream. Subjective visual measurements are still important, so we'll call out tests that generate a bad-looking stream.

Open Broadcaster System

There are several software encoding applications, but we chose Open Broadcasting System (OBS) due to its flexible tuning options, detailed output logs, and broad compatibility with streaming services. We're using the x264 software encoder, along with YouTube Gaming for our streaming service. Any run that reports frames dropped due to networking interference is discarded.

Our ultimate goal is to develop a test that measures CPU performance, so we select parameters that remove the most obvious bottlenecks. Gaming at 1920x1080 with an EVGA GeForce GTX 1080 FE side-steps a GPU limitation (as much as possible). Encoding overhead isn't as high with lesser video cards that generate fewer frames per second. We also test with a 10 Mb/s upload rate, though you can stream at 6 Mb/s or less. Our Internet connection would accommodate up to 35 Mb/s uploads. To vary game selection, we chose Grand Theft Auto V, Middle-earth: Shadow of War, and Battlefield 1 for our tests.

There are several other scenarios we could have added to increase the complexity of our testing, such as a simultaneous video stream from a webcam, recording the game to the host system, or streaming to multiple services at once. We went with just one service to reduce the number of variables...at least for now.

Finding the best streaming options requires some tuning for every game and hardware configuration. There is a delicate balance between game performance on the host system and stream quality for the remote viewer, so fine-tuning is needed to yield the best mix. We picked somewhat general settings that offered a good range of performance by our subjective measure. We also stuck with options that'd establish a level playing field for a wide range of test systems. Just be aware that there are plenty of knobs to turn, some of which could offer better performance than the ones we use (lowering the stream to 30 FPS, for instance, cuts encoding overhead significantly)

Tuning the encoding presets is one of the most direct ways to adjust streaming performance and quality for your system's capabilities. Slower encoding increases compression efficiency, which provides better output quality and reduces compression artifacts. OBS has 10 presets ranging from "ultrafast" (the lowest-quality setting with the least computational overhead) to "placebo" (offering the best streaming quality and consuming the most host processing resources). The placebo setting is aptly named; there is certainly a rapidly diminishing rate of return on stream quality after passing the "slower" preset (two ticks before placebo). More strenuous settings can quickly cripple even powerful processors, particularly if you are streaming from a single host system. Placebo with care.

We split our test groups into three different classes. After evaluating a few Core i3- and Ryzen 3-class processors and determining that they can't stream effectively at our settings, we chose Ryzen 5 and Core i5 models for our entry-level systems. We used the "veryfast" encoding setting for this class of CPU. Naturally, higher-end processors, such as our Ryzen 7/Core i7 and Threadripper/Core i9 chips, offer more performance, so we use the "faster" and "fast" settings, respectively, for brawnier CPUs.

Because we're testing with different encoding presets, you cannot compare test results for the different classes directly.

Test System & Configuration

Hardware

Intel LGA 1151 (Z370)Intel Core i5-8600K, Core i7-8700KMSI Z370 Gaming Pro Carbon AC2x 8GB G.Skill RipJaws V DDR4-3200 @ 2666 and 3200 MT/sAMD Socket AM4AMD Ryzen 5 1600X, Ryzen 7 1800XMSI Z370 Xpower Gaming Titanium2x 8GB G.Skill RipJaws V DDR4-3200 @ 2667 and 3200 MT/sIntel LGA 1151 (Z270)Intel Core i5-7600K, Core i7-7700K MSI Z270 Gaming M72x 8GB G.Skill RipJaws V DDR4-3200 @ 2666 and 3200 MT/sAMD Socket SP3 (TR4)AMD Ryzen Threadripper 1950XAsus X399 ROG Zenith Extreme4x 8GB G.Skill Ripjaws V DDR4-3200 @ 2666 and 3200 MT/sIntel LGA 2066Intel Core i9-7900X, Core i9-7980XEMSI X299 Gaming Pro Carbon AC4x 8GB G.Skill Ripjaws V DDR4-3200 @ 2666 and 3200 MT/sAll EVGA GeForce GTX 1080 FE 1TB Samsung PM863 SilverStone ST1500-TI, 1500W Windows 10 Creators Update Version 1703Corsair H115i

MORE: Best CPUs

MORE: All CPUs Content

Intel Core i5 & AMD Ryzen 5

We generated our stream using the "veryfast" encoder preset for this class of processors.

AMD Ryzen 5 1600X

Intel Core i5-8600K

Intel Core i5-7600K

Battlefield 1

First, we run a set of baseline tests to gauge performance without an active YouTube stream. Intel's Core i5-8600K fares best, but the Core i5-7600K and Ryzen 5 1600X aren't far behind.

The story changes when we add streaming to the mix. All three CPUs lose varying amounts of performance, with Ryzen 5 1600X ending up on top (though not by much).

We have two entries in our charts for the Kaby Lake-based Core i5-7600K: normal and high-priority. First, let's talk about the normal results. Intel's 4C/4T architecture hamstrings the Core i5-7600K, and it loses nearly half of its average frame rate. Unsurprisingly, this has an impact on in-game frame time variance and unevenness measurements. Those numbers don't tell the entire story, though. Battlefield 1 is basically unplayable due to hitching and stuttering during our stream.

The stream quality chart shows that Core i5-7600K only delivers 23.6% of its frames during the test, which means it drops an astounding 76.4% of the frames. This also results in a completely unwatchable stream. Flip over to the CPU utilization charts; they tell a profound story. In short, Core i5-7600K doesn't have any headroom available to handle the parallelized encoding workload throughout most of our benchmark.

The OBS software has several different settings to optimize performance, and adjusting process priority is a common tactic to improve streaming on Kaby Lake-based processors. So, we selected the high-priority setting (marked HP Stream in the charts). Effectively, this allows the encoder to steal cycles from the game engine, resulting in lower frame rates. But the encoder successfully processes 100% of the frames. Our Core i5-7600K fell to 51.6 FPS, and its 99th percentile measurements nosedived. On the flip side, less performance means fewer frames to encode per second, which in turn boosts encoding efficiency. Surprisingly, 91.05% of the frames landed within the desirable 16.667ms range.

We could always dial-back the encoder preset to accommodate Intel's lackluster -7600K, but similarly-priced processors in the test pool handle these quality settings adeptly. While streaming, Core i5-8600K provides nearly the same in-game average frame rate as AMD's Ryzen 5 1600X and it encodes 100% of the frames. The -8600K's streaming quality isn't as impressive as Ryzen's: 89.44% of frames land within the desirable range and ~10% of frames fall either above or below the threshold. In our opinion, the stream looked fine, but wasn't as smooth as the Ryzen 5 1600X's output.

A glance at the 99th percentile FPS chart tells us that Ryzen 5 1600X delivers the smoothest gaming experience during our streaming session. It also has plenty of reserves left in the tank, even during the streaming workload. In other words, you could probably increase the encoding preset to a higher level than the competing processors.

AMD's Ryzen CPUs don't overclock as well as Core processors, though. We tuned each model to see if the Intel's available headroom turned the tables. The x264 encoder uses AVX instructions extensively, so it serves as a nice stress test for overclocking stability. No doubt, Coffee Lake's AVX offset could come in handy, though it might also result in lower frequencies during streaming if you activate the feature (or leave it on Auto in the UEFI).

Overclocking nudges the Core i5-7600K ahead of Ryzen 5 1600X during our baseline gaming test, but it does little to rectify the issues encountered during our streaming workload. Four threads just can't hack it. Streaming suffers even after adjusting the priority status. The -7600K does serve up a higher percentage of frames with the overclocked settings and normal prioritization, but it still costs the chip game performance. Priority adjustments have little bearing on gaming smoothness: the -7600K's 99th percentile measurements are terrible with both settings.

A 4.9 GHz overclock does boost the Coffee Lake-based Core i5-8600K's 99th percentile measurements. However, it still trails Ryzen 5 1600X. The tuned -8600K is certainly competitive thanks to a compelling gaming experience, but it can't match the Ryzen 5 1600X's streaming performance. It's apparent the encoder appreciates lots of threads, so Ryzen's 12 are a big advantage.

The frame time chart summarizes in-game performance well. Kaby Lake-based processors scribble their way across with extreme variance. The Core i5-8600K is also easy to spot during our streaming workload. Meanwhile, Ryzen 5 1600X provides a much more consistent in-game experience.

Grand Theft Auto V

Grand Theft Auto V scales well with additional host processing resources and tends to favor Intel architectures. It isn't surprising to see the Core CPUs lead during our baseline tests. But we weren't expecting the Core i5-8600K to maintain its advantage over the competition while streaming.

The Core i5-8600K delivers 100% of its frames during the test, though only 89.71% fall into the perfect 60 FPS bucket. We see a nearly even split of 5% above and below that mark. Visually, the stream looks fine. The -8600K takes more of a hit in the 99th percentile measurements, though it still leads the rest of our contenders.

