Sign in with
Sign up | Sign in
System Builder Marathon, Q4 2013: A $2400 PC That Costs $2700
By ,
1. Making Tough Choices In Volatile Markets

System Builder Marathon, Q4 2013: The Articles

Here are links to each of the four articles in this quarter’s System Builder Marathon (we’ll update them as each story is published). And remember, these systems are all being given away at the end of the marathon.

To enter the giveaway, please fill out this SurveyGizmo form, and be sure to read the complete rules before entering!

Day 1: The $800 Gaming PC
Day 2: The $1600 Enthusiast PC
Day 3: The $2400 Performance PC
Day 4: Performance And Value, Dissected

Introduction

Picking a top-performing build for the last System Builder Marathon of 2013 was as easy as looking to the triumphs and tribulations of the previous quarter's configuration. My very first issue was that, if I hoped to compete on the basis of value, I was spending too much money.

Unfortunately, that's the way it always seems to go for me. Any high-end PC I put together costs too much once I blow past the value sweet spot, which previous analysis has shown to be under $1000. But rather than toss out the idea of a high-end build in its entirety, we decided to compromise this quarter, spending three (rather than four) times as much as the lowest-price model.

My last stab at the crown benefited from the Sandy Bridge-E architecture's extra cores, but didn’t overclock very well. Today, the Ivy Bridge-E design takes its place. Our motherboard of choice wasn't very tweakable last time around, so it gets replaced with an award-winning alternative from a past round-up. And I had to fight my DDR3 memory back in Q3 too. So, I went back to using tried-and-true modules from a different manufacturer.

I also revisited my awards list for a closed-loop liquid cooler, and a new case was picked to support the oversized cooler. Now I was pretty confident that I could bolster the productivity-oriented tests. Time to move on to the games...

Although they were powerful, the three GeForce GTX 760s I used didn't give me the scaling I was hoping for. To explore this further, I went back and wrote SLI Scaling: Can Three GeForce GTX 760s Beat Two 780s? The problem was that the two 780s I would have wanted instead were overpriced. AMD forced Nvidia’s hand by dropping its $400 Radeon R9 290 into the market, making the GK110-based board more tenable. That left us to choose between the second-fastest cards from both companies.

Q4 2013 $2400 Performance PC Components
ComponentModel NameOriginal PriceCurrent Price
ProcessorIntel Core i7-4930K: 3.4 to 3.9 GHz, Six Core, 12 MB Shared L3 Cache$580 $580
Graphics2 x Asus R9290-4GD5 Radeon R9 290 4 GB (CrossFire)$800 $1060
MotherboardASRock X79 Extreme4: LGA 2011, Intel X79 Express$230 $220
MemoryG.Skill Ripjaws X F3-14900CL9Q-16GBXL: DDR3-1600 C9, 16 GB (4 x 4 GB) $139 $180
System DriveSanDisk Ultra Plus SDSSDHP-256G-G25: 256 GB, SATA 6Gb/s SSD$175 $170
Storage DriveWestern Digital Green WD20EZRX: 2 TB, SATA 6Gb/s Hard Drive
$90 $90
OpticalPioneer BDR-208DBK: 15x BD-R, 16x DVD±R$70 $60
CaseFractal Design Define R4 Black Pearl$110 $100
PowerSeasonic M12II SS-850AM: 850 W Semi-Modular, ATX12V v2.3, 80 PLUS Bronze$120 $135
CPU CoolerThermaltake CLW0217 Water 2.0 Extreme$95 $95
PWM FanAntec Spot Cool Blue LED Fan $11
  Total Cost $2409 $2701

The choice was made easier by the fact that R9 290s sold for $400, while the GeForce GTX 780 still went for $500. The caveat, of course, is that AMD's reference cooler isn't the most highly regarded thermal solution around. Chris and I have an ongoing rivalry concerning these things, as I’m primarily concerned about case heat, while he’s easily annoyed by noise.

My configuration was priced at $2409 when I placed my order from Newegg. However, some of the parts are a lot more expensive today. Let's talk about those components first.

