Cooling a computer isn’t terribly complicated in and of itself. But if you want to do it right, selecting the best cooling fan can be a daunting task. It’s easy to throw money at Noctua, and we wouldn’t fault you for doing so, but that strategy gets very expensive very quickly. And not everyone digs the beige-and-brown theme – and the company flat-out refuses to do RGB.

Of course, you can opt to go the other easy route and just use what’s included with the case and cooler you bought. But that’s also rarely the optimal solution, and many premium cases come with no fans or just one for exhaust. Ideally, you want a fan that’s quiet, effective at moving air, attractive, and affordable. And since you’ve landed on this page, chances are you bought one of the cases that doesn’t come with enough fans, or you’re not happy with the included fans' performance or audio profile, and are looking for something better.

Recently, we’ve picked up fan testing again, and we’re working hard to test a bunch of different models to figure out the best options in this very crowded field. Over the past several months, we’ve worked in cooperation with industry experts to flesh out a benchmarking methodology, and we expect to update this page regularly with new models we’ve tested.

Our Testing methodology is built on real-world applications, and there are a handful of scenarios a fan needs to perform well in. Naturally, we’re testing airflow in an unrestricted environment as a ‘vanilla’ test, but this is often not indicative of performance inside a PC case.

What’s also not indicative of real-world performance is static pressure testing. Yes, it’s an industry standard, but it really only measures how much pressure a fan can generate when blowing into a closed tube, or up against a wall in a more real-world scenario. It does not account for the changes in acoustics when a fan is mounted on a radiator, heatsink, or behind a mesh intake.

Instead, we’ve decided to measure a fan’s acoustic performance not only when unrestricted and suspended mid-air, but also when strapped to a radiator and again when behind a mesh intake. The results from this paint a much more nuanced, real-world picture, often with interesting outcomes.

For more information about how we test PC fans, scroll to the bottom of this page.

Quick List

Best PC fans you can buy today

Best silence optimized fans

Noctua’s NF-A12x25 G2 is game-changing, offering the quietest, most refined fan we’ve ever seen. In our testing, it delivered whisper-quiet performance and managed an impressively huge amount of airflow despite running at very low RPMs.

Managing such big airflow despite not needing to spin fast to accomplish it means it’s also not a very turbulent fan, which translates to chart-topping performance when installed on a radiator or behind an intake mesh. In fact, it’s the only fan we’ve tested thus far that ran quieter on a radiator than when suspended mid-air without any load.

In addition to pulling off airflow miracles, the NF-A12x25 also offers a seriously advanced internal hub, featuring Noctua’s refined, tried-and-tested SSO2 bearing, a new 3-phase etaPERF motor, and a new PWM controller with SupraTorque functionality. This sounds like a bunch of marketing talk, but after testing the fan, we can tell you with confidence that it has zero audible mechanical noise.

Best budget silence optimized fans

The Arctic P12 Pro is an interesting fan, because it offers performance that is, as far as the charts go, nearly neck-and-neck with Noctua’s NF-A12x25. Except, the Arctic does something Noctua doesn’t: It only costs $8.50 for one fan.

The price you pay for the Arctic fan is, in part at least, in its level of refinement. It uses a plastic ring to strengthen the impeller. And, in order to achieve its airflow, it relies on spinning up to higher speeds. Consequently, it’s a more turbulent fan than the Noctua, so it takes a bit more of a hit when mounted close to a mesh intake. It also produces a few mechanical noises, but even so, it’s a highly capable fan.

If, however, you just want a fan to move lots of air quietly, and you don’t want it to cost much, the Arctic P12 Pro is in a league of its own.

Best RGB fans

It feels a bit weird to place essentially the same Arctic fan on the Best list twice, but it truly is that good. The P12 Pro RGB is, for all intents and purposes, the exact same fan as the non-RGB variant, except with a white translucent impeller so that the RGB lighting from the hub shines through.

It performs on par with the non-RGB variant in our testing, offering wicked performance at up to 3000 RPM and a fantastic noise-to-airflow ratio. The best part: it barely costs more than the non-RGB variant.

Best High-Performance fans

The Phanteks T30 (review here) is one of the fans that somewhat copied the NF-A12x25 – it featured a similar blade shape, and also used a LCP (liquid crystal polymer) plastic for the impeller. Except, it wasn’t quite a copy, and now, even half a decade later, it’s still king of the hill in the ultra-high-performance fan space.

Why is it so good? Phanteks cheated. I wish I could say they didn’t, but they did. The industry standard for a 120mm fan is 25mm thickness. So what did Phanteks do? They made a 30mm fan. In fact, Phanteks was so proud of this fan that when they sent it out for review, they put together a wind tunnel kit for us because they wanted everybody to know Phanteks meant business and had defeated Noctua at their own game. And they did – it performed notably better than Noctua’s finest when tested for noise-to-airflow.

Moreover, the good stuff doesn’t stop at just 5mm of extra thickness. The company also built in a mode selector at the back of the hub that lets you select quiet, regular, and advanced modes. If you ask me, though, advanced should be called Final Destination mode it lets spin all the way up to 3,000 RPM. With its added thickness, high speed, and LCP material, it’ll gobble air and your fingers if you’re not careful.

Choosing the best fan for you

Fans. They spin. They move air. They make noise. Sometimes they have lights, and more recently, some have LCD screens. It may seem like a silly thing to obsess over, but picking the right, or wrong, fan can make a major difference in a PC build.

Not everybody looks for the same thing in a fan, and whereas picking the Best CPU or Best GPU is often a matter of what’s best-performing for the dollar in your allocated budget, fans don’t evolve at nearly the same pace, so picking one is much more of a subjective exercise. With that said, depending on your situation, there can be a few key considerations that can make the choice easier.

• If you’re someone who comes home from work, whips on a headset, and plays games for an hour, noise levels or MTBF (durability) figures may not matter much to you.
• Alternatively, if you’re someone who works on your PC in silence all day long, it may be worth investing in more refined fans that offer less wind and mechanical noise, with a more durable motor.
• Personal sensitivity to certain noise types and frequencies also plays a role.
• Whether the PC is placed on the desk or on the floor is another factor to consider. Fan noise can be less noticeable from the floor.
• Some fans are designed to perform better on radiators and heatsinks, whereas others perform better as case intake or exhaust. Pick accordingly.
• Fans with daisy-chaining connectors can ease installation and provide tidier cable management.
• Cases with tinted windows require fans with brighter RGB for a visual spectacle.
• Mini-ITX PCs, though fantastically small, can benefit more from a fan upgrade to quiet them down than big full-tower E-ATX machines.

More about how we test PC fans

Our PC fan testing consists of multiple phases. First, we test a fan’s vanilla acoustics by suspending it mid-air with four pieces of string so that no frames or holders interfere with the airflow. We start with a basic measurement, noting whether the fan breaks through the noise floor at its minimum speed, then we record the RPM for every whole dB between 34 and 44 dB(A). Finally, we record the maximum RPM and its associated noise level.

We then repeat these tests with a radiator mounted behind the fan, and again with a mesh intake in front of the fan. We do this because radiators and intake mesh fundamentally change a fan’s acoustics and the RPM they must spin at to achieve the desired noise targets.

After the acoustic data, we mount the fan in our wind tunnel to collect airflow data.

The windspeed is measured with an anemometer. However, although the fan’s impellers may push air, no air comes from the space behind the motor hub, and fans don’t generally push air in only one direction. In fact, the space immediately behind a motor hub usually becomes a low-pressure pocket, which is why we need a wind tunnel – to form the airstream into one cohesive, non-turbulent column of moving air.

Inside the wind tunnel (at the halfway point and near the end) are two air-straighteners, which together with the length of the tunnel at 77cm (just over 30 inches) neatly combs the airstream into something we can measure, without creating tons of backpressure that would affect the readings.