AMD's Ryzen 5 1600X starts out with the lowest average frame rate during our baseline and falls to 60.7 FPS during the stream. It doesn't suffer too badly in 99th percentile measurements. Meanwhile, the chip provides stellar streaming performance, dropping no frames and almost landing at a smooth 60 FPS.

The Core i5-7600K provides a decent gaming experience during normal streaming, churning out 79.2 FPS. That's because it really isn't encoding frames, though—it drops 84.9% of them. The stream is unwatchable by any measurement. Again, we can boost the -7600K's encoding performance to 100% by specifying high priority, but then the gaming experience is terrible. In fact, large portions of the scene simply do not render correctly.

Overclocking the Core i5-7600K doesn't help; it's still entirely saturated during the streaming test. But tuning does benefit Intel's Core i5-8600K, which leads the pack in both average and 99th percentile metrics. Its streaming quality is decent as well.

The Ryzen 5 1600X's performance improves after overclocking, too. It doesn't provide the absolute best frame rate while streaming, but it is perfectly playable. Also, Ryzen 5 1600X still offers the smoothest stream, if only just barely.

Middle-earth: Shadow of War

Intel's Core i5-7600K continues to disappoint. It serves up an average of 51.4 FPS while streaming (using the high priority setting) and encoding 100% of the frames. Unfortunately, game smoothness goes out the window again. The 15.9 FPS 99th percentile measurement tells us everything we need to know.

The Ryzen 5 1600X offers decent gaming performance while streaming, but that Coffee Lake-based -8600K continues leading.

AMD's Ryzen 5 1600X offers the best of both worlds, though Core i5-8600K is also very capable. Perhaps tuning the encode could get more performance from the Intel chip. Then again, you could say the same thing about Ryzen 5, which offers the best mix of streaming and frame rates in its class. It's obvious from these results that a Kaby Lake-based Core i5 owner needs to dial back streaming settings dramatically or turn to faster hardware.

MORE: Best CPUs

MORE: All CPUs Content

AMD Ryzen 7 & Intel Core i7

Paying more for a processor should also get you more performance, right? With that in mind, we're stepping up to the "faster" encoder preset for this CPU class.

AMD Ryzen 7 1800X

Intel Core i7-8700K

Intel Core i7-7700K (Kaby Lake)

Battlefield 1

The Core i5-7600K disappointed us in the previous round of tests, and Core i7-7700K picks up where the i5 left off. This chip just isn't well-equipped for streaming, despite the additional four threads enabled by Hyper-Threading. The -7700K dropped 94.3% of our frames using the normal settings, which comes as a side effect of excessive CPU utilization. We triggered the high priority setting to try rectifying this situation, but were presented the same bipolar behavior as Intel's Core i5-7600K. There just isn't a straightforward way around saturated CPU cores. You can either stream well or game well, but you can't do both with Kaby Lake-based processors and OBS streaming at 1920x1080/60FPS.

The Ryzen 7 1800X bears down with eight cores and 16 threads to provide solid streaming and gaming performance.

Core i7-8700K, which brings Coffee Lake up to a Hyper-Threaded six-core design, also performs well in this test. Overall, it offers the best gaming performance while streaming, 99th percentiles included. It even ekes past Ryzen 7 1800X with 100% of frames encoded to the stream. AMD's Ryzen 7 delivers 99.9% of the frames, and dropping 0.1% doesn't concern us. The stream is still smooth.

Overclocked, Ryzen 7 1800X delivers an ever-so-slightly better stream. But the Core i7-8700K leads in overall gaming performance. Of course, the 1800X's 117.8 FPS average, while streaming, is plenty impressive. Both processors prove up to this task.

Our encoding workload pegs the -7700K's cores at 100%, while the high priority settings help bring down CPU utilization. Simply, that's because the CPU has fewer frames to encode as a result of lower gaming performance.

Grand Theft Auto V

The Core i7-8700K takes a healthy lead during the baseline and streaming tests. It even provides more in-game performance while streaming than an unencumbered Ryzen 7 1800X.

Ryzen 7 1800X again achieves lower gaming performance, but a slightly superior stream than Intel's -8700K. The differences are pretty much imperceptible, though. Kaby Lake, on the other hand, simply can't compete.

Turning up the clocks benefits Coffee Lake greatly. Core i7-8700K offers the best overall performance, while Ryzen 7 1800X is competitive.

Middle-earth: Shadow of War

Coffee Lake continues to impress with strong in-game and streaming performance. The Ryzen 7 1800X does encode a higher percentage of frames than -8700K within the desirable 16.667ms range, but it also generates fewer frames. AMD's Ryzen does have more headroom to spare during our streaming workload though, so it'd likely accommodate more taxing encode presets.

We find that the stream frame time variance isn't very noticeable unless the 16.667ms percentage drops below a 90% threshold, such as what we see from Intel's Core i7-7700K. The -7700K does serve up better frame rates while it's streaming, even besting AMD's Ryzen 7 1800X. But that goes to show why we need to keep an eye on performance from every angle. Core i7-7700K drops 40% of its frames during this test! That makes any gaming performance advantage meaningless.

Ryzen 7 1800X musters a solid 75.4 FPS during the streaming test and provides a crisp stream to YouTube, as well.

Overclocking amplifies the trends we've been talking about. The same inherent strengths and weaknesses are still apparent. Ryzen 7 1800X and Core i7-8700K are both solid choices for streaming.

The Core i7-7700K demonstrates lower CPU utilization than usual during this test. Surprisingly, that doesn't equate to a usable combination of in-game and streaming performance, though, suggesting other architectural influences may be affecting our numbers.

MORE: Best CPUs

MORE: All CPUs Content

AMD Threadripper & Intel Skylake-X i9

You expect the best when you drop $1000+ on a CPU for gaming and streaming simultaneously. So, we shifted to the "fast" setting for this round of tests.

AMD Ryzen Threadripper 1950X

Intel Core i9-7900X

Intel Core i9-7980XE

Battlefield 1

AMD didn’t design Threadripper for "just" gaming, particularly at lower resolutions. But an intense streaming workload might expose more of the architecture's benefits.

Threadripper does have several configurable modes to tailor to its response to various tasks, as we outlined in our AMD Ryzen Threadripper 1950X Game Mode, Benchmarked article. Game Mode and Creator Mode both impact various titles differently, and we expect those same trends to carry over to our streaming benchmark. As such, we tested the Threadripper 1950X both ways.

The Threadripper 1950X pulls into range of Intel's Skylake-X models during our baseline tests. Creator Mode (CM) exposes all 32 of the 1950X's available threads, yielding more potential horsepower than Game Mode (GM). The 1950X's Creator Mode also makes it possible to encode 100% of the test run's frames, while Game Mode drops 1.6% of them. That isn't a huge sacrifice, but Game Mode also causes Threadripper to average 26 FPS fewer. That means it isn't encoding as many frames, either. We would have expected higher frame rates from a lighter encoding workload, but disabling half of the 1950X's threads in pursuit of higher game performance doesn't always work well when you're streaming, too.

The 18C/36T Core i9-7980XE doesn't perform as well as the 10C/20T Core i9-7900X during the streaming tests. Rather, the -7980XE stumbles in the smoothness department as its 99th percentile frame rates fall below the Threadripper 1950X, despite a lead in average frame rates.

Aside from Threadripper 1950X in Game Mode, all of these processors encode 100% of the frames.

The Core i9-7980XE's game performance is higher than the Threadripper models during our streaming test after bumping up clock rates. It also delivers a much better 99th percentile measurement.

AMD's Threadripper configurations fare better after tuning, too. Both encode 99.9% of the frames we send their way, which is adequate for a quality stream.

Grand Theft Auto V

Grand Theft Auto V's in-game performance goes Intel's way during the baseline benchmarks. Moreover, Core i9-7980XE redeems itself when it comes time to stream. That chip does command quite a premium though, so its advantage doesn't necessarily represent the best value.

Again, Threadripper's Game Mode just doesn't appear to be ideal for streaming. Game Mode provides better baseline frame rates in this title, but it falls to the bottom of our chart once we start encoding video.

Middle-earth: Shadow of War

Once again, Intel's processors offer the best in-game and streaming performance. This isn't entirely surprising; Threadripper may even be overkill for this type of enthusiast workload. We'd expect the architecture to handle workstation-class production workflows more adeptly.

Core i9-7900X does encounter a hiccup as its 99th percentile scores fall below Threadripper 1950X's during the streaming workload. The 1950X in Creator Mode also experiences some variance, with 20.38% of its frames falling below the 16.667ms threshold.

Even after a bit of tuning, Core i9-7900X provides a higher average frame rate than the 1950X in Creator Mode during our streaming benchmark, but can't match AMD's 99th percentile performance.

The Threadripper 1950X's difficulty in Creator Mode while streaming Middle-earth is even more pronounced. Some aspect of OBS doesn't agree with Creator Mode and this one game. We ran the tests several times to ensure it was a repeatable phenomenon.