2. Graphics, Memory, And CPU

Graphics: Two Asus R9290-4GD5 In CrossFire

The choice was easy a month ago. If AMD's $400 Radeon R9 290 could roughly match the performance of Nvidia’s GeForce GTX 780, an $800 pair of 290s should outperform September’s trio of GTX 760s at $900. More performance for less money was the theory, before Hawaii-based boards shot up in price online and the increase was blamed on a crypto-currency gold rush. Now we're stuck with two $530 cards that are identical to the $400 boards we purchased a month ago. Talk about buyer's remorse. 

Read Customer Reviews of Asus' R9290-4GD5 Radeon R9 290


Choosing a brand wasn’t hard, since all of the cards we've seen in stock follow AMD's reference design. Since Chris’s best retail sample came from Asus, I went that route as well.

DRAM: G.Skill Ripjaws X F3-14900CL9Q-16GBXL

Several of our memory reviews have shown that the best gaming experience comes from DDR3-2133 with optimized timings. Those same articles showed some of our benchmarks slowing down when we used DDR3-2400, probably because motherboards offset higher data rates with relaxed timings that are more difficult to optimize.

The problem with my previous system was that its DDR3-1600 wasn’t overclockable. In fact, I had to use overclocked voltage levels just to guarantee stability at its rated settings. I’m not going to make that mistake again!

Read Customer Reviews of G.Skill's Ripjaws 16 GB DDR3 Memory Kit


G.Skill sells the same modules in various colors, with various heat spreaders (Ripjaws or Ares), and under various model numbers. I’ve been using these 4 GB DIMMs for a couple of years, and find that, while they don’t always win round-ups, at least they overclock with a fair amount of consistency. Unless there’s a problem with the motherboard or the CPU’s on-die memory controller, I expect to reach DDR3-2133 while only paying for DDR3-1866.

Formerly a good value, these modules are also 30% more expensive than the day we placed our order. Ouch!

CPU: Intel Core i7-4930K

With the same 12 MB shared L3 cache as its award-winning predecessor and the advanced architecture of its flagship sibling, Intel’s Core i7-4930K is an easy choice to replace my previous machine’s Core i7-3930K.

Read Customer Reviews of Intel's Core i7-4930K


I’m also expecting a little more forgiveness from this part when it comes to overclocking, whether that comes from lower power consumption or the 22 nm manufacturing process. A 200 MHz-higher baseline clock rate is the only actual evidence I have to support my enthusiastic expectations.

3. Motherboard, CPU Cooling And Case

Motherboard: ASRock X79 Extreme4

September’s build was only moderately successful with ASRock’s mid-market X79 Extreme6. So why would I choose a lower-model board from the same vendor? And why would I pay an extra $20 for a product that, based on its specifications, should be inferior?

Read Customer Reviews of ASRock's X79 Extreme4


Well, there’s that award I gave the platform almost two years ago, for starters. My hope that its less expensive and better-equipped sibling was fixed through firmware was dashed last quarter. So now I'm hoping that this pricier version is as good as the sample I reviewed way back in 2012.

CPU Cooling: Thermaltake Water 2.0 Extreme

I’m once again relying on personal experience. Thermaltake’s Water 2.0 Extreme was the top-performing sample of our 2013 round-up series.

Read Customer Reviews of Thermaltake's Water 2.0 Extreme


I expect this closed-loop solution to seriously outperform my September build’s big heat sink, particularly matched up to the right kind of case.

Case: Fractal Design Define R4

You could argue that the best enclosure for Thermaltake’s Water 2.0 Extreme would be anything able to support its radiator up front, used as an intake. On the other hand, the right case for AMD's Radeon R9 290 might be something with extra noise dampening to muffle the fan that spins up to obnoxious levels under load.

Read Customer Reviews of Fractal Design's Define R4


Fortunately for me, Fractal Design’s Define R4 offers both traits. No wonder it earned our Recommended Buy award back in 2012.

4. Power Supply, SSD, Hard Drive, And Optical Drive

Power Supply: Seasonic M12II SS-850AM

By now, my reduced budget is starting to feel a little constrictive. But power is one place where I cannot cut corners. I estimated that I’d need at least 800 watts, and the unit I chose to reach that level in September is now too expensive. Thankfully, Seasonic was ready to take care of my needs.