Using the RPM datapoints from the acoustic testing phase, we now set the fan to each RPM point and record the associated wind speed. We repeat this with a radiator behind the fan, and again without the radiator but with the mesh intake in front of the fan. Of course, the wind speed isn’t actually a useful measure (yet) of how much air a fan can displace, until we convert it to meters cubed per hour (m³/h).This testing yields up to 84 datapoints per fan (assuming the fan can achieve each point), which tells us enough about a fan's capabilities, quirks, and acoustics to judge whether it's one of the best PC fans you can buy.

Not long ago, we tested two popular case fans, pitting Noctua’s NF-A12x25 G2, an industry legend, against the also well-liked Arctic P12 Pro. We had some surprising results, as the Arctic fan almost kept up with the much pricier Noctua. However, Arctic also sent us some P12 Pro’s of the RGB variety, so we thought it would be fitting to put Artic’s colorful spinner up against some competitors in its own price category.

In this round of testing we’ve rounded up five 120mm RGB fans that don’t cost a whole lot, so here, you won’t be seeing Lian Li’s fancy LCD-infused, wireless, infinity-mirror fans. Instead, we’re pinning the P12 Pro up against Be Quiet’s Light Wings LX, Fractal Design’s Aspect 12 RGB, Phanteks’ M25 G2 RGB, and finally, Levelplay’s Combat Fan 120 A-RGB. This last one was sent to us as a surprise, and I thought it was fitting to include in the roundup.

Let’s dive straight into the testing.

Testing Methodology

For evaluating PC fans, there are two main terms you should be aware of: Airflow and Static Pressure.

Airflow is easy – how much air does a fan move (through a wind tunnel in our tests) when not burdened by a radiator, mesh, or any other restriction.

Static pressure can be measured by placing the fan on a wind tunnel, closing the end, and measuring the pressure generated with a sensor, which spits out a value in millimeters of water. The higher the number, the better a fan is likely to be at overcoming restrictions.

However, I find that a fan’s measure of static pressure isn’t actually all that useful – all it tells you, really, is how well a fan can push air up against a wall, which is obviously not how you want to position the back of your PC. Instead, I prefer to test for noise and airflow through a radiator and intake mesh, as well as the standard suite of unrestricted tests for a more nuanced, real-world picture.

Testing Results of 5 Budget RGB Fans

(Not) Breaking the Noise Floor

On the day of this testing, the noise floor in my office sat between 32.4 and 32.7 dBA, and I am happy to be able to write that all five fans, when run at their minimum speed, were completely inaudible and did not break the noise floor, even with the dB meter placed at just 20 cm from their hubs.

This used to be something I only saw in premium fans, where manufacturers have designed quiet hubs, so I was delighted to see this kind of performance from budget-friendly RGB fans.

Airflow is not recorded for these tests, as none offer enough air speed for the anemometer to register.

dBA VS. Airflow: Unrestricted

The first series of tests I subject the fans to is measuring their acoustics in an unrestricted environment. I target every whole dB between 34 and 44 dB(A), and record the associated RPM. Then, I put that same fan on our wind tunnel, and record the windspeed at said RPM figures. This is then calculated over to cubic meters per hour for airflow.

In this test, Arctic’s P12 Pro leads the pack by a significant margin, followed by Be Quiet’s Light Wings LX. Phanteks’ M25 G2 RGB only performs a smidge worse than Be quiet’s fan.

Note the dip in the curve of Fractal Design’s Aspect 12 RGB at 37 dB(A). This fan creates quite an annoying resonance at certain speeds, where it gets louder at a certain RPM, and then quiets down again once past this RPM. I suspect that if I were to test at even smaller increments, the graph would become quite wobbly.

Most fans exhibit this kind of behavior to some extent, where they create unwanted noises at certain RPMs – even Arctic’s fan does this, but the level to which the Aspect 12 RGB made this noise was really quite bothersome.

Similarly, for the Phanteks’ fan (while its curve looks alright), I was unable to test at 35 dBA, which is why its line starts at 36. No matter how much I fidgeted with its speed, I simply could not get the dB meter to hit 35 – it was either over, or under, but not on.

The Levelplay Combat fan had the same issue, but at 41 dB(A), where I just couldn’t hit the mark.

All of this is not to say that Arctic’s fan is without imperfections either, though. This sample exhibited a light high pitch at certain RPMs which didn’t affect the readings, but I did notice, and at speed it started wobbling – a sign that it is unbalanced.

The predicament: Whether to test with more fans

I’ve wondered throughout testing these fans whether I should test with more fans and take the average results, but I have two problems with this: firstly, testing this thoroughly already takes quite some time, and more importantly, averaging out the results would mask defects.

In a premium fan from Noctua, we expect near-perfection. However, these are RGB fans that cost no more than $15 each. There will be some variance in issues between samples, but not averaging out the results means you get to see the fan’s performance for what it really is with one specific fan. Some of these behaviors will be down to their design and consistent across samples, while other issues may differ between units.

I do request three samples for each fan to review, however. If I do run into any issue like those mentioned above, I grab the other samples and quickly run them from lowest to max RPM and judge the results. If they exhibit similar behavior, I roll with the data, but if they to suddenly sound much cleaner, then I know there’s a manufacturing defect at play.

dBA VS. Airflow: Radiator Restricted

Pin these fans on a radiator, and something very interesting happens. Four of the fans perform nearly the same as each other, with the Arctic P12 Pro standing out by a huge margin.

Note that for this test, the acoustics are measured separately from the acoustics without the radiator, and the fans follow similar noise-to-RPM patterns.

dBA VS. Airflow: Mesh + Filter Intake Restricted

When placing the fans behind a mesh intake, things again get interesting. This is the most challenging test of the three, as placing intake mesh in front of a fan fundamentally changes the way they sound – intakes punish airflow more than radiators, not because they’re more restrictive, but because placing something in front of a fan has a much more profound effect on the noise than something behind it, and this is where a fan’s blade design really starts to shine.

The Arctic P12 Pro continues to lead the race here, with Be Qquiet’s Light Wings LX following behind. The Combat fan, despite being from a lesser-known brand, does OK for itself.

However, Fractal Design’s Aspect 12 RGB and Phanteks’ M25 G2 put down appalling performances on this test, offering only 60% of the P12 Pro’s airflow at the same noise levels.

Maximum performance

Finally, the fans get a test at full blast.

This isn’t a useful test to compare noise-to-performance efficiency of the fans, but if you’re looking for a fan with low maximum speed to keep noise levels down without having to make a custom fan curve, Levelplay’s Combat Fan is your best pick. On the other hand, if you’re looking for maximum peak performance, Arctic’s P12 Pro RGB is a beast in a league of its own that can spin at well above 3000 RPM.

The dBA figure doesn’t say everything

Just because a fan performs well on the charts doesn’t necessarily mean it may be pleasant to listen to. One example of this is the Combat Fan from Levelplay. It’s not a terrible performer, but at certain speeds, something starts humming at an annoying mid-frequency. It was one of those frequencies that echoes through the room, and if I weren’t testing it but had it in my PC, it would drive me mad, wondering what the source of the noise in the room is.

The Fractal Design Aspect 12 RGB also suffers from a painfully dirty noise profile that did show up on the charts, as it limited the fan’s ability to reach the speeds it otherwise would have had it not made these noises. It sounds a bit raspy, almost like there’s grit in the motor, and it had similar resonant noises as the Combat Fan.

The M25 G2 didn’t have weird noises – it was just inefficient in its noise-to-airflow ratio.