MORE: Best CPUs

MORE: All CPUs Content

Final Analysis

AMD democratized access to high core counts with an attractively-priced Ryzen portfolio. In a way, we can thank the company for Intel's newfound interest in competing on a $/core basis. Just look at the difference in our results from Kaby Lake to Coffee Lake. As a result, high-quality software encoding on a gaming PC is becoming more realistic to mainstream gamers.

Our testing is indicative of general performance trends. Given enough time and energy, you could almost certainly improve upon our results. Part of that is by design: we're using these settings to compare large groups of processors against one another on a level playing field, as opposed to wringing the most performance out of any one processor. While we can't use our benchmarks to make definitive statements about the possibilities with each chip, we can draw some fair conclusions about how certain architectures behave.

Encoding is a parallelizable workload. If software encoding is your primary goal, you'll definitely want to seek out CPUs with lots of cores and simultaneous multi-threading capabilities. Intel's quad-core Kaby Lake models illustrate how chips that once offered class-leading gaming performance can fall apart during streaming. You can boost their performance by using less intensive quality presets, lowering the streaming frame rate, or sacrificing some quality with GPU acceleration. However, competing processors offer much more performance than Kaby Lake at the same presets and roughly the same price.

Many streamers place video quality over maximizing the frame rate of whatever game they're playing, so your own priorities will largely dictate how you tune your system. In fact, turning on v-sync may be a good way to balance streaming and gaming performance.If you seek the highest in-game performance while you stream, Intel's Coffee Lake-based Core i7-8700K is a good fit. The Ryzen 7 1800X is also competitive and tends to offer better streaming performance. Using our settings, the 1800X also had more CPU headroom leftover for more taxing encode settings, if desired. Granted, some of that extra horsepower is due to the 1800X's lower gaming performance, which means there are fewer frames to encode.

Two extra cores on the Coffee Lake-based Core i5 certainly help its standing, but the lack of Hyper-Threading has a definite impact on streaming performance. In the end, a six-core Core i5-8600K is forced to battle the 12-thread Ryzen 5 1600X, which offers a more balanced profile. Overclocking does help Intel somewhat. It can't overcome the advantage AMD gets from a more thread-heavy architecture, but it shrinks the gap somewhat in streaming workloads.

If you're really serious about streaming and gaming at the same time, the highest-end desktop CPUs are an option. Just expect to pay dearly for them. Most enthusiasts are better served by mainstream processors. Intel's Core i9 models generally provide better performance than the Threadripper 1950X, but they cost more, too. The 1950X is a solid value choice that also offers a diverse range of capabilities.

There are plenty of other solid options for gaming/streaming, and this introductory round of tests only focused on high-end models from each family. We'll expand our testing to locked SKUs as we work through coming CPU reviews.

MORE: Best CPUs

MORE: All CPUs Content

ASRock X299E-ITX/ac Skylake-X Motherboard Review

Thomas Soderstrom — Wed, 29 Nov 2017 14:00:00 +0000

Features & Specifications

With its ability to cram up to 18 physical cores into the tiny Mini-ITX form factor, the X299E-ITX/ac is ASRock’s second take on putting big processors into smaller boxes. So unusual was its predecessor that we gave it our best in class award just for being the only board in its class, even though it only provided two DIMMs to a CPU series designed for four channels. The new board addresses that weakness:

That’s not to say that the X299E-ITX/ac could be the perfect LGA 2066 solution for everyone who only needs a single PCIe x16 slot, because the board does not support Kaby Lake-X (Core i7-7740X or Core i5-7640X) processors. Arguing that those buyers should instead consider their Coffee Lake options would take us off topic: Let’s see what the X299E-ITX/ac can do for Skylake-X!

Specifications

By supporting only Skylake-X processors, ASRock assures that the X299E-ITX/ac would always have access to at least 28 CPU-based lanes, and ASRock wants to make sure everyone knows that its single PCIe x16 slot can alternatively feed two x16 graphics cards at x8-mode via a two-slot riser. Eight of the remaining lanes go to two M.2 slots, and a quick check of the connections under Device Manager shows the other four CPU lanes going . . . nowhere. The third M.2 slot is fed by the PCH.

The X299E-ITX/ac has only four USB 3.0 ports and zero USB 2.0 ports. I usually like to keep around the latter just to connect a keyboard and mouse without consuming any of the PCH’s precious HSIO lanes. It’s likewise missing another port that doesn’t need HSIO: the PS/2 keyboard/mouse port that so many stubborn gamers use for a legacy keyboard or mouse. Not that HSIO would be a problem for this board, since two of its M.2 slots are fed directly by the CPU. So few HSIO devices are fed by the chipset on this tiny board that we would have accepted two more USB 3.0 ports without concern for resources, but instead we find empty space above the USB 3.1 Gen2 (10Gb/s) ports that we like to reserve for high-speed external storage.

ASRock placed SATA, along with USB 3.0 and 2.0 front panel headers, on a removable riser card at the front of the board to conserve space. The CPU’s tiny voltage regulator heatsink is crammed beneath that riser and has a tiny air gap to assure proper ventilation. Screws connect the card to two tabs of the sink to assure rigidity.

A second mini riser is designed for easier removal, thereby easing access to its M.2 slot. It's identical to a SO-DIMM slot, and anyone who claims to have “accidentally” put a memory module in this odd location should probably be rebuffed. The riser also includes both Intel Gigabit Ethernet controllers and a USB 3.0 hub that forces all four I/O panel USB 3.0 ports to share the bandwidth of a single connection . . . despite the chipset’s under-utilized HSIO. An ASRock representative explained that there wasn’t enough space to run traces from the four unused CPU lanes to the riser card’s M.2 interface, but here we see four USB ports running from the riser card to the I/O panel. Odd. Perhaps the alternative would have required another PCB layer?

Two M.2 slots under the X299E-ITX/ac are limited to 80mm drives, and unlike the riser card slot, there are no 30, 42, or 60mm alternatives. ASRock needs no excuse for this limitation, as additional mounting points would have gotten in the way of DIMM traces. We also see the USB 3.1 Gen2 controller, along with the Super I/O controller and several voltage regulation components.

The X299E-ITX/ac provides an advantageous layout despite a few sacrificed connections. For example, the Front Panel Audio connector is found next to its I/O panel jacks, thereby preventing its connector from being smashed flat by a graphics card. The connector located where we expected to find HD-Audio instead appears to be for a serial port breakout plate and, being undocumented, is easily dismissed. A TPM connector is found between the rear riser card I/O panel’s USB 3.1 Gen2 ports. The upper rear corner features rear-fan and VROC jacks, while the upper front corner features RGB and front-panel LED/switch connections. Two more fan connectors are at the center of the front edge, along with legacy LED and beep-code speaker connections. A switch above the PCIe x16 slot selects whether the I/O panel’s button works as Power On/Off or CLR_CMOS.

The X299E-ITX/ac includes printed documentation, four SATA cables (two with a single right-angle end), a Wi-Fi antenna with self-adhesive Velcro attachment strip, a case badge, driver disc, I/O shield, and SSD mounting screws.

MORE: Best Motherboards

MORE: All Motherboard Content

Software & Firmware

Although its driver disc provides needed drivers to get the system working and online, ASRock’s software comes by way of its App Shop application. Here we find download links to its RGB LED control applet, a custom-interfaced version of the cFos network utility called XFast LAN, the self-explained Restart to UEFI applet, Sapphire’s graphics card overclocking utility, ASRock APP Charger USB device charging mode applet, A-Tuning overclocking suite, a bunch of freeware, and a Norton Security trial. The application’s BIOS & Drivers download tab polls ASRock servers for updates, and its Settings tab adds a tickbox to disable auto-load at Windows boot.

ASRock RGB LED is nothing more than a Windows app that accesses its firmware settings, its GUI a reflection of firmware. Though the board lacks on-board lighting, a test of its RGB output shows normal function (including color-cycling modes).

Likely due to thermal concerns, the X299E-ITX/ac does not have any factory-configured overclocking profiles saved in its firmware. Since A-Tuning pulls these from firmware, its EZ OC section is blank. Furthermore, since Auto Tuning uses EZ OC profiles as a baseline, it's likewise not enabled.

A-Tuning’s OC Tweaker menu still works, providing Windows access to a full range of UEFI settings. We confirmed functionality of CPU multiplier, BCLK, and voltage settings.

Other A-Tuning menus provide system monitoring and fan control. We couldn’t find any reading for the motherboard voltage regulator, and the reported “M/B Temperature” was up to 60°C below that of the voltage regulator.

Firmware

The X299E-ITX/ac opens to its Advance Mode GUI, where we were able to easily set a stable CPU frequency of 4.40 GHz and a DDR4-3800 data rate. DDR4-3800 SO-DIMMs? Jump to the next page to see them.

We had to manually increase power limits to accomplish any overclock under heavy CPU loads. The board would even force our CPU below its minimum Intel Turbo Boost ratio when loaded with Prime95 small-FFTs at default settings.