Read Customer Reviews of Seasonic's M12II SS-850AM PSU


Semi-modular and manufactured by one of the industry’s most-reputable firms, the only feature Seasonic’s value-priced M12II-850AM forces me to give up is a small amount of efficiency. Coming in just shy of the Gold certification with an 84.52% full-load efficiency rating, that’s not much of a sacrifice.

SSD: SanDisk Ultra Plus SDSSDHP-256G-G25

Still shaving nickels to collect scrap metal, I turned to our Best SSDs For The Money column. First recognized for its performance and value in SanDisk Ultra Plus SSD Reviewed At 64, 128, And 256 GB, this drive remains atop our list of 256 GB models.

Read Customer Reviews of SanDisk's Ultra Plus 256 GB


Hard Drive: Western Digital Green WD20EZRX

A 200-300 GB SSD fills up quickly with programs and games, and most users still need additional space for all their movies, pictures, work files, and documents. September’s build had 3 TB of extra storage, but budget restrictions scaled this build back to 2 TB.

Read Customer Reviews of Western Digital's Green WD20EZRX


Apart from compatibility issues with a few older systems and some shipping damage complaints, Western Digital's 2 TB Green drive looks fairly good from Newegg's user reviews. It's not a top performer, but does serve up plenty of capacity quietly, and features 64 MB of cache to help it along.

Optical Drive: Pioneer BDR-208DBK

Short on cash but still full of dreams, I looked to previous System Builder Marathon stories to find a cheap, full-featured Blu-ray burner for redundant storage and media playback. The trick was to find an OEM drive with the software suite typically lacking from OEM packages.

Read Customer Reviews of Pioneer's BDR-208DBK


Unfortunately, I lost the bet on this one. Unlike its BDR-206DBK, Pioneer's BDR-208DBK really doesn’t include software. It’s still a good drive though, and I really couldn’t afford to pay any more than its low $60 price.

5. Radiator Installation

If you saunter over to page five of Quiet Gaming Cases, Part 2: Corsair, Fractal, And Gigabyte, you'll find a walk-through of the basic hardware installation in Fractal Design's Define R4, and our review of Thermaltake’s Water 2.0 Extreme shows how the CPU mates with the motherboard. Now we just need to figure out how to put that cooler into this case!

The Define R4 has dual-fan radiator mounts on both its top panel and front intake fan bracket. Indeed, front mounting gives us the coolest air for reduced CPU temperatures, even if that option requires removing the bottom and center hard drive cages. A second set of mounting holes lets us reposition the lower hard drive cage. But the upper cage can't be used once the lower one gets moved.

Fractal Design doesn’t mention this side bracket, which is designed to keep the cage from flexing front-to-rear. It’s not needed with the center cage taken out, and the bracket itself must be removed to make room for the front-mounted radiator.

One of the screws securing that bracket isn’t accessible from the fan filter door, so the entire front panel has to be pulled off. Thankfully, Fractal employs a snap-off design. After removing the four bottom screws and two front screws, the lower drive cage falls out.

A single top latch secures the Define R4's fan bracket, and side latches clip its fan in place. Sliding out the filter, we find screw holes.

This dual-fan radiator can be mounted at several heights, but only the lowest mounting holes allow its tanks to slide past case obstructions.

6. Finishing The Build

With the radiator installed onto the bracket and the hard drive cages removed, it’s time to pop the assembly back into the case.

I usually install heat sinks before securing the motherboard. However, my closed-loop system is already attached to the radiator, which we just mounted to our chassis. Although the CPU and memory modules could have been dropped in after getting the board into place, I added them first out of habit.

Designed for three-way SLI, the third PCIe x16 slot on ASRock’s X79 Extreme4 borrows eight lanes from the second slot. That’s still fast enough for every card we’ve tested, and enjoys the advantage of better spacing and, consequently, cooling. The only potential problem is that the hard drive cage I relocated overlaps the space consumed by these cards by less than 1/8”.

The cage is flexible enough that this installation would have worked, but it probably would have transferred noise from the card's fan through the rest of the enclosure. Rather than sticking a strip of foam between the two components, I decided to instead use the motherboard’s middle slot for my second Radeon R9 290.