Judging the RGB, feature set and pricing

Fractal Design Aspect 12 RGB

Fractal Design’s Aspect 12 RGB fans have the worst RGB of the lot. The blades are almost opaque, which makes the RGB on anything but the hub rather dim compared to the competition. It also does not have a zero-RPM mode – when fed a 0% PWM duty, it will still spin, and with a minimum speed of about 500 RPM; I consider this a problem. Yes, it’s quiet at its minimum speed, so that’s not an issue, but high minimum speeds combined with no Zero-RPM implementation can lead to excessive dust buildup, as the constantly whirring blades accomplish nothing but to pull small particles towards your filter and / or fan.

This fan is available for $51.47 as a 3-pack, which brings the per-unit cost to $17.16.

Combat Fan 120 A-RGB

Levelplay’s Combat Fan 120 A-RGB comes with a ring around its blades, which makes for a cool look. Despite the RGB coming from the hub, it still lights up the ring moderately. It also responds correctly to a Zero-RPM signal.

Pricing for the Levelplay Combat fan sits at $12.23 per unit.

Be Quiet! Light Wings LX

Be Quiet’s Light Wings LX fans feature good RGB and a slick blade design, but no Zero-RPM feature. That being said, the minimum RPM is extremely low, tested at just 213 RPM, so I can forgive the lack of a zero-RPM mode.

Pricing sits at $47.90 for a 3-pack, bringing the per-unit price to $15.97.

Phanteks M25 G2 RGB

Phanteks M25 G2 RGB, although not being the quietest fan, offers a properly complete feature set that makes a compelling case for forgiving its performance shortcomings. The RGB is gorgeous and it comes with an infinity-mirror hub. The fans can also clip on to each other, daisy-chaining in a not-quite-wireless fashion but certainly better than the competition. The fan even comes with plastic covers for the screw holes for a tidy appearance. It also responds correctly to a 0% PWM duty signal.

Pricing sits at $36.99 for a 3-pack, which is a surprisingly good price if you consider the amount of effort that went into their design. It may not be the quietest fan, but it sure is the prettiest and comes with most bits and pieces to complete the set.

Arctic P12 Pro RGB

Arctic’s P12 Pro lights up bright and uniform, offering a fantastic visual spectacle. It also comes with a proper Zero-RPM implementation, but that’s where its feature set ends.

This fan costs just $28.49 in a 3-pack, which is a mighty good deal that’s hard to pass up at $9.50 per fan.

Picking a winner

It would come as no surprise if I said that Arctic’s P12 Pro was the winner of this test, but I’m not sure about that. The P12 Pro RGB is clearly the best performer, but it was a tad unbalanced and had some light high-pithed whining. It also clicked a few times upon startup, but most of the fans did that. Mechanical noises.

If you’re looking for the best value offering, the Arctic P12 Pro RGB is hard to pass up with the 3-pack costing just $9.50 per fan.

The Light Wings LX, although being the second-place fan in testing for noise levels, is my favorite of the bunch. Why? It’s clean! It makes a crisp, clean sound ---just a whoosh, with no extraneous noises on top. It’s a slightly louder whoosh, but a clean whoosh. Of course, that’s not to say it may be your favorite. Personally, I’m sensitive to these kinds of noises, and I’m willing to pay a little extra for a clean-sounding PC, but that little bit extra is quite a bit if you tally it up over multiple spinners.

Costing $15.97, this puts the be quiet! Light Wings LX right on the edge between budget and mid-range RGB fans.

If, on the other hand, you’re someone who doesn’t mind noise too much, and wants to build something that you’re truly proud to show off, Phanteks’ M25 G2 fans do away with the focus on silence, and prioritize looks with bright RGB, an infinity mirror, seamless daisy-chaining, and even a mechanism that clips the spinners into one unit and covers for the screw holes.

I’m someone that works many hours a day in silence, with my open air PC on my desk near me, so I’ll prioritize silence any day of the week. But if you’re a gamer, keep the PC on the ground, or wear a headset all the time anyway, the quality of the RGB may be far more important.

If picking a winner is so difficult, should we pick a loser?

If there’s one clear conclusion I can draw from this batch of testing, it’s that Fractal Design’s Aspect 12 RGB is on my do-not-buy list. It offers no real redeeming features to make up for its shortcomings – it produces a dirty, inconsistent noise, performs disappointingly compared to the competition, has no Zero-RPM mode, and even the RGB doesn’t look all that great. It doesn’t even come with vibration-damping rubber pieces on the corners.

I honestly don’t understand what it’s doing in Fractal Design’s product portfolio at all, other than making the company’s other fans look good. And the wildest thing is, it’s also the most expensive fan tested here, at $17.16 a piece.

In this respect, fans are much like tires for cars: they come in all shapes and sizes, each with different purposes, from different manufacturers, and they all seem to do the same thing. And much like tires, in the world of fans there are a few great ones, many good ones, and some to avoid entirely.

PC components manufacturer Phanteks showcased one of the most striking PC cases at CES this year. At first glance, the new Exo 626s looks like an open-frame case; however, it comes with dedicated chambers for the PSU, motherboard, graphics card, and a 360mm AIO liquid cooler for the CPU.

The idea behind this case is that with each component having a separate chamber, airflow is easier to manage, and heat from one part doesn’t affect the others. From what we gathered on our visit in Vegas, the GPU chamber can fit large-sized graphics cards going as big as 4-slots. It comes with an RGB LED strip and a mesh finish at the top and the bottom, allowing the card to breathe easily. The PSU chamber at the bottom features a similar RGB strip along with some I/O ports at the front, including two USB Type-A ports, a USB Type-C, and a 3.5mm audio jack.

The AIO chamber also features a mesh finish at the front and is positioned next to the motherboard with a provision to route the water cooling tubes directly into the motherboard chamber. Lastly, the motherboard chamber comes with a glass panel at the front, while the back panel gives access to the AIO radiator and the rear side of the motherboard with a provision to mount a 120mm fan for additional cooling.

Unfortunately, the case is available as part of a pre-built only, as it is made in partnership with CyberPowerPC, with no confirmation on a standalone retail version at the moment. The pre-built will be offered with an Intel Core Ultra 285K CPU, an RTX 5080 GPU, an Asus TUF Z890 motherboard, 32GB of DDR5 dual-channel memory clocked at 6400 MHz, a 2TB WD Black SN8100 PCIe Gen 5 SSD, a Bitspower 360mm AIO liquid cooler with ARGB lighting, and a Corsair RM1000X 1000W 80+ Gold power supply unit. As for the pricing, it will be retailing at $3,500.

While the Exo 626s looks impressive, locking such an innovative case behind an expensive pre-built feels like a missed opportunity. This is exactly the kind of uncanny airflow-focused design PC enthusiasts would love to experiment with in their own custom builds. Hopefully, Phanteks takes note of the interest and considers releasing the Exo 626s as a standalone case for consumers and not just system integrators.

We're now on Day Two of our Computex 2024 coverage, and the news isn't showing any signs of slowing down. Today, we have some insights from Intel on its next-generation Xeon 6 processors, which will be available in low-power/efficient and high-performance variants to take on AMD's potent EPYC family. There are new power supplies from Seasonic and Super Flower (including a 2800W monster) and a massive new 57-inch Dual 4K monitor from Cooler Master. 

Intel's Xeon 6 processors are on deck to challenge AMD's EPYC family Earlier this week, AMD announced its new 3nm Turin-based EPYC processors, which are available in up to 192-core configurations. Today, however, is Intel's turn to shine with its Xeon 6 processors. The Xeon 6 family will be split into families, with the first being "Sierra Forest," featuring efficiency-optimized E-Cores with up to 144 cores are available now (288-core models will debut in 2025). The second family, "Granite Rapids," will feature up to 86 P-cores and launch during the third quarter of 2024 (up to 128 P-core SKUs arrive in 2025).