The X299E-ITX/ac’s UEFI supports XMP, along with a full range of data rates, primary through tertiary timings, and memory boot settings. It even has an extra menu to enable the viewing, selection, and application of the memory’s SPD and XMP programming.

The Voltage Configuration menu allows changes to motherboard settings, while the FIVR Configuration menu adjusts voltages inside the CPU. We were able to sustain our desired 1.15V CPU core at full AVX load using the board’s Level 2 Load-Line Calibration setting. Conversely, the DIMM setting of 1.340V produced a measured 1.354V at the slots, and the lower setting was required to keep within the 1.355V limit we chose to keep overclocking comparisons fair.

The UEFI Setup Style setting is found within the Advanced menu of Advance Mode, and users who don’t want to see all these settings can choose to enter firmware in EZ mode next time. The F6 keyboard function also toggles between these GUI styles.

The Tool menu includes RGB LED header control, an outgoing email client for tech support, a utility for copying RAID drivers from the disc to a USB flash drive, a utility to enter UEFI Flash mode, a utility to poll servers for new UEFI versions and download them to a USB flash drive, and a network configuration utility.

ASRock motherboards provide a UEFI utility that determines a fan’s operational range and alters its automatic slope accordingly, and two of its fan headers are even rated for 1.5A (rather than 1A) output. Those two headers also provide a voltage regulation option for speed control, and users are welcome to set their own fan slopes by clicking the FAN-Tastic Tuning tab to bring up an additional sub-menu.

Users who choose EZ Mode GUI can still set boot order, enable XMP, and access tools menus for Instant Flash firmware updating, Internet Flash firmware downloads, and Fan-Tastic Tuning.

MORE: Best Motherboards

MORE: All Motherboard Content

How We Test

Test Hardware

Though the firm already sold dual-channel kits up to DDR4-3200, G.Skill produced a new quad-channel SO-DIMM kit specifically to address the enhanced capabilities of ASRock’s X299E-ITX/ac. Rated at DDR4-3800, its F4-3800C18Q-32GRS will compete directly against the company's DDR4-3866 long DIMMs that have long been used to evaluate ATX boards.

Sound	Integrated HD Audio
Network	Integrated Gigabit Networking
Graphics Driver	GeForce 382.53

The tremendous heat produced by the Core i9-7900X required us to use nothing less than our award-winning Fractal Design Celsius S24 liquid cooling system sample. Getting its fans to blow sufficiently over each motherboard’s voltage regulator required additional design consideration: Of our leftover case review samples, Cooler Master’s HAF-XB had the best layout.

Concerned about our attempts to overclock using its tiny voltage regulator heat sink, ASRock sent along a BP-WBMASRX299EI monoblock from Bitspower. Using it would invalidate our thermal comparisons, so we reserved this RGB-lit part for an additional bonus test that will be discussed in our overclocking comparison. Because our Fractal Design Celsius S24’s pump is integrated into its CPU block, ASRock added the BP-LXAIO240-RGB Leviathan Xtreme All In One 240.

The Leviathan Xtreme 240 includes a two-way RGB splitter, but the monoblock did not, thus our choices were to light the monoblock from the motherboard, or the monoblock and one fan, or both fans but not the monoblock. The Leviathan Xtreme 240 also includes power adapters and an RGB controller of its own, but using it along with the motherboard's controller would have caused the LED groups to operate out of sync with each other. Rather than attempt to set a matching static color between the water block and fans using two different controllers, I chose to leave the fans unlit.

Comparison Products

ASRock X299E-ITX/ac

ASRock X299 Taichi

Gigabyte X299 Aorus Gaming 3

The X299E-ITX/ac is compared to the two most similarly-configured boards from the controller standpoint, the X299 Gaming Pro Carbon AC from MSI and ASRock's X299 Taichi. All models provide a range of overclock settings that exceeds the capabilities of our hardware, so this should be a solid test!

Benchmark Settings

Synthetic Benchmarks and Settings
PCMark 8	Version 2.7.613Home, Creative, Work, Storage, Applications (Adobe & Microsoft)
3DMark 13	Version 4.47.597.0Skydiver, Firestrike, Firestrike Extreme Default Presets
SiSoftware Sandra	Version 2016.03.22.21CPU Arithmetic, Multimedia, Cryptography, Memory Bandwidth
DiskSPD	4k Random Read, 4k Random Write128k Sequential Read, 128k Sequential Write
Cinebench R15	Build RC83328DEMOOpenGL Benchmark
CompuBench	Version 1.5.8Face Detection, Optical Flow, Ocean Surface, Ray Tracing
Application Tests and Settings
LAME MP3	Version 3.98.3Mixed 271MB WAV to mp3: Command: -b 160 --nores (160 Kb/s)
HandBrake CLI	Version: 0.9.9Sintel Open Movie Project: 4.19 GB 4k mkv to x265 mp4
Blender	Version 2.68aBMW 27 CPU Render Benchmark, BMW 27 GPU Render Benchmark
7-Zip	Version 16.02THG-Workload (7.6 GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=9"
Adobe After Effects CC	Release 2015.3.0, Version 13.8.0.144PCMark driven routine
Adobe Photoshop CC	Release 2015.5.0. 20160603.r.88 x64PCMark driven routine (light and heavy)
Adobe InDesign CC	Release 2015.4, Build 11.4.0.90 x64PCMark driven routine
Adobe Illustrator	Release 2015.3.0, Version 20.0.0 (64-bit)PCMark driven routine
Game Tests and Settings
Ashes of Singularity	Version 1.31.21360High Preset - 1920x1080, Mid Shadow Quality, 1x MSAACrazy Preset - 1920x1090, High Shadow Quality, 2x MSAA
F1 2015	2015 Season, Abu Dhabi Track, RainMedium Preset, No AFxUltra High Preset, 16x AF
Metro Last Light Redux	Version 3.00 x64High Quality, 1920x1080, High Tesselation, 16x AFVery High Quality, 1920x1080, Very High Tesselation, 16x AF
The Talos Principle	Version 267252Medium Preset, High Quality, High Tesselation, 4x AFUltra Preset, VeryHigh Quality, VeryHigh Tesselation, 16x AF

MORE: Best Motherboards

MORE: All Motherboard Content

Benchmark Results & Final Analysis

Using firmware revision 1.0, the X299E-ITX/ac was our first sample to completely adhere to the 140W TDP given by Intel for the Core i9-7900X processor. Rather than being a green initiative, this appears to be a way to cope with voltage regulator heat with the extra small heat sink. Intel XTU shows the board throttling below the expected threshold, and increasing firmware power limits to “maximum” still only gets it to the 170W previously seen in the ultra-efficient Prime-X299 Deluxe. The extra power enables an extra 300 MHz at max AVX load (Prime95 small-FFTs), though none of our performance benchmarks are brutal enough to reveal this. (Note: ASRock released firmware version 1.30 to make these limits correspond to builder-selected cooler modes (liquid or air) two weeks after our analysis.)

Meanwhile, the X299 Gaming Pro Carbon AC completely disregards power limits to improve performance (229W), the X299 Aorus Gaming 3 comes in at a much more reasonable 189W, and even ASRock’s own X299 Taichi hits 183W from the CPU. We benchmarked the entire series at motherboard defaults, save for enabling any disabled power-savings features and disabling any manufacturer-enabled overclocks (ie, fixed-ratio Intel Turbo Boost “enhancements”).

Synthetic Benchmarks

Synthetic benchmarks are a great way to see if someone is cheating or something is broken. They're more than just toys for reviewers and can also be used as in real-world diagnostics.

Anyone looking for the impact of ASRock’s lower default power limits won’t see them in PCMark or 3DMark, as these tests don’t significantly stress the CPU's AVX registers.

One might expect a big drop in Sandra’s Cryptography benchmark, but the performance difference in Hashing is slightly less than the 9% indicated by Prime95. That’s because the Cryptography bench requires high throughput, and the X299E-ITX/ac claws some of its score back with improved memory bandwidth.

The X299E-ITX/ac performs well in Cinebench, and even better in Compubench. Still, it’s not gaining enough to set off any alarms yet.

3D Games

While all the boards compete on-par in Ashes, Ultra Preset scores keep going up in F1 2015. Broader implementation of Intel Turbo Boost Max 3.0 may be the reason, as we restored a previous image just to make sure the improvement wasn’t coming by way of game updates. Our “High Preset” test is far more bandwidth-dependent, and the differences nicely match those seen above in Sandra Memory Bandwidth.

Unlike ASRock, high-end MSI boards come with an audio software suite that can help gamers locate their opponents, as well as synthesizing a 3D environment through stereo speakers and headset. Talos is compatible, yet there is a performance penalty for using the feature. A chart from our previous review shows how fast the board can be with Nahimic Audio Solution disabled.

Timed Applications

If you thought you’d see the difference in power ceiling reflected in timed application length, that’s just not happening. It appears that only a few synthetic benchmarks are capable of stressing the AVX registers enough to reach that power limit.