The pump body is upside down because that's the cleanest way to run the hoses. My only concern was that the air bubble present in all closed-loop coolers might be recirculated. It should get trapped in the radiator, but the upside-down radiator doesn’t leave much room for the air bubble either. The best way to find out if the bubble is small enough to trap within the top of the radiator tank is to start the system and listen for noise that lasts more than a few minutes.

Another alternative would have been to mount the radiator on top of the chassis. That probably would have worked well in an exhaust configuration, especially with the blower-style coolers used on our graphics cards, since they push GPU heat out of our case. Mounting the radiator up top also provides more room for the cooler’s air bubble and would have firmed up the case’s somewhat-flexible roof.

7. Overclocking Through Firmware

The motherboard in my September build had trouble setting a fixed-ratio, fixed-voltage overclock, even a couple years ago when we reviewed it. Those issues were never completely addressed. The board I'm using in today's story doesn't have that problem. However, I chose to set a variable overclock anyway.

The main reason I'm going with a variable overclock is that it allows me to retain most of the CPU’s power-savings features at idle. The second reason is that, at my target maximum voltage, this CPU only runs at my target frequency with one or two cores subjected to a load. I set 4.6 GHz for one or two cores, 4.5 GHz for three or four, and 4.4 GHz for five- or six-core loads.

Chosen for its overclocking consistency, G.Skill’s DDR3-1866 once again reaches DDR3-2133 after switching from its default CAS 9 to CAS 10.

I decided to limit CPU core voltage to 1.30 V to increase processor longevity. To get there without disabling power-saving features, I used an offset-mode setting of +140 mV.

Memory voltage was the toughest variable to figure out, since my experience with these modules shows that they top out around 1.55 to 1.60 V. Stable through most of our benchmarks and all stress tests at stock voltage, an increase to the motherboard’s 1.585 V setting was needed to push it through one metric.

8. Final Touches

The best way I know of to maintain a high graphics core clock rate is to increase its power limit and fan speed. I was able to do this with the GPU set to 1100 MHz and the DRAM pushed to GDDR5-5600. Of course, you have to remember that AMD's PowerTune technology starts scaling back frequency once the Hawaii GPU hits 94 degrees and fan speed tops out at factory-defined levels.

AMD tries to keep the noise of its under-built thermal solution under control by letting its GPU get fairly hot, at which point the fan spins up to around 60% duty cycle. The fan only spins faster if it detects a thermal crisis. But those settings wouldn’t let these GPUs run near their limits.

I chose a maximum fan speed of 100% at a target maximum of 80° Celsius, even though AMD would have us believe that's overkill for a GPU that can run at maximum performance at higher temperatures.

Even though my CPU's temperatures remained low enough to prevent thermal throttling, the system would lose performance anyway after several minutes under full load. I opened the case and reached inside to find a scalding hot sink on the motherboard’s voltage regulator! Placing the left-over intake fan as a top-panel intake over the voltage regulator should have worked, but the blades whizzing past the fan grille sounded like a very loud hornet’s nest. Flipping the fan over to create exhaust helped cut the noise, but also reduced airflow over the voltage regulator. Fortunately, I kept a solution in reserve.

Voltage regulator sinks are typically designed to work in conjunction with a CPU fan. But liquid cooling moves those fans away from the processor interface. Even non-traditional gaming system makers like Lenovo recognize that problem, and have addressed it. Lacking a custom-fit solution or even so much as screw channels to secure a fan upon the X79 Extreme4’s PWM cooler, Antec’s universal SpotCool fits the build.

The case’s original intake fan remains mounted as exhaust above the motherboard, even though it’s no longer required. A little extra cooling never hurt.

As a result of my aggressive GPU fan settings, maximum system noise increased from 40.1 to 54.1 decibels at one meter after overclocking, illustrating Chris' notion that these cards either run too loud, or can't be cooled amply using the reference solution to keep them at maximum speed. At least the stock setup proves my advice about choosing the right case to muffle noise from the graphics cards.

9. Benchmarking Configurations

We're hoping that newer components and better overclocks will give the new system enough performance to overcome recent price spikes, yielding an overall value victory!

It's also worth reiterating that our benchmark suite is in the process of evolving. There are a lot of older tests still present today, but we're also running the new stuff for use next quarter.