• Intel Launches 144-core' Sierra Forrest' Xeon 6 CPUs, Granite Rapids Follows in Q3

New Power Supplies from Super Flower and Seasonic
In the world of power supplies, Seasonic and Noctua have collaborated to launch the Prime TX-1600 Noctua Edition. This 1,600W power supply has been infused with a 120mm Noctua NF-A12x25 fan that spins at up to 2,000 RPM, an optimized grill fan, and includes 12V-2x6 connectors. It also comes in a brown finish that is indicative of the Noctua influence.

But if you need even more power, look no further than the Super Flower Leadex SF-2800F14HP 3.1. This PSU is rated for a staggering 2800W and supports up to four GeForce RTX 4090 graphics cards using 12V-2x6 connectors. Despite the insane power rating, Super Flower says the Leadex SF-2800F14HP 3.1 only supports a single PC.

• Seasonic's latest big and silent 1,600W PSU gets a quieter 120mm Noctua fan upgrade
• 2800W Super Flower PSU boasts four 12V-2x6 power connectors — good for quad GPU setups used for AI training, not gaming

CAMM2 Memory Standard Backed by Kingston and MSI
Both Kingston and MSI are embracing the CAMM2 standard for desktop motherboards. The new space-saving standard is aimed primarily at laptops as a lower-profile alternative to SO-DIMMs (they are 57 percent thinner). However, it will also have a place in the desktop world, as witnessed by Kingston's Kingston Fury Impact DDR5 CAMM2 modules. They first will be available in 32GB and 64GB capacities, with 128GB and 256GB capacities arriving at a later date.

MSI has already announced a desktop motherboard supporting CAMM2 modules, which left many people scratching their heads. Now we see why the company opted for this route: the space-efficient design of CAMM2 allowed it to design a special waterblock to cool the memory directly.

• Kingston unveils new CAMM2 RAM modules at Computex — bold new RAM form factor comes to PC on MSI and Asus boards
• A curious use case for CAMM2 memory modules on desktops: Liquid cooling

Phanteks and InWin Have New PC Cases on Display
If you want to upgrade your gaming rig's enclosure, we have two intriguing options. First is the Phanteks Evolv X2, which has glass panels on three sides, which gives you a nice wraparound look at the hardware lurking inside. Power connectors are hidden, the fans are recessed, and a secret channel hides all those pesky cables. The result is an incredibly clean design that also aids in airflow. Despite the good looks and trick design features, Phanteks says the case will retail for just $150.      

For something a bit more unorthodox, there's the Infinite Signature case from InWin. It features a large 180-degree curved glass panel. The case is also motorized, allowing it to rotate 45 degrees to give you a better look at the internal components.

For something more mainstream, there's the F3 microATX case that can be fitted with plastic, wood, or steel panels to give it your unique and personal touch. Pricing starts at $149 and goes up to $160 if you want wood paneling.

• Phanteks Evolv X2 is the perfect showcase PC chassis — floating motherboard tray and recessed fans a delight
• InWin flexes its glass-bending skills with new Signature case — Infinite and F3 revealed at Computex 2024

ASRock and Cooler Master Showcase Exciting New Gaming Monitors
Computex has always been a big draw for gaming monitors; this year is no exception. ASRock drew in eyeballs with its new Phantom Gaming PG032UFS2B, which features a 32-inch 4K OLED panel with a 240 Hz refresh rate. However, if you dial the resolution back to 1080p, the maximum refresh rate jumps to 480 Hz. 

If 480 Hz isn't fast enough for you, there's the PG27FFX2A, which is claimed to be the world's fastest IPS panel, with a refresh rate of 520 Hz (at 1080p).

For those who want an absolutely massive monitor to dominate your desk and give you an immersive gaming experience, there's the Cooler Master GP57ZS. While it's still a prototype, the GP57ZS is a Dual-4K (7,680 x 2,160) resolution monitor boasting an insane 57-inch diagonal. With a 1000R curve and 2,304 Mini-LED backlit zones, we're eager to get our hands on a future production sample.

• ASRock's 32-inch, 480 Hz OLED monitor is a feast for the eyes and your router
• Cooler Master showcases a massive 57-inch super ultrawide dual-4K Mini LED gaming monitor

The Best of the Rest

• ASRock shows off new AMD Ryzen 9000 motherboards
• Ducky One X keyboard first to use Cherry's innovative induction switches
• MSI launches AI+ laptops for gamers and creators, boasting 100 to 542 TOPS

Phanteks put on a dazzling show at Computex 2024, showing off a slate of new hardware headlined by the Evolv X2. The Evolv X2 is a triumphant return to the Evolv line, a perennial favorite of Tom's Hardware. 

With its three glass sides, curved metallic top and bottom, recessed fan areas, and floating motherboard tray, the Evolv X2 is a feast for the eyes. A look around the back reveals no secrets, with the motherboard cables tucked perfectly into inset cable channels. Add to this the trademark height of an Evolv case and options for gold accents, we thought this case would cost north of $250. But shockingly, Phanteks targets $150 for the Evolv X2's MSRP. 

The interior of the showroom PC looked clean as a whistle, thanks to a mix of industry trends and clever design. The X2 supports the new wave of back-connecting motherboards (BTF, Project Zero, Stealth etc), where all I/O and power connections to the motherboard are hidden away on the rear of the PCB. The motherboard rests a few inches deeper than the rear panel, creating a secret rear channel for hiding away cable clutter. And of course, it wouldn't be a Phanteks case without some serious RGB integration.

If thermals are your concern, Phanteks did some good work in accommodating a solid airflow stream. With higher than average ground clearance, three bottom intake fans can pull good air in to be expelled by the 4 exhaust fans. While enthusiast hardware may get hot, a mainstream system should perform fine. Still, we'll have to put the case through our testing to say for sure.

If you're looking for a case to showcase your build without spending $200 or more, the Evolv X2 looks to be a tough case to beat. Its innovative styling and floating motherboard tray will likely make this case a mainstay of the midrange market upon its release. We are basing our coverage off of the prototype model, but Phanteks expects to have the case ready for launch later this year.

Whether it's going for an all-white build or just having a really nice case to build in, this excellent budget case from Phanteks is a steal at this sale price. However, it does require mailing in a rebate to claim this offer, which does offer a little hassle. The Phanteks Eclipse G360A on sale is also only available with the white color option.

One of our favorite cases that we've tested and an Editor's Choice winner, the Phanteks Eclipse G360A is on sale for just $59 at Newegg after redeeming a $20 mail-in-rebate. See our review of the Phanteks Eclipse G360A for more detailed information and benchmark testing results.

It is a mid-tower ATX case with plenty of room for all of the latest components and is spacious enough to fit radiators measuring up to 360mm in either the front or top of the case. With the ability for such cooling potential, this budget case is capable of housing some high-end components. In testing, we found the Phanteks Eclipse G360A had excellent thermal performance thanks to the mesh front panel enabling good airflow. Still, the front panel also didn't have the best filtering, so things could get a little dusty over time if you are operating in a less-than-ideal environment. 

There's an easy-to-access front panel on the top of the case, which includes 2 x USB 3.0 ports (No USB-C), a mic and headphone port, power controls, and a button for switching lighting modes of the included D-RGB fans and D-RGB LED strip. So there are plenty of options for configuring the lighting setup for your rig. 

Although this case is roomy and easy to build in, one negative that we found with our review unit was that the design of the cable routing wasn't the greatest, and if you have shorter cables on your PSU, some routing might get a little tight. Overall, though, this is a fantastic case at its original MSRP price, so being able to pick the Phanteks Eclipse G360A for just $59 makes it a no-brainer if you're looking to build a good-looking PC on a budget. 