Power, Heat, & Efficiency

Because the synthetic benchmarks are capable of greater power draw than real world benchmarks, I’ve been using the average of idle and full load power to approximate efficiency differences. That’s worked on everything except notebooks and . . . X299 platforms. In order to test the X299E-ITX/ac in the same manner as the other boards, I had to use the uncapped power settings shown at the top of the page to generate these charts.

With that explanation out of the way, you’ll probably want to know what the numbers were with the caps in place. The X299E-ITX/ac’s full-load power drops to 206W at 3.30 GHz, and its temperatures deltas drop to 30/29K over ambient, with the caps in place.

The X299E-ITX/ac is far more efficient than its larger rivals, and this is evident even without excessive power throttling.

Overclocking

Overclocking the X299E-ITX/ac was easy, but testing that overclock was a complicated mess. To begin with, our default configuration has enough airflow to work with its tiny voltage regulator sink, if barely: The actual voltage regulator temperature was around 120°C! And while the monoblock dropped the voltage regulator temperature by around 60°C, it allowed the CPU to reach its thermal throttle point. Thus, our highest stable overclock was a test of several hours using our default Celsius S24 cooler and the motherboard’s included, undersized voltage regulator heat sink.

Performance is the reason to overclock, but some manufacturers have used extremely slow secondary and tertiary timings to improve memory overclocking capability. The result has occasionally been a performance reduction at ultra-high data rates, so we added a bandwidth chart.

Indeed, performance for the X299E-ITX/ac dropped significantly when pushed even slightly past the memory’s rated XMP settings, so I stopped at DDR4-3800. The reward for stopping there is an incredible 71 GB/s transfer rate!

Final Analysis

The starting price for an enthusiast-class X299 motherboard is around $250, and X299 Aorus Gaming 3 hit that price point (based on MSRP, rather than constantly fluctuating retail price points). The X299 Gaming Pro Carbon AC and X299 Taichi justify their higher prices with enhanced features, and the X299E-ITX/ac combines the Pro Carbon AC's 867mb/s controller with the Taichi's dual Gigabit Ethernet, but loses a bunch of connectors and the hardware that supports those to the Mini ITX form factor.

Perhaps the economies of volume have pushed X299E-ITX/ac pricing as ASRock attempts to spread development cost across the sale of fewer parts...but I just can’t bring myself to give the X299E-ITX/ac a value award. And I shouldn’t even have to think about that, because once again ASRock has produced the only board in its class.

By being the only LGA 2066 motherboard available to Mini-ITX builders, the X299E-ITX/ac automatically becomes the only possible choice for these builds. The price isn’t terrible either, and the performance level is spot-on for the Core i9-7900X even with the board’s lower-than-average default power ceiling. It’s not perfect, but it’s a little better than good enough to retain its award, and an impressive feat of packaging.

MORE: Best Motherboards

MORE: All Motherboard Content

ASRock X299 OC Formula Review

Thomas Soderstrom — Tue, 10 Oct 2017 13:00:00 +0000

Layout & Features

Many of our readers have told us that they’re sick of useless gadgets on their high-end motherboards. Everything from excess USB 2.0 ports to Wi-Fi and RGB lighting has been derided as a waste of money by people whose primary goals include performance, compatibility, and overclocking. Although we typically suggest lower-priced models to those buyers, overclocking is often a sticking point. So, what do we get if a manufacturer treats overclocking as a primary feature?

Former overclocking world champion and current ASRock employee Nick Shih has been pouring his time into the X299 platform to deliver optimal results from an ASRock motherboard, and the firm’s latest OC Formula is the result. Onboard RGB lighting is limited to the PCH sink and two cable outputs, just to please those who still want to light up their case. Resource shares are limited to the placement of SATA M.2 modules that steal SATA ports, since hardly anyone would limit their M.2 slot to SATA anyway.

Specifications

The OC Formula has many visible similarities to ASRock’s mid-priced X299 boards, but going by visual cues is like comparing the AC Ace to the Shelby AC/Cobra. A quick look at the missing memory slots and additional PCIe slot are your first clues to the serious changes underneath, and the I/O panel gap where you might expect to find a Wi-Fi card doesn’t even have solder points on the X299 OC Formula.

That’s not to say the I/O panel is sparse, as it still uses an ASM3141 USB 3.1 Gen 2 controller to serve up a Type-C and Type A port from two PCIe 3.0 lanes, in addition to its four USB 3.1 ports, legacy USB 2.0 and PS/2 ports for your keyboard and mouse, dual Intel-based Gigabit Ethernet ports, CLR_CMOS and BIOS Flashback buttons, digital optical out, and five analog audio jacks.

Function buttons and switches are moved up to the top front corner, adding overclocker-specific features such as BFG (boot failure guard, aka safe boot mode), BCLK +/-, boot retry, and a “Menu” toggle button to change from BCLK to CPU or cache ratio, to the more-traditional power and reset button. Extra switches are also there for overclockers, including for switches to disable graphics cards if one of them causes the system to hang (from overclocking, of course), a “Slow Mode” switch that can help LN2 users get past “cold bug” boot problems by clocking the system down, and an LN2-mode button that’s supposed to allow firmware to handle the latter task automatically during the boot process. While we’re staring at that corner, we’d might as well mention the VROC connector that allows RAID firmware modules to be added to the CPU’s integrated PCIe controller.

Since both M.2 slots use the chipset’s controller, the VROC connector is only useful with devices that mount in traditional PCIe slots. The X299 OC Formula adds a 4-lane, open-ended slot that might be useful for storage, but that slot only functions when 44-lane CPUs are installed. The middle slot might be even more useful for storage, with its x8 interface, but that slot also requires a 44-lane CPU. 28-lane processors “make up” for the fact that they aren’t 32-lane by dropping the bottom slot to x4 mode, and 16-lane processors lose the second x16-length slot in addition to the center slot and x4 slot.

Since the bottom slot drops to x4 mode with 28-lane or 16-lane CPUs, 4-way SLI is limited to 44-lane CPUs. CPUs with 28 lanes can run 4-way CrossFireX or 3-way SLI, and while 16-lane CPUs can run either SLI or 3-way CrossFireX, the latter requires the third card to take the slow path from the chipset. The total of 18 2-lane pathway switches located between the various slots can thus be attributed in part to Intel’s decision to add a 16-lane CPU to an X299 platform that shouldn’t have been designed for such constraints. Those 16-lane Kaby Lake X processors also have but two memory channels, reducing the X299 OC Formula to two functional memory slots.

All eight of the X299 OC Formula’s SATA connectors face forward to avoid card interference, even the two that are located above all the cards. A front-panel USB 3.1 (10Gb/s for Type-C ports) header is found just above the centerline of the top card, just below the USB 3.0 front-panel header, which is just behind a forward-facing PCIe power connector. The purpose of that last connector is to feed extra power to PCIe slots, although few cards need it. Another noteworthy header is the second front-panel audio output, which is connected in common with the first, but turned 90° towards the bottom edge for improved cable management.

The X299 OC Formula includes a 4-way SLI bridge, two 3-way SLI bridges at different slot spacing, an HB SLI bridge, an I/O shield, four SATA cables, and even a socket backplate that makes it possible to use an ancient LGA-1366 cooler with this LGA-2066 motherboard. That last item may be important to LN2 users who haven’t updated the mounting hardware of their favorite pot.

MORE: Best Motherboards

MORE: All Motherboard Content

Software & Firmware

There haven’t been any major changes to ASRock’s software suite since our Gaming i9 review, although the X299 OC Fomula doesn’t include the Gaming i9’s Creative audio software. A minor update has added firmware compatibility to ASRock’s RGB LED software application. Users can set the heat sink light to several colors, including rainbow cycles, and patterns that include breathing and strobing. The two RGB outputs can be set either synchronously or asynchronously depending on the user’s taste.

Formula Drive replaces F-Stream and A-Tuning applications of ASRock’s other motherboard series, although these are essentially different names for the same application. The application was not able to access firmware-based overclocking profiles, but its manual tuning worked as expected. Maximum settings are shown based on the limits of firmware.

Firmware opens by default to the advanced UEFI interface, where the “Main” page has a section that allows users to store favorite settings.

Moving over to the OC Tweaker menu, we find Nick Shih’s LN2 profiles, ASRock’s regular “Optimized CPU OC” settings, and several submenus to control overclocks.

Every Nick Shih OC is based on a 50x CPU ratio at 1.45V CPU core, with other voltage levels varying with CPU model. For example, the 7800X through 7900X configurations use a 2.3V CPU input, while the 7920X through 7980EX use 2.45V.

“Optimized” overclocks are far more limiting, with the 4.2 GHz setting having negative core multiplier offsets of 2x for AVX2 and 3x for AVX3 mode, and the 4.8 GHz setting offsetting the core multiplier by 8x and 15x, respectively. We like to test the core at a true maximum load, and found that the AVX setting of “Auto” gave us no AVX offset whatsoever.