Test Hardware Configurations
 Q4 $2400 PCQ3 $2550 PC
Processor
(Overclock)
Intel Core i7-4930K: 3.4 to 3.9 GHz, Six Physical Cores
O/C to 4.40-4.60 GHz, +140 mV
Intel Core i7-3930K: 3.2 to 3.8 GHz, Six Physical Cores
O/C to 4.20 GHz, 1.25 V
Graphics
(Overclock)
2 x Asus Radeon R9 290: 947 MHz GPU, GDDR5-5000
O/C to 1100 MHz, GDDR5-5600
3 x EVGA GeForce GTX 760: 980-1033 MHz GPU, GDDR5-6008 O/C to 1130 MHz GDDR5-6680
Memory
(Overclock)
16 GB G.Skill DDR3-1866 CAS 9-10-9-28
O/C to DDR3-2133 CL 9-10-11-10, 1.585 V
16 GB Mushkin DDR3-1600 CAS 9-9-9-24
Not Overclockable
Motherboard
(Overclock)
ASRock X79 Extreme4: LGA 2011, Intel X79 Express
Stock 100 MHz BCLK
ASRock X79 Extreme6: LGA 2011, Intel X79 Express
Stock 100 MHz BCLK
OpticalPioneer BDR-208DBK: 15x BD-RPioneer BDR-2208: 15x BD-R
CaseFractal Design Define R4 Black PearlLian Li PC-9NA
CPU CoolerThermaltake CLW0217 Water 2.0 ExtremeNoctua NH-D14 SE2011 
Hard DriveSanDisk SDSSDHP-256G-G2 256 GB, SATA 6Gb/s SSDMushkin Chronos Deluxe DX 240 GB, SATA 6Gb/s SSD
PowerSeasonic SS-850AM: 850 W Semi-Modular, 80 PLUS BronzeCorsair HX850: 850 W Modular, ATX12V v2.3, 80 PLUS Gold
Software
OSMicrosoft Windows 8 Pro x64Microsoft Windows 8 Pro x64
GraphicsAMD Catalyst 13.12Nvidia GeForce 326.80 Beta
ChipsetIntel INF 9.3.0.1026Intel INF 9.3.0.1026
Benchmark Configuration
3D Games
Battlefield 3Campaign Mode, "Going Hunting" 90-Second Fraps
Test Set 1: Medium Quality Defaults (No AA, 4x AF)
Test Set 2: Ultra Quality Defaults (4x AA, 16x AF)
F1 2012Steam Version, In-Game Test
Test Set 1: High Quality Preset, No AA
Test Set 2: Ultra Quality Preset, 8x AA
Elder Scrolls V: SkyrimUpdate 1.5.26, Celedon Aethirborn Level 6, 25-Second Fraps
Test Set 1: DX11, High Details No AA, 8x AF, FXAA enabled
Test Set 2: DX11, Ultra Details, 8x AA, 16x AF, FXAA enabled
Far Cry 3V. 1.04, DirectX 11, 50-Second Fraps "Amanaki Outpost"
Test Set 1: High Quality, No AA, Standard ATC., SSAO
Test Set 2: Ultra Quality, 4x MSAA, Enhanced ATC, HDAO
Adobe Creative Suite
Adobe After Effects CS6Version 11.0.0.378 x64: Create Video which includes 3 Streams, 210 Frames, Render Multiple Frames Simultaneosly
Adobe Photoshop CS6Version 13 x64: Filter 15.7 MB TIF Image: Radial Blur, Shape Blur, Median, Polar Coordinates
Adobe Premeire Pro CS6Version 6.0.0.0, 6.61 GB MXF Project to H.264 to H.264 Blu-ray, Output 1920x1080, Maximum Quality
Audio/Video Encoding
iTunesVersion 11.0.4.4 x64: Audio CD (Terminator II SE), 53 minutes, default AAC format 
LAME MP3Version 3.98.3: Audio CD "Terminator II SE", 53 min, convert WAV to MP3 audio format, Command: -b 160 --nores (160 Kb/s)
HandBrake CLIVersion: 0.99: Video from Canon Eos 7D (1920x1080, 25 FPS) 1 Minutes 22 Seconds
Audio: PCM-S16, 48,000 Hz, Two-Channel, to Video: AVC1 Audio: AAC (High Profile)
TotalCode Studio 2.5Version: 2.5.0.10677: MPEG-2 to H.264, MainConcept H.264/AVC Codec, 28 sec HDTV 1920x1080 (MPEG-2), Audio: MPEG-2 (44.1 kHz, Two-Channel, 16-Bit, 224 Kb/s), Codec: H.264 Pro, Mode: PAL 50i (25 FPS), Profile: H.264 BD HDMV
Productivity
ABBYY FineReaderVersion 10.0.102.95: Read PDF save to Doc, Source: Political Economy (J. Broadhurst 1842) 111 Pages
Adobe Acrobat XVersion 11.0.0.379: Print PDF from 115 Page PowerPoint, 128-bit RC4 Encryption
Autodesk 3ds Max 2013Version 15.0 x64: Space Flyby Mentalray, 248 Frames, 1440x1080
BlenderVersion: 2.67b, Cycles Engine, Syntax blender -b thg.blend -f 1, 1920x1080, 8x Anti-Aliasing, Render THG.blend frame 1
Visual Studio 2010Version 10.0, Compile Google Chrome, Scripted
File Compression
WinZipVersion 17.0 Pro: THG-Workload (1.3 GB) to ZIP, command line switches "-a -ez -p -r"
WinRARVersion 4.2: THG-Workload (1.3 GB) to RAR, command line switches "winrar a -r -m3"
7-ZipVersion 9.28: THG-Workload (1.3 GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=5"
Synthetic Benchmarks and Settings
3DMark 11Version: 1.0.3, Benchmark Only
PCMark 8Version: 1.0.0 x64, Full Test
SiSoftware Sandra 2013Version 2013.10.19.50, CPU Test = CPU Arithmetic / Cryptography, Memory Test = Bandwidth Benchmark
10. Results: 3DMark And PCMark