Phanteks just launched a 5.5-inch LED with an IPS display to be used inside your case. This is an excellent option for those who want to showcase more than just their computer's RGB lighting hardware, liquid cooling setups, and fancy cables to make your edge-to-edge fish tank-looking glass PC cases look less boring while complimenting its aesthetics.

These have a 2160 x 1,440 resolution, 60 Hz refresh rate, 450 nits brightness, and 99% sRGB color profiles. The casing is plastic with two color options and is lightweight (Phanteks only lists the gross weight, so it should be much less than 337.5 grams).

What Makes This More User-Friendly Than Any Mini Displays

Many PC enthusiasts have been incorporating mini displays inside their PC cases to display a particular video or as a system health monitor, even before all glass panel chassis were a thing. Their multiple mounting options with two viewing angles and DC power input make these mini displays more convenient for this purpose.

The monitor has magnets placed within the monitor's casing, allowing you to have an almost flush mount on any steel surface (like the PSU Shroud). But the screw-mount feet allow more freedom. It provides a 90-degree or 15-degree tilt, and it is also magnetic. Additionally, because of the mounting holes on the display and the in-display magnets, you can display it horizontally or vertically. If this doesn't satisfy you, the feet provide screw holes, letting you attach this monitor to a 120mm fan from within. The mounting method is limited to horizontal orientation when mounted on top/ bottom mounted fans and the vertical layout when mounted from within the front and rear case fans.

Since it uses a 5v mini USB connector for power, it provides cleaner cable management than typical mini displays with a power brick powered by a wall socket. A SATA or Micro-USB cable lets your system's PSU power up this display internally. All the adequately sized wires are provided, though you'd need to buy longer cables should you use this as an external monitor away from the case.

Phantek's purpose-made LED has a recessed cable management area for the HDMI and a SATA-to-mini USB connector. The monitor's OSD lets you change the display orientation, set resolution, set the display to extend, hide the taskbar, and prevent cursor display on this screen, something you wouldn't find in typical mini displays.

Phanteks provides all the needed screws, HDMI cable, and a SATA-to-mini-USB cable, giving you a complete package. With a two-year warranty and the option to choose in white or black casing, you get everything you need to have a mini display inside your PC case. Pre-orders can be placed on Phantek's website, which sells these for $109 each. But once this is available, we may see these listed on Amazon and other online retailers.

Lian Li has filed a patent suit in California requesting action against Phanteks (h/t OC3D). The Taiwanese aluminum case stalwart alleges that its Dutch rival has willfully infringed upon its patented fan design innovations. Specifically, Lian Li indicates that Phanteks D30 series fans infringe upon US patent 10,690,336 B1, which was awarded to Lian Li in June 2020. The patent describes a method of daisy chaining RGB fans using a simple, compact connector.

The UNI range of fans from Lian Li has been quite successful. Part of this product line’s appeal may be due to its “revolutionary” interlocking modular connectors, making them “cableless” and great for daisy chaining in systems while minimizing cable clutter.

Perhaps Phanteks was ‘inspired’ by the Lian Li fans, or at least the idea behind them. However, it is hard to believe the idea of minimizing cabling while maximizing RGB and airflow in a system is unique to Lian Li. Also, perhaps the Phanteks D30 series is too similar to Lian Li’s designs to have been truly independently originated. These and similar questions will probably be central to the upcoming deliberations in the US District Court for the Central District of California.

The court case was filed on September 8, 2023, and reportedly comes about three months after Lian Li sent a letter to Phanteks about what it thought were patent-infringing products. Phanteks has continued to market and sell its D30 fans since May, and it seems like a lack of acknowledgment of Lian Li’s claims has precipitated this lawsuit.

What Lian Li hopes to achieve with its legal action is a judgment confirming that Phanteks has infringed its US patent. If successful, Lian Li will seek damages to date, plus a sales injunction on infringing products sold without a license.

We aren’t patent experts, so it is hard to know whether Lian Li’s court action has a great chance of success. The overall idea shared by the rival fan daisy chain designs seems to be similar, but the execution might be different enough to save Phanteks. If Lian Li is successful here, it might also trigger further litigation involving products like the Thermaltake SWAFAN EX and Corsair iCUE Link systems.

From the floor at Computex, Phanteks has announced the launch of two new chassis to its NV line, the full tower NV9 and the mid-tower NV5. It has also announced the Glacier One D30 AiO cooler and D30 140 mm fans.

Phanteks NV9: Bigger and Better

The NV9 is a full tower chassis (276 x 572 x 607 mm larger than the NV7) with an MSRP of $299 and an October 2023 release date. With "near-seamless tempered glass panels" to the front and one side of the case, we get a clear view of the hardware inside. With support for up to three 420 radiators and a 480 radiator on the side, your rig will have ample cooling options. Talking of radiators, a modular side bracket enables easy installation and maintenance of your radiators.

The NV9 has improved support for E-ATX motherboards up to 280 mm and a simplified storage are affording more space and flexibility. Two additional accessories are GPU brackets (one a premium model) to mount your GPU away from the motherboard.

• Support for fans up to 11 x 140 mm or 12 x 120mm
• Radiators: Top 420/360, Side 420/480, Bottom 420/360, Rear 280/240
• Storage: 3 x 3.5 inch hard drive, 2 x 2.5 inch SSD or 6 x 2.5 inch SSD total
• Dimensions: 276 x 572 x 607 mm
• Colors: Satin Black / Matte White
• Availability: October 2023
• MSRP: $299

Phanteks NV5: The New Mid-Tower

Just like the NV9, the NV5 has a near-seamless glass view on two sides of the case, offering a view of your motherboard and RGB. The NV5 has support for eight 120 mm fans and at the top there is space for a 360 radiator, as well as space to the side. A bottom fan provides cool air for your GPU, and Phanteks claims that the NV5 has a "spacious interior to accommodate modern GPUs". A PSU cover provides easy cable management, and DRGB provides the subtle (or not) lighting that we all crave.

The NV5 is for enthusiasts who like the design of the NV7, but want more of their desks back, less fans or more money in their pocket. The chassis features support for ATX motherboards, but not E-ATX due to a GPU support bracket. An anti-sag GPU bracket is included in the kit. 

Cooling can be via up to eight 120mm fans, or your choice of 360 or 120 radiators. Storage is up to three 3.5 inch drives, or four 2.5 inch SSDs.

Phanteks NV5 is slated for an August 2023 release, with an MSRP of $129.

• Support for eight fans up to 120 mm in size
• Radiators: Top 360, Side 360, Bottom 120, Rear 120
• Storage: 3 x 3.5 inch HDD, 4 x 2.5 inch SSD
• Dimensions: 237 x 476 x 526 mm
• Colors: Satin Black / Matte White
• Availability: August 2023
• MSRP: $129

Glacier One D30 and D30-140 Fans

The Glacier One D30 AiO high performance cooler provides three D30 fans and a large copper base with a fin array for use with high-end CPUs. Its design provides maximum clearance from the motherboard and RAM and comes in either a 360 or 240 configuration. The MSRP is expected to be $149.

The D30 fans are now available in 140 mm (up from 120 mm originally) and can be daisy-chained to provide a single connection for all of the fans. The fans come in black and white, as well as normal and reverse models. The MSRP is expected to be $29.

Both of these products will be released in September 2023

Even if you buy one of the best power supplies, you may not like the cables that come in the box. To deal with the problem, you can either buy extension cables or look for custom cables that work with your particular make and model of PSU. 

But if you're going to hate the pack-in cables anyway, why not just buy them separately? That's the philosophy behind the new Phanteks revolt line of power supplies that the company has introduced at Computex 2023.

Available in capacities ranging from 1,000 to 1,600 watts, the Revolts are based on Seasonic's Vertex platform and meet all ATX 3.0 specifications. But to get the cables, you'll need to buy them from Cablemod, a company that specializes in custom power supply cabling.