Using the same heat-determined 1.15V CPU core as we’ve used on other boards, we reached the same 4.40 GHz overclock at 10-core, 20-thread, Prime95 AVX load. Additional setting changes were not required, as the OC Formula’s default power and thermal limits fully supported this processor at those settings.

The X299 OC Formula reached a super-high DDR4-4122 overclock at full four-DIMM stability, proving the usefulness of its single-DIMM-per-channel slot layout. Users who want to tweak their memory beyond basic SPD or XMP timings will find the full range of settings farther down the DRAM Configuration submenu.

Merely changing the CPU ratio mode from “Auto” to “Manual” causes the X299 OC Formula to change its CPU Input voltage from “Auto” to “2.10V Fixed.” Since several of the “Optimized” settings were based on a 1.90V fixed input, I simply changed this value back to “Auto.” The “Level 2” load-line calibration produced the expected stability and thermal values we’ve previously seen when overclocking this CPU, and the 1.35V DIMM setting produced a measured 1.353V at the DIMM slots.

The “Advanced” menu of UEFI’s “Advanced” mode includes a setting (near the bottom) to change the UEFI boot interface back to “Easy” mode. One can also get into “Easy” mode from “Advanced” mode by pressing the keyboard’s F6 function key.

The “Tools” menu of UEFI contains everything mentioned in our Gaming i9 review, including an RGB menu identical to that of ASRock’s software application. The software application operates by changing the value in firmware.

Only two of the X299 OC Formula’s headers can be switched from PWM to voltage-based fan control, and those same two headers are tagged for use with pumps up to 1.5A. The other headers have a 1.0A limit. You can set compatible fans using various programed profiles within the FAN-Tastic Tuning submenu.

Although the X299 OC Formula can be set to open its “Easy Mode” interface by default, that setting is found within “Advanced Mode.”

MORE: Best Motherboards

MORE: All Motherboard Content

How We Test

Test Hardware

Sound	Integrated HD Audio
Network	Integrated Gigabit Networking
Graphics Driver	GeForce 382.53

Comparison Products

ASRock X299 OC Formula

Asus Prime X299-Deluxe

MSI X299 XPower Gaming AC

Benchmark Settings

Synthetic Benchmarks and Settings
PCMark 8	Version 2.7.613Home, Creative, Work, Storage, Applications (Adobe & Microsoft)
3DMark 13	Version 4.47.597.0Skydiver, Firestrike, Firestrike Extreme Default Presets
SiSoftware Sandra	Version 2016.03.22.21CPU Arithmetic, Multimedia, Cryptography, Memory Bandwidth
DiskSPD	4k Random Read, 4k Random Write128k Sequential Read, 128k Sequential Write
Cinebench R15	Build RC83328DEMOOpenGL Benchmark
CompuBench	Version 1.5.8Face Detection, Optical Flow, Ocean Surface, Ray Tracing
Application Tests and Settings
LAME MP3	Version 3.98.3Mixed 271MB WAV to mp3: Command: -b 160 --nores (160 Kb/s)
HandBrake CLI	Version: 0.9.9Sintel Open Movie Project: 4.19 GB 4k mkv to x265 mp4
Blender	Version 2.68aBMW 27 CPU Render Benchmark, BMW 27 GPU Render Benchmark
7-Zip	Version 16.02THG-Workload (7.6 GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=9"
Adobe After Effects CC	Release 2015.3.0, Version 13.8.0.144PCMark driven routine
Adobe Photoshop CC	Release 2015.5.0. 20160603.r.88 x64PCMark driven routine (light and heavy)
Adobe InDesign CC	Release 2015.4, Build 11.4.0.90 x64PCMark driven routine
Adobe Illustrator	Release 2015.3.0, Version 20.0.0 (64-bit)PCMark driven routine
Game Tests and Settings
Ashes of Singularity	Version 1.31.21360High Preset - 1920x1080, Mid Shadow Quality, 1x MSAACrazy Preset - 1920x1090, High Shadow Quality, 2x MSAA
F1 2015	2015 Season, Abu Dhabi Track, RainMedium Preset, No AFxUltra High Preset, 16x AF
Metro Last Light Redux	Version 3.00 x64High Quality, 1920x1080, High Tesselation, 16x AFVery High Quality, 1920x1080, Very High Tesselation, 16x AF
The Talos Principle	Version 267252Medium Preset, High Quality, High Tesselation, 4x AFUltra Preset, VeryHigh Quality, VeryHigh Tesselation, 16x AF

MORE: Best Motherboards

MORE: All Motherboard Content

Benchmark Results & Conclusion

Intel specified a 140W TDP for our Core i9-7900X CPU, but the CPU can’t operate within that limit at both full speed and full AVX load. While companies that make heat-constrained units such as servers may be stuck with the inevitable performance drop, the firm hasn’t been so finicky with the way retail motherboard makers treat enthusiast models. Of the boards we’ve tested, only the Prime X299-Deluxe from Asus stayed within Intel’s stated limit.

Given the X299 OC Formula’s target market, we were a little surprised to see it come in second place in TDP. It also takes second place in our measurement from the wall socket, though that metric is impacted by additional components, as even a 4W Wi-Fi controller can force a position swap. The X299 OC Formula consumes only 3W more power on average than the Prime X299-Deluxe, but also has far fewer on-board controllers to feed.

Yet our power measurements were taken with all CPU-integrated power-saving technologies enabled, and the target market for the X299 OC Formula is likely to disable most, if not all, green features. Doing so drastically alters the power needs of any motherboard, and would thereby negate the following benchmarks. On the other hand, it’s always nice to have a “stock” baseline, no?

Synthetic Benchmarks

Synthetics used to be a great way to catch people cheating, until everyone started cheating by various degrees. It’s still a good way to look for problems, though.

PCMark shows a slight deficiency in the Creative scores of both the X299 OC Formula and X299 XPower Gaming AC. Will this have an impact on real-world-based Creative Suite and MS Office task completion times?

The two most-miserly boards (Asus and ASRock samples) also had the lowest Integer x32 throughput in Sandra Multimedia. Let’s see if we find matching deficiencies in any of our real-world benchmarks!

3D Games

Though it looks great in Ashes, the X299 OC Formula falls behind at our medium preset of F1 2015. That setting is normally affected by memory performance, yet there weren’t any deficits detected in Sandra Memory Bandwidth.

The X299 OC Formula also comes up a little short in Talos. The MSI XPower Gaming AC’s performance was impacted by inclusion of Nahimic Audio solution, and a look at its original benchmark chart shows how it leads when that software is disabled.

Timed Applications

The X299 OC Formula and Prime X299-Deluxe fell slightly behind in Handbrake, commensurate to their power savings and Sandra Multimedia deficits.

The X299 OC Formula took significant losses in both Adobe CC and MS Office timed tests, which are both driven by PCMark. These losses are also spotted in one of eight game tests, and appear to have something to do with memory. After examining both benchmarks, all firmware settings, the PCMark suite, and even the F1 2015 test, the best guess I could come up with was that the Beta BIOS 1.11 may need further development. I normally refrain from using Beta firmware, but this is the only version ASRock has made public, and I always put public firmware first.

Heat & Efficiency

We saw at the top of the page that the X299 OC Formula consumes barely more energy than the miserly Prime X299-Deluxe, and those low power numbers produced low temperatures to match.

Big advantages in energy consumption are offset slightly by performance deficits to put the X299 OC Formula in second place, behind the Prime X299-Deluxe for efficiency.

Overclocking

Overclocking is the X299 OC Formula’s purpose, yet our CPU can’t transfer heat fast enough from its core to its heat spreader to take advantage of the extra voltage needed for a high core O/C. What it can do is drive ultra-fast memory, and the X299 OC Formula obliges with a record-pace overclock. Remember, this memory is “only” rated at XMP-3866.

Overclocking may be the purpose of the X299 OC Formula, but performance is the purpose of overclocking. The problem for ASRock is the MSI’s X299 XPower Gaming AC already set a bandwidth record at a mere DDR4-3838, and the OC Formula’s DDR4-4122 isn’t able to break that record.

The X299 XPower Gaming AC also costs more than the X299 OC Formula, which is a hard pill to swallow given the OC Formula’s already-high $400 tag. Yet all of the competing boards have additional features, and even people who don’t want those features must admit that they cost something. I could go through a list of the features you may not want, such as premium Wi-Fi on all three competitors, enhanced RGB lighting on all three competitors, the thermistors for remote heat monitoring included by all three competitors, the included PCIe x8 to 2x PCIe x4 M.2 adapter card offered by MSI . . .

And the X299 XPower Gaming AC wasn’t even the cheapest premium board. That honor belonged to the Gigabyte sample. On a features basis, ASRock’s own Fatal1ty Professional Gaming i9 would be the board to beat these three competitors. Indeed, that sample already received my recommendation.

The only award the X299 OC Formula could apply for is Editor’s Choice, but the only thing I’d probably choose it for is to reach the highest-possible memory overclock. I may still do that. Yet without a performance benefit for that added clock speed, I’m left wishing that the OC Formula was also a better value in overclocking. To get there, it would need to cost around 10% less.