The new build posts big numbers in 3DMark, especially when we overclock it. September’s configuration sure appeared powerful at the time. However, lackluster overclocking held it back.

Beyond graphics performance, we also see that the newer Core i7 facilitates better stock and overclocked results in 3DMark's CPU-oriented Physics module.

The new machine’s cheaper SSD nearly matches the high-end part used last quarter, shaving a few dollars off of the build cost in the process. See? Sometimes value does win!

As with September’s build, I reverted to Microsoft’s generic AHCI driver in overclocking mode to boost storage performance beyond what Intel's workstation-oriented RSTe allows.

11. Results: SiSoftware Sandra

As we'd expect, the new machine's Ivy Bridge-based CPU is barely faster than the Sandy Bridge-E-based Core i7-3930K at stock settings. But it overclocks better. The difference is quite apparent in SiSoftware's Sandra Arithmetic and Cryptography metrics (the latter of which uses host processing power for the SHA-based hashing test).

The new system has faster RAM, but Sandra’s bandwidth scores still appear exceptionally high. Perhaps the newer CPU’s memory controller is just that much better? The bandwidth result affects our measure of encryption/decryption in the Cryptography test, too.

12. Results: Battlefield 3 And Far Cry 3

Capped at 200 FPS, Battlefield 3 finally starts to cut the new machine’s cards loose at 5760x1080. Stepping up to the Ultra-quality preset yields a more taxing load down at 4800x900. The big takeaway is that two R9 290s in CrossFire are overkill unless you're running a triple-screen configuration (or an expensive Ultra HD display).

Far Cry 3 appears to demonstrate a CPU bottleneck on the new system, even with its faster CPU. Greater processing overhead associated with AMD's graphics cards (something we've seen before) could be part of the problem. But we know how to circumvent this: ratchet up the detail settings, shifting more load onto the GPUs. Once we begin to lean hard on the graphics subsystem, the new machine fares a little better.

13. Results: F1 2012 And Skyrim

When I use pretty much any high-end combination of graphics cards, F1 2012 becomes either DRAM- or CPU-bottlenecked. Fortunately, the new system’s CPU and memory are both superior to last quarter's components. Unfortunately, though, I had to disable CrossFire to run the game at maximum performance.