Cablemod has a custom configurator that lets you choose exactly which cables you want, their color and their style. The company will also be selling bundles targeted at the Revolt line.

However, it's important to note that Cablemod isn't exactly cheap. When we priced out four cables -- a 24-pin ATX, 8-pin EPS, dual-SATA and 8 pin PCIe -- using the configurator, the cost was $99 and you probably need more cables than that. 

Phanteks says that, for people who can't or don't want to buy from Cablemod, it will sell two cable kits: a PCIe Gen 5 kit for $79 and a complete kit for $179. That's without the PSU.

The power supplies themselves will carry MSRPs of $199, $249 and $399 for the 1,000, 1,200 and 1,600 watt units respectively. The 1,000 and 1,200 watt units will come in black or white while the 1,600 watt one will only be available in black. All three will be available in August. 

When you do the math, the Revolt seems pretty expensive. You can get an ATX 3.0 and PCIe 5.0 compliant 1,000-watt power supply for less than $200 today from reputable companies like Corsair and MSI and those come with all the cables in the box. 

For the entry-level Revolt, you'll likely be spending over $300 to get there (we don't know exactly how much Cablemod will charge for its cheapest kit). But you will get the cables you want.

PC hardware is becoming more powerful (and power-hungry) than ever. And because of that, high-end components are larger than ever. Since the Nvidia RTX 4090 was released last year, only some case companies have released chassis designed to fit such a gigantic card. And to some extent that’s reasonable. Should every case be big enough to house a a graphics card that’s longer than an Xbox Series S and costs over $1,500?

But that said, if you are spending that much on a graphics card and comparable high-end components, there’s a decent chance you might want to show off your hardware. And that’s where Phanteks comes in with its NV7. It’s a full tower that’s slathered with glass mesh and RGB. While that might not sound all that different than many other cases released in recent years, the Phanteks NV7 manages to look and feel fresh, with lots of customization, attractive looks, and a $220 price that’s reasonable – although you will have to bring your own fans, which of course adds to the cost unless you have some you’re going to carry over from a previous build.

Specifications of the Phanteks NV7

Features of the Phanteks NV7

When I first heard about Phanteks’ NV7, I if the company had just jumped on the O11 hype train. However, after spending some time building and testing this chassis, it’s clear differs greatly from Lian Li’s flagship case line.

The NV7 has the most cooling space of any case I have reviewed, save for the Cooler Master HAF 700 EVO, but that case is bigger in every dimension and much more expensive. Twelve fans can fit in the NV7, and that’s without sacrificing any storage space.

Speaking of fans, none come with the case. That’s become increasonly common with cases designed as show pieces, like Lian Li’s O11 Dynamic. Of course, you’ll need to add fans of your own, which will increase your build budget unless you have spinners you want to carry over from a previous build.

That said, Phanteks sent along a few sets of its new D30-120 fans, which sport an internal RGB ring and clip together like the Lian Li’s Unifans, reducing the need for fiddly cables and connectors. The fans come in white or black and with regular or reversed blades. The reversed blade models allow for a more attractive look in places where the rear of the fan would normally be visible. For those who want the best aesthetics possible, this is a nice option. But Thermaltake’s Swafan EX line solves this issue by including two sets of fan blades in the box and letting you swap them out yourself rather than having to buy two different fan types.

Installing Phanteks’ fans in this case, was easy. The side fan mount is held in by one thumb screw at the top of the case. A similarity the NV7 shares with the HAF is its bottom tray, which can be rotated a few degrees to angle the bottom fans forward if you want.

As usual with Phanteks, this case comes with a plastic box full of sorted extra screws to get your build going. And the company also includes a GPU anti-sag bracket, though it’s not great. While it is adjustable to fit the width of most GPUs, it wobbles like crazy and covers the third fan on our Gigabyte RTX 3070 TI Gaming OC. This didn’t seem to affect the thermals, but I’d recommend getting a better bracket if you’re installing a huge GPU.

Like most of Phanteks’ cases, the RGB is sprinkled here and there and never overwhelming, which I like. The case features an RGB strip that runs along the edge of the motherboard tray. According to Phanteks, the RGB in this chassis uses a new D-RGB controller, which offers dual-channel output. The RGB is very customizable, thanks to the four control buttons at the top of the chassis. However, because there are four buttons, control of the light show can be a bit convoluted.

One area where the design feels a bit lazy is on the rear side panel, which needs to be screwed in from the top of the case with a tiny screw. The screw is so small that I can easily see it being lost during the building process, so you’ll want to keep a keen eye on it in case it rolls.

After removing the rear side panel, you’ll see the unique storage caddies in the NV7. Instead of the usual cage, you’ll find in nearly all mid-tower and larger cases, both the 2.5 and 3.5-inch drives are seated within a door that can be removed in case you go the m.2 route. I really like how the “storage doors” look- it’s really clean. The amount of drives the NV7 can hold is generous by today’s standards: six 2.5-inch and one 3.5-inch drive, or two 2.5-inch and two 3.5-inch drives.

By now, you’re probably asking where the power cable connects. Everything from the power cable to the motherboard IO is hidden behind a mesh door, and the cables are looped out from the bottom. While I appreciate that Phanteks is thinking outside of the box here, I found this to be annoying because it adds an extra step to assembling the build. Connecting a skinny USB cable is one thing, but looping a thick power cable through the cable channels is awkward, and I don’t like it.

Often, when you get a full tower, the IO is abundant, but that’s not the case with the NV7. You just get two USB 3.0 Type-A ports and one USB 3.2 Gen 2 Type-C port, alongside a headphone/mic combo jack set at an angle on the bottom front. The power and RGB buttons live up top, and while the controls there are nice, it would have been nice to have some ports up there as well.

Internal Layout

The Phanteks NV7 is a big boy, measuring 20.9 x 10 x 23.1 inches (LWH). That makes it roughly two inches taller than the Lian Li O11 Dynamic XL. So obviously, the NV7 isn’t limiting your options for GPU length or CPU tower cooler height. The NV7 can house GPUs up to 17.7 inches (450mm) and CPU coolers up to 7.28 inches (185mm). But with all this space, I would opt for one of the best AIO coolers here.

That said, this case can certainly hold its own on the air-cooling front, which is evident in our testing results later on.

Testing Hardware

Our testing hardware to uses Intel’s 12 Gen “Alder Lake” platform, specifically a Core i7-12700KF, which is cooled by a Noctua U12s air cooler. Our graphics card is a Gigabyte RTX 3070 Ti Gaming OC.

Acoustic Results for the Phanteks NV7

Our acoustic tests consist of three scenarios: We run the CPU at full load, the CPU and GPU at full load, and an optimized mode. The CPU full load test runs the CPU and case fans at their maximum speed. For the CPU and GPU full load acoustic test, we also stress the Gigabyte RTX 3070 Ti Gaming OC and set the fans at 75% speed, because in gaming the fans never run at 100 percent and are far too loud when they do.

For the optimized mode, we run the GPU fan speed at 30% and run the CPU and included case fans at the lowest speed that they will spin.

Because the Phanteks NV7 does not include any fans, we installed two Noctua AF120 fans- one on the side and the other in the rear for exhaust for testing. As you can see, the NV7 isn’t the most quiet case on the market, at least in our test configuration. Maybe there’s a reason why Phanteks didn’t include fans after all.

Thermal Results for the Phanteks NV7

For the thermal tests, all case and CPU fan speeds are set to 100%. The Core i7-12700K is set at a fixed 4.7GHz clock at 1.3v on all performance cores to ensure consistent power consumption across test scenarios. Letting the GPU run at 75% fan speed enables it to maintain its power target while sticking to one set reasonable fan speed, so that the temperature is the only variable.