MORE: Best Motherboards

MORE: All Motherboard Content

MSI X299 XPower Gaming AC Motherboard Review

Thomas Soderstrom — Mon, 02 Oct 2017 13:00:00 +0000

Features & Specifications

Enthusiast-class motherboards often have a little extra, including extra room to put extra features such as extra connectors and buttons. MSI’s X299 XPower Gaming AC is the latest in the company's own oversized enthusiast-class boards, but before we get to what that extra space gives you, let’s consider for a minute what the board is—at 10.7”-deep by 12” top-to-bottom—and what it isn’t.

It isn’t 13” deep, so the EATX label doesn’t fully apply, even though MSI uses it. It isn’t XL-ATX, because those boards have an extra 0.8 to 1.3” at the bottom edge for an eighth slot. Size wise, it’s the least of these two formats. And I only stress this small fact because it fits both types of PC case, along with any ATX case that has an extra 1.1” of free space at the motherboard’s front edge.

I also avoided saying 12” high, because that would likely infuriate anyone who came into the enthusiast market from the server industry.

Without further ado, here’s what the X299 XPower Gaming AC offers:

Specifications

On the I/O panel, the X299 XPower Gaming AC gets you two extra USB 3.1 Gen 1 (aka USB 3.0) ports and an extra RJ45 network jack, the latter served up by Intel’s venerable i211AT PCIe-based Gigabit Ethernet controller. We still find Intel’s high-end 867Mb/s Wi-Fi solution, Type A and Type C 10Gb/s USB 3.1 (aka Gen2) ports, five analog audio jacks, optical S/PDIF, a PS/2 and two USB 2.0 ports for legacy peripherals (keyboards, mice, and printers), a CLR_CMOS button, and a USB Flashback+ jack for flashing firmware without the need of a compatible CPU or RAM.

On the surface we find three M.2 slots with heat spreaders, the top slot featuring MSI’s FROZR design with quick-release latch. The third M.2 takes data directly from the CPU, providing full PCIe 3.0 x4 bandwidth without the sharing issues of the chipset’s DMI pathway. This design also prevents 16-lane (Kaby Lake-X) processors from supporting SLI, since the remaining slots are stuck at x8/x4 instead of the required x8/x8 mode. Not that anyone should feel compelled to put a Kaby Lake-X on this $400 board: Skylake-X processors make full use of the board’s capabilities by providing 44 or 28 lanes, depending on the model.

There’s also an extra pair of SATA ports and a U.2 connector for off-board PCIe x4 storage. The added controller for those SATA ports shares a pathway with the PCIe x1 slot, so that users must choose which interface they want most. SATA sharing doesn’t end there, as the center M.2 slot gets its four HSIO resources for PCIe x4 mode from SATA ports 5 through 8. A SATA M.2 drive placed there consumes only port 5. And a SATA drive in the top M.2 slot consumes SATA port 1.

Generously equipped with 10 four-pin fan headers that can be manually or automatically switched between PWM and voltage-based controls, X299 XPower Gaming AC also adds a new-generation 10Gb/s USB 3.1 front panel header to its pair of USB 3.0 headers. Of these, the bottom USB 3.0 header could possibly get in the way of a four-way graphics card installation, particularly if the bottom card is more than 9.5” long. Perhaps it’s good then that hardly anyone uses the second USB 3.0 interface?

Overclockers who prefer open systems can make use of the X299 XPower Gaming AC’s overclocking buttons, with BCLK step sizes that can be altered in UEFI. A quick adjustment knob also lets those who don’t know much about overclocking use one of MSI’s factory-configured overclocking profiles. Voltage check points closer to the front edge let you detect the rail voltages of the CPU, DRAM, VSA, VCCIO, and include taps into Core and Ring voltage lines. Next to those are LEDs that indicate initialization of the firmware (Boot), graphics card, DRAM, and CPU to show overclockers which component is stopping the boot process. A Port 80 display serves the same purpose and is easier to see from a distance.

Not related to overclocking but more useful to some of my colleagues, the VRAID connector enables VROC modules for a CPU’s integrated PCIe controller.

Ironically called “Chest Plate,” an anodized brushed-aluminum plate is MSI’s solution for an ugly backside that presents itself in certain glass cases. It’s also said to reduce flex, which is likely why it spans the extra inch forward of standard ATX standoffs on this 10.7”-deep design. As a practical matter, if the X299 XPower Gaming AC’s forward edge were to be flexed far enough to touch the motherboard tray, its “Chest Plate” would prevent circuit contact.

Like most MSI motherboards, the X299 XPower Gaming AC includes a hefty load of documentation and stickers. Unlike its cheaper siblings, it also includes a significant installation pack. We find two magnetic antennas for its 867Mb/s Wi-Fi controller, an RGB Y-cable for its single lighting header, a case badge, a pack of 3D X-Mounting standoffs for custom facades that users can 3D print from available online designs, a thermistor cable that connects to an onboard header for remote component monitoring, an HB-type SLI bridge, MSI’s M.2 Xpander-Z PCIe to M.2 adapter card, and six SATA cables including three with a single right-angle connector.

MSI’s M.2 Xpander-Z adapts eight CPU lanes to two four-lane M.2 cards, and serves as another reason for buyers to choose Skylake-X (44 or 28 lanes) over Kaby Lake-X (16-lanes).

MORE: Best Motherboards

MORE: All Motherboard Content

Software & Firmware

MSI has made few changes to its software suite since we covered it in our X299 Gaming Pro Carbon AC review, but we did grab a few supplemental screenshots. MSI Mystic Light changed our LED colors to red and didn’t respond to custom color changes until we updated, after which it became unresponsive to most input (even flash patterns). We’ll wait for yet another update.

OC Mode of MSI’s Gaming App still references the motherboard’s first OC profile, which is still an underclock for our processor but at 3.8 GHz rather than the 3.5 GHz of the former test subject. Other shortcuts still refer to more useful apps as described in that review.

And now for the good news: MSI Command Center has been fully updated to work with X299 motherboards. We controlled CPU clocks and core voltage with ease. On the other hand, DRAM clocks are still stuck.

The “Game Boost” setting above can be set to hardware or software mode, accessing seven different factory-programed overclocks from firmware. Since it’s purely firmware controlled, changes require a reboot.

Applets launched from the Advanced key of Command Center include advanced voltage control, system fan settings, advanced DRAM timings, and a very useful temperature monitoring schematic. DRAM timing adjustments worked without crashing the program for the first time I can remember, though these firmware-based adjustments require reboot.

The “Settings” key of Command Center launches a system event recorder, an alarm settings menu, and mobile control for MSI’s Command Center smartphone app.

Command Center’s “Information” key links its Motherboard, Memory, and CPU info applets, along with detailed HW Monitor readings.

X299 XPower Gaming AC Firmware

MSI Click BIOS 5 still opens to Easy mode by default, but now remembers the mode it exited from. That means if you press F7 to enter advanced mode, change a few things, save and exit, you’ll be greeted with Advanced mode the next time you enter UEFI.

The OC menu also opens into Expert mode by default. The Misc Setting drop-down includes Enhanced Turbo mode by default, and this must be disabled to benchmark using Intel’s default Turbo Boost ratios. That setting disappears when you manually select a multiplier.

Our CPU reached its expected 4.40 GHz clock at 1.15V, where substantially higher voltage levels that might have kept it stable at higher frequencies would have also allowed it to overheat under full load. DRAM voltage is only adjustable in 10mV increments: The 1.340V setting allowed measured voltage to exceed our 1.355V test limit, forcing us to choose 1.330V to achieve 1.348V as measured at the DIMM slot.

Memory timing adjustments are too numerous to write about, and can be set at once or split into separate A/B and C/D channel timings. The X299 XPower Gaming AC extracted DDR4-3838 from our DDR4-3866 set at XMP timings.

Since 4.40 GHz is within the standard range of our Core i9-7900X, the X299 XPower Gaming AC treated it that way and maintained certain power savings controls. The 45x multiplier brought us an unregulated 4.50 GHz, but the voltage required to maintain it caused our CPU to thermal throttle. Getting all cores to operate simultaneously at 4.40 GHz under an AVX load required setting the AVX offset to zero and setting the VCore Loadline Calibration control to “anything other than default,” whereupon we found Mode 3 to be closest to a level setting.

Additional submenus show CPU capabilities and DRAM timings.

The CPU Features menu doesn’t have very many power savings control settings. We enabled those for our performance and power evaluations, then disabled several prior to overclocking analysis.

The M-Flash button isn’t a menu, but a command to reboot into a special firmware-flashing GUI.

Six custom firmware configurations can be saved as overclocking profiles or exported to a file on a USB flash drive.

All 10 of the X299 XPower Gaming AC’s fan headers automatically detect whether or not a PWM-controlled device is attached, and all 10 can alternatively be configured manually between PWM and voltage-based speed control. Fan control profiles can be configured automatically, manually, or set to full speed.