Skyrim does really well on the new machine's higher-end processor and memory subsystems; scaling appears to correspond to CPU frequency. Again, two R9s are way more than you need in this title.

14. Results: Audio And Video Encoding

Single-threaded iTunes and LAME benefit from the Q4 machine’s faster CPU. The small advantage you see at stock settings comes from Ivy Bridge's minor improvements over Sandy Bridge. The larger jump between the overclocked configurations is attributable to our newer processor's higher frequency ceiling.

HandBrake and TotalCode Studio also respond appropriately to the new machine’s higher frequency, and they likewise also benefit from the Ivy Bridge architecture's subtle tweaks.

15. Results: Adobe Creative Suite

Adobe After Effects scales well enough on the new machine to expose a slight preference for its Ivy Bridge architecture, while Photoshop’s CPU-based filters scale upward only in proportion to its higher frequency. Photoshop’s OpenCL-based filters prefer the older machine’s Nvidia-based graphics cards though, and our results swing slightly in the opposite direction.

16. Results: Productivity

ABBYY's FineReader, Blender, and 3ds Max all benefit from the newer build's Ivy Bridge architecture, while Visual Studio's performance is more closely related to clock rate. Interestingly, though, if we threw a quad-core processor into the comparison charts, you'd also see that each one of these tests scales based on core count as well.

17. Results: File Compression

In contrast, our compression-based benchmarks tend not to scale quite as well based on available host processing resources. You typically get small increases in 7-Zip from a six-core CPU, but the other two titles don't benefit as much. We are using an older version of WinZip though, and our latest version of this benchmark (with WinZip 18 Pro) is more responsive.

But because both of these are systems with six IA cores, it's a little surprising that the newer machine scales better in 7-Zip compared to WinRAR. The gains could be coming from changes made to the Ivy Bridge architecture or better memory. I benefit either way.

Our OpenCL- and CPU-based WinZip benchmarks don’t benefit from the new machine’s advanced hardware, handicapping its higher clocks. It's only when we run the most intensive EZ script that I get the advantage I was looking for.

18. Power, Heat, And Efficiency

This quarter's machine enjoys several hardware-oriented advantages over its predecessor, and yet it barely consumes more power. Its 85%-efficient power supply requires 1000 W of input to produce an 850 W output.

With everything loaded simultaneously, the new machine barely stays under the rated capacity of its power supply.

Anyone worried about future upgrades needs to understand that we set up unrealistically-high loads to represent a worst-case scenario, and that Seasonic is famed for producing units with a moderate amount of over-capacity headroom.

The new machine’s stock configuration includes a 94° Celsius target GPU limit, and it sticks there regardless of ambient temperature. While that makes this single reading insignificant, all of the other configurations do allow max fan speed and are therefore variable with ambient conditions.

Notice how far the GPU temperature drops when I crank its fan up as high as it'll go. This would have been key to keeping the GPU at its full rated performance level, and was instrumental in achieving the highest possible performance from my overclock.

My new build only yields 4.6% more performance at its stock settings than its predecessor, while consuming 8.1% more energy. The most damning numbers came from a handful of games that weren't particularly kind to AMD's graphics cards, yet certainly reflected the power-hungry Hawaii GPU. At least its overclocked efficiency is an improvement over what we saw last quarter.

19. Value Conclusion

If you were wondering why all of my charts listed “original price” under the current build, that’s because I decided to break down performance-per-dollar twice: first using the prices we paid when we bought the components, and then with the current jacked-up prices. I’m absolutely certain that the original $2400 price of this machine smashes the value of last quarter's $2550 juggernaut. But would the same hold with R9 290s selling for more than $500 and the memory significantly more expensive as well?

With around 95% the price and 105% the performance, this quarter's machine originally set me up for a 10% value lead at its stock settings. The $300 price spike pushes its value rating to 1% below what I achieved three months ago. But my current effort also overclocks better. Consequently, I see a 30% value lead at November prices that drops to a 17% advantage using the figures we'd pay today.

If these were pure gaming machines, we wouldn’t need six-core CPUs. On the other hand, it is nice to see how much better the new system ranks at 5760x1080. AMD’s Radeon R9 290 really is a fast card, and two are even better. It’s a shame that a digital gold rush is taking these out of the hands of so many gamers.