I have to say, for such a long and untraditional airflow path, the NV7’s thermal performance was excellent. As mentioned in the acoustic test, we mounted one fan on the side and another in the rear. Our fan setup was not ideal for a case this large, but given that the airflow direction is L-shaped, the thermal performance is surprisingly good and has a ton of potential if you add more fans.

Bottom Line

Phanteks has released some interesting and affordable cases over the last couple years, but nothing ground-breaking like the Enthoo Pro II. But despite the weird cable routing design, the Phanteks NV7 is a breath of fresh air in the show case PC market that’s largely been dominated by O11 variants and the Hyte Y60.

The NV7 also pairs well Phanteks’ many custom water cooling components, with a fill port up top and a drain port in the bottom rear, plus loads of space for multiple radiators and tubes for all major components. Its $220 price is appealing as well. You’ll have to spend more to bring your own fans. But chances are if you can afford the kind of high-end components and custom cooling options this case was designed for, you can afford to pay a little extra for some nice fans as well.

MORE: Best PC Cases 2022

MORE: Best Mini-ITX Cases 2022

MORE: Best PC Cases 2022

MORE: Best Mini-ITX Cases 2022

Phanteks, known for its strong budget cases, has brought some new products to CES 2023, including two new chassis for budget and performance builds.

The flagship announcement is the Phanteks NV7, an ATX case that the company says has a "strong and defining outer frame that captures and accentuates the hardware components." In addition, it appears to have RGB around the area where the motherboard goes, which would certainly highlight it. 

Phanteks also says the chassis helps avoid a "traditionally messy" rear side and that you can see all angles with glass panels. We'll have to try building in it to see how neat this case can keep cables. It will release in February for $219.99 / € 219.90

What might be the most exciting is the Eclipse 300A, which appears to be the company's next budget case at $59.99 / €59,90. It has a mesh front panel to keep dust out but allow maximum airflow into the system. Phanteks claims that this case can fit GPUs up to 390 mm long. It should hit shelves in March.

The company also has a series of new cooling options, including the D30-120 fans ($29.99 for one, $84.99 for a triple pack), which have lighting on both sides of the fan blades and can be linked together for a cleaner installation. In addition, the Glacier One T30 Gen 2 has 240 mm ($189.99) and 360 mm radiator ($259.99) options (and the pictures show off some '80's retro lighting on the pumps), and there are Polar air coolers with up to six heat pipes.

For fully water-cooled builds, the Glacier R260 reservoir is arriving in March and can come with a D5 pump preinstalled in either black ($239.99) or white ($249.99), though you can also get it without the pump at $159.99.

Updated Jan. 5 at 12:08 p.m. ET with pricing and release dates.

Phanteks knows what it takes to land a spot on our best PC cases list. Its Eclipse P360A has occupied our best budget ATX case spot for more than a year now. But the new Eclipse G360A is an attempt to improve on budget excellence. And for the most part, the company succeeds here, while still keeping the price fairly low at $99.99.  

The Eclipse G360A isn’t a major departure from its predecessor. But for about $10 more, it brings in a third aRGB fan, delivers PWM support for all three spinners, and makes room for 360mm radiators at the top and front, to better cool current and upcoming high-end components. Combine that with the mesh front panel that already delivered good airflow with just two fans--but now packing in three--and the Eclipse G360A is a great case for the price.

It’s not without its quirks though. Cable routing in particular could certainly be improved. And there’s still no USB-C port. But with very good cooling performance (that gets even better if you move one of the three fans to the back for exhaust, or bring your own), and tasteful RGB, this is one of the best cases you can grab for around $100.  

Specifications of the Phanteks Eclipse G360A

Features of the Phanteks Eclipse G360A

Taking the G360A out of the box, the first thing I noticed was how similar it looks to the Phanteks P350X, a case I used to own. This isn’t a bad thing because the P350X was an affordable case with good build quality and good enough airflow. The similarities with that older case can be seen in the Eclipse G360A’s tempered glass side panel, PSU shroud and the RGB strip sandwiched between the two. While the tempered glass side panel isn’t tinted, it is darkened around the edges, giving it a rounded corner look like a an old-school tube TV or CRT monitor–although obviously, the glass panel is flat. In between the PSU shroud and glass side panel is a thin RGB strip that’s subtle and attractive.

More RGB can be found on the three included case fans, which are both PWM-coltrolled and aRGB, which is crazy for a case that’s just $100. A three-pack of PWM aRGB fans can cost as much as this case–or more depending on the brand and if you buy during a sale or not.

But while the G360A glistens with its aRGB fans, it’s not quite as shiny when it comes to attention to detail. During my initial disassembly of the case, I was happy to see that the tempered glass side panel features captive thumbscrews. On our review unit, the rear side panel did not have a captive screw, but Phanteks assured us that in shipping models, both side panels will feature captive screws.

The oddities continued when I installed a standard-sized PSU (Corsair RM850X). The hard drive cage was too close to the PSU for cable comfort, and moving the cage is a big pain. To move the drive cage you need to take it apart, which requires the removal of multiple screws on the top of the power supply shroud and underneath the case. Again, having a simple thumb screw or two here to allow for removal of the cage would make things so much easier.

At the top of the case, you’ll find two USB 3.0 Type-A ports (5Gbps), two buttons for RGB color and mode adjustment, a power button and a microphone/headphone jack. I would’ve preferred to see at least one USB Type-C port here. The power button is at the front of the case, above the top fan.

Like its older sibling, the P360A, the G360A features a big mesh front panel. But this time Phanteks added some personality by stamping a design on it that looks like something between a U and a V. Not everyone is going to love it, but at the very least it looks less plain than the front of the P360A. There’s still no real front fan filter here, which can be an issue. But so long as you regularly pop off the front to clean the mesh and the fans and blow out the interior of the case a couple of times a year, you should be fine. That said, if you live in an area with lots of dust or pollen in the air, you may want to pay more for a case with an easily removed dust filter.

Internal Layout of the Phanteks G360A

Even though the G360A isn’t the biggest ATX chassis, it has more room than you’d expect–which is good considering today’s power-hungry components. The G360A supports up to three 120mm or two 140mm fans at the top, along with radiator support up to 360mm (the previous P360A only supported 240mm). The front of the G360A got the same treatment as the top, with radiator support up to 280 or 360mm. The rear of the case is unchanged but good enough with support for a 120mm fan or radiator.

Today’s best graphics cards are already huge, and most indications are that next-gen GPUs will be even bigger. But with 15.7 inches (400mm) of GPU clearance in the G360A, we had no issues installing our triple-fan Gigabyte RTX 3070 Ti Gaming OC, and most next-gen cards will likely fit here as well. CPU tower cooler support comes up a bit short in the G360A–literally–at 6.3 inches (162mm). Your big Noctua NH-D15 will not fit, though most mid-sized coolers will; our Noctua U12s fit just fine. And with motherboard support up to 280mm, some E-ATX motherboards will fit. But as always, for true E-ATX support without blocking important cable cutouts, you should opt for a larger full-tower case when building an E-ATX system, or expect some complications and limitations.

Behind the motherboard tray, you’ll find two preinstalled sleds for two 2.5-inch drives. If you want to install other SATA drives, there’s a cage for two 2.5- or 3.5-inch drives under the PSU shroud. But if you want to move or remove it to make more room for PSU cables, it’s a chore since screws both on top of the PSU shroud and underneath the case hold it together.

While I didn’t have any issues closing the rear side panel on the G360A after building, routing the PSU cables to the motherboard was difficult. I really struggled to route the 24-pin cable, to the point where I barely had any slack on the wire. And the cutout for routing the EPS 12V CPU cable could be an even bigger issue.