MSI Board Explorer shows detected devices, but apparently isn’t able to detect whether USB or SATA devices are installed. It shows all internal ports occupied, though none were.

MORE: Best Motherboards

MORE: All Motherboard Content

Test Configuration

Test Hardware

Sound	Integrated HD Audio
Network	Integrated Gigabit Networking
Graphics Driver	GeForce 382.53

Benchmark Settings

Synthetic Benchmarks and Settings
PCMark 8	Version 2.7.613Home, Creative, Work, Storage, Applications (Adobe & Microsoft)
3DMark 13	Version 4.47.597.0Skydiver, Firestrike, Firestrike Extreme Default Presets
SiSoftware Sandra	Version 2016.03.22.21CPU Arithmetic, Multimedia, Cryptography, Memory Bandwidth
DiskSPD	4k Random Read, 4k Random Write128k Sequential Read, 128k Sequential Write
Cinebench R15	Build RC83328DEMOOpenGL Benchmark
CompuBench	Version 1.5.8Face Detection, Optical Flow, Ocean Surface, Ray Tracing
Application Tests and Settings
LAME MP3	Version 3.98.3Mixed 271MB WAV to mp3: Command: -b 160 --nores (160 Kb/s)
HandBrake CLI	Version: 0.9.9Sintel Open Movie Project: 4.19 GB 4k mkv to x265 mp4
Blender	Version 2.68aBMW 27 CPU Render Benchmark, BMW 27 GPU Render Benchmark
7-Zip	Version 16.02THG-Workload (7.6 GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=9"
Adobe After Effects CC	Release 2015.3.0, Version 13.8.0.144PCMark driven routine
Adobe Photoshop CC	Release 2015.5.0. 20160603.r.88 x64PCMark driven routine (light and heavy)
Adobe InDesign CC	Release 2015.4, Build 11.4.0.90 x64PCMark driven routine
Adobe Illustrator	Release 2015.3.0, Version 20.0.0 (64-bit)PCMark driven routine
Game Tests and Settings
Ashes of Singularity	Version 1.31.21360High Preset - 1920x1080, Mid Shadow Quality, 1x MSAACrazy Preset - 1920x1090, High Shadow Quality, 2x MSAA
F1 2015	2015 Season, Abu Dhabi Track, RainMedium Preset, No AFxUltra High Preset, 16x AF
Metro Last Light Redux	Version 3.00 x64High Quality, 1920x1080, High Tesselation, 16x AFVery High Quality, 1920x1080, Very High Tesselation, 16x AF
The Talos Principle	Version 267252Medium Preset, High Quality, High Tesselation, 4x AFUltra Preset, VeryHigh Quality, VeryHigh Tesselation, 16x AF

MORE: Best Motherboards

MORE: All Motherboard Content

Benchmarks & Final Analysis

We’re putting the overclocking section of today’s analysis atop the page since it’s what you want to see most. This also gives me an excuse to top the page with the firmware settings of each competing motherboard. Some readers will certainly want to jump past boring benchmark charts to the conclusion, but you never know when something weird will happen there.

Frequency And Voltage Settings

MSI X299 XPower Gaming AC

Asus Prime X299-Deluxe

Gigabyte X299 Aorus Gaming 7

MSI’s X299 XPower Gaming AC takes third place in a hotly-contested DRAM overclocking race, where its competitors use more voltage. That’s MSI’s fault, since the voltage regulator is adjustable only on 10mV steps, forcing me to choose 1.348V rather than a test-violating 1.358V. The 5mV increments of competitors simply let me choose something between those two options.

Of course the X299 XPower Gaming AC matched its competitors in CPU overclocking, which is why I mentioned DRAM first. It also matched competitors in base-clock overclocks, but only at the 1:1 boot strap. X299 XPower Gaming AC firmware doesn’t have a boost strap adjustment, and setting BCLK to 125 MHz failed to initialize an automatic change.

If performance is the point of overclocking, we might as well end our analysis here. The X299 XPower Gaming AC has the highest memory bandwidth ever extracted from our consumer motherboards. And given that a 4.40 GHz CPU overclock performs virtually identically across all boards, that’s the one place where any single product could really stand apart from the rest.

Power Metrics

Since we assume all overclockers will want to set up clocks themselves, we take every possible measure to eliminate automatic overclocking from our performance tests. We’d hate for someone to assume that “if brand X performs Y% better than brand Z at stock, it will also perform that much better overclocked,” because this simply isn’t true. Yet enabling all available CPU-based power-saving features and disabling automatic overclocks is no longer a perfect method to achieve similar results, because Intel is allowing motherboard manufacturers to violate its stated 140W TDP on consumer-market enthusiast motherboards. The Core i9-7900X rarely runs at its full performance level at full load but is instead clocked down to stay within a power and heat envelope: Expanding beyond that envelope allows power inconsistencies that we can’t properly compensate.

Seeing that everyone but Asus uses an unlocked TDP by default, we’d like to see how much this costs us at the power plug before moving on to see how it impacts performance.

Even if we assume that the extra features of the X299 XPower Gaming AC and X299 Aorus Gaming 7 costs a few watts, both far exceed the X299 Gaming i9, and even that model appears a power hog compared to the Intel standard maintained with the Prime X299 Deluxe.

All of this means that even with all power savings features enabled and all overclocking techniques disabled, the X299 XPower Gaming AC and X299 Aorus Gaming 7 are running close to the limit of this CPU’s thermal interface. Of course, we could try to force compliance by using a 140W CPU cooler, but that’s not a realistic limitation of the enthusiast performance market. Let’s see how these things affect performance.

Synthetic Benchmarks

For all of its extra power consumption, an uncapped TDP does little to alter 3DMark and PCMark scores. The X299 XPower Gaming AC leads the closest-power-level X299 Aorus Gaming 7 through 3DMark, but that could be due to its enhanced memory performance.

It’s not until we reach Sandra Memory Bandwidth that we see just how great the X299 XPower Gaming AC’s memory bandwidth advantage is. Remember that this is at DDR4-2133 defaults, not even the XMP settings that were too high for this particular board to use.

3D Games

We expected a board with the memory bandwidth advantage of the X299 XPower Gaming AC to wipe the floor with its competitors in the memory-sensitive F1 2015 benchmark, but were surprised to see it also take a noteworthy lead at the lower test settings of Ashes. The reason it’s noteworthy isn’t that it’s visible in game play, but rather that it falls outside of the normal variation between consecutive tests.

The X299 XPower Gaming AC includes Nahimic audio solution, with its ability to virtualize 3D through headphones and help you locate your enemies in games. The only game in our suite affected is Talos, where the darker bars show how much lower the score is when Nahimic is running. Since Nahimic is a major feature, the lower numbers are used in our final tally.

Timed Applications

The X299 XPower Gaming AC burns through our mixed workloads with flair, but drops back a little in Adobe CC and MS Office workloads. In the balance we expect . . . balance.

Efficiency

The performance penalty imparted by Nahimic in one of our games offset the performance gained in F1 2015. Similarly, performance losses in Adobe CC and MS Office offset gains in mixed workloads. The 1% combined performance deficit is pitted against its 16% above-average power consumption level to produce sub-par efficiency.

Final Analysis

People like value charts that show “bang for the buck,” but added features don’t produce a performance bang. We try to make the comparison fair by comparing similarly-priced boards and then discussing feature differences.

The X299 Aorus Gaming 7 and X299 Gaming i9 are the boards for MSI to beat with superior features. All three have at least dual Gigabit Ethernet and some kind of 802.11ac controller, but X299 Professional Gaming i9 also adds 10 Gb/s Ethernet as a third wired network interface. That’s hard to match on the networking front, let alone beat. That same board has a weaker 433 Mb/s Wi-Fi controller, but that concession hardly puts a dent in the value of its wired interface configuration.

The Gigabyte board has five x16-length slots but can connect three cards in SLI, even when equipped with a 44-lane CPU. ASRock’s sample is wired for four-way Crossfire and SLI, but slot spacing issues limit most cards to 3-way configurations. The MSI X299 XPower Gaming AC can do 4-way CrossFire or SLI with a 44-lane CPU, 3-way with a 28-lane CPU, and . . . no-way SLI with a 16-lane CPU because four of its lanes are dedicated to M.2. Talk about a mixed bag.

Thus, while the X299 XPower Gaming AC has all the features needed to justify its price, it can’t convincingly beat its competitors in features for the money. You’ll probably buy it anyway if you need 4-way SLI in addition to its high-end features, and we might even be tempted by its high memory bandwidth. I’d even give it an award for that last accomplishment, if any of our awards were applicable. Yet as I sit here looking at its PCIe 3.0 x8 to dual M.2 adapter card, I think to myself, I can’t find one of these for sale anywhere. That’s something. And if you’re in the market for 3-way CrossFire or SLI and three CPU-driven M.2 drives, that card may be the only thing the X299 XPower Gaming AC needs to get your stamp of approval.

MORE: Best Motherboards