If the end of your CPU power cable doesn’t split in half and you have to push the whole eight-pin plug through the hole, you should try to route this cable before screwing the motherboard down. If not, you may have to remove the top of the case, as I did, in order to get the connector through with enough slack to plug in the cable. But that was a huge hassle, as it involved removing 13 screws. Yes, that’s 13 screws. While we wouldn’t necessarily expect it at this price, this case definitely isn’t tool-free or particularly modular. Again, this all could have been avoided if Phanteks had just made this cutout a little bigger.

Phanteks Eclipse G360A Cooling

The previous Eclipse P360A currently holds the belt for our best budget ATX gaming, case for its great airflow and RGB. While the G360A is a clear successor to the P360A, Phanteks could’ve taken the easy way out by just including three 3-pin aRGB fans here, but instead the company delivered three aRGB and PWM-controlled spinners instead. And as we are about to see, it all comes together for some quite good performance.

2022 has been an interesting year for case fans. We’ve seen the 160mm fans in the Fractal Torrent Compact, the revival of semi-translucent 200mm SickleFlow fans in the Cooler Master HAF 700 Evo and now the D-aRGB fans in the G360A. Teh fans here aren’t ground-breaking like the company’s T30 spinners. But for a $100 case that includes three of them with RGB lighting, I’ll take them with open arms.

The fine mesh front panel on the G360A does two things: It lets an immense amount of air into the chassis, while allowing the RGB of the fans to shine out. Phanteks didn’t try to reinvent the already excellent P360A with the newer G360A, and that was probably a smart move. All it took was an increase in cooling capability and a few other little changes (like room for bigger radiators) to make this case better for 2022 and beyond.

Testing Hardware

We’ve updated our testing hardware to use Intel’s 12 Gen “Alder Lake” platform, which has shaken up the desktop landscape (and taken prominent spots on our best CPUs for gaming list). We’re now using a Core i7-12700KF, which is being cooled by a Noctua U12s air cooler. Our graphics card is a Gigabyte RTX 3070 Ti Gaming OC.

Acoustic Results for the Phanteks Eclipse G360A

Our acoustic test consists of three scenarios: We run the CPU at full load, the CPU and GPU at full load, and an optimized mode. The CPU full load test runs the CPU and case fans at their maximum speed. For the CPU and GPU full load acoustic test, we also stress the Gigabyte RTX 3070 Ti Gaming OC and set the fans at 75% speed, because in gaming the fans never run at 100 percent and are far too loud when they do.

For the optimized mode, we run the GPU fan speed at 30% and run the CPU and included case fans at their lowest spinning speed.

For an airflow-focused case, the acoustic results for the G360A aren’t bad, but I did notice a significant increase in decibel levels with our Gigabyte RTX 3070 Ti Gaming OC’s fans ramped up to 75% speed. With that said, if you’re buying this case, chances are that noise isn’t a big issue to you. Plus, it’s not like this chassis is as loud as some pricier cases, like the Corsair iCUE 5000T.

Thermal Results for the Phanteks Eclipse G360A

For the thermal tests, all case and CPU fan speeds are set to 100%. The Core i7-12700K is set at a 4.7GHz clock at 1.3v on all performance cores to ensure consistent power consumption across test scenarios. Letting the GPU run at 75% fan speed enables it to maintain its power target while sticking to one set reasonable fan speed, so that the temperature is the only variable.

The Phanteks G360A comes with three PWM aRGB fans, and they perform just as good as they look, especially when combined with the fine mesh front panel on this case.

Because the G360A comes with three fans, we tested this case both in its default configuration with all three up front, and with one of the spinners in the rear of the chassis for exhaust. As you can see, our results from the stock configuration are still impressive– trading punches with the Fractal Torrent Compact, which comes with two large 160mm fans. However, after moving the bottom fan to the rear of the case, our CPU temps improved by a significant six degrees (while increasing the GPU temperature one degree). So while moving a fan to the back should drop your CPU temperatures, the G360A delivers good thermal performance either way. Your best option may be to carry over an old 120mm fan from a previous build and install it at the back, while leaving the prettier RGB spinners up front for the best intake and looks.

Bottom Line

Phanteks has a history of delivering value and performance with its cases, as evidenced by the Eclipse P360A that’s been our favorite budget ATX gaming case for well over a year. But its time in that spot has now passed, as Phanteks has delivered a better option with the Eclipse G360A. 

At $100, it costs about $10 more than the current price of the previous Eclipse, but you get an extra RGB fan and support for larger 360 radiators at the top and front. And the G360A features PWM on all three fans for improved efficiency, noise, and control. If that isn’t worth an extra $10 to you, then you may want to scoop up the P360A while it’s still available. 

Despite some quirks in its design, the Eclipse G360A is a truly great case for gamers on a budget. Heck, even if money isn’t a major issue, it’s still nice to get good airflow and attractive RGB without spending a lot. Sure you could pay more for something else, but the Eclipse G360A is proof that you don’t have to spend a lot of money to get a great breezy box for your PC parts.

MORE: Best PC Cases 2022

MORE: Best Mini-ITX Cases 2022

At CES 2022, PC cases and cooling specialist Phanteks has taken the wraps off an intriguing new case. The Phanteks Evolv Shift XT name gives a clue to its heritage, but this model can "Shift" more than its brethren. This mini-ITX case can be physically user-adapted to your choice of three sizes. Thanks to its adaptability, Phanteks promises the Evolv Shift XT provides "Maximum Cooling, No Compromises."

Buyers of Mini ITX cases often face a sticky situation balancing the goal of a compact PC system with the need for a certain amount of power and expandability. Once you make a choice, you are normally stuck with what are sometimes quite severe limits, particularly with a mini-ITX chassis. The Phanteks Evolv Shift XT can help remove the potential for such buyer's remorse.

The central idea behind the Phanteks Evolv Shift XT seems to be that a modest PC build with lower power / higher efficiency components can settle for the most compact option. If you add a powerful GPU, you can crank up the height of the case, giving GPU fans more room to breathe. Finally, a higher third level allows users to add in a large AIO radiator as well. It looks good too, with its tempered glass infinity mirror exposed front panel.

Some key specs of the Phanteks Evolv Shift XT, beyond its triple modes, are its 2.5mm anodised aluminium build, its standard fitting of ultra fine mesh dust filters, and a convenient radiator bracket installed in the top of the case. In the video up top, you can see how all the different internal components relate to one another and the way the top radiator section of the case swings on a hinge at the back to provide decent system access. With the side panels removed, you can get at and upgrade/replace all major components quite easily.

One of the biggest performance components in a gaming PC is usually the graphics card. Despite its Mini ITX form factor the Phanteks Evolv Shift XT can fit up to triple slot designs that are up to 325mm in length, providing ample space for most users. Phanteks says that a PCIe Gen4 riser cable is included in the box, and it will be necessary to fit your GPU vertically, as pictured.

One thing we must be clear about is that the Phanteks Evolv Shift XT doesn't feature any mechanised system to expand and contract the vertical height of the case. It isn't a hugely expensive flight of fancy like some of the exotic articulated InWin cases that can be popular at trade shows. The height of this Mini ITX case is a build decision you will make as you put together your PC. However, you can be comfortable knowing that you can change the height later as you see fit, or as needed.

Availability and pricing

At the time of writing we haven't got the full specs of the Phanteks Evolv Shift XT, only some outline details that Phanteks wanted to share at CES ahead of a full-on launch in Feb. The Evolv Shift XT will hit retail with an MSRP of €169.90 / £159.90 / $169.99. It will be available in Satin Black and Galaxy Silver finishes.

Phanteks also recommends pairing with the new Evolv Shift XT case with its also new for CES 2022 Revolt SFX PSU series, available in 650W Gold, 750W Gold, or 750W Platinum. We aren't certain whether you will need an SFX PSU for the Evolv Shift XT chassis, or if that will just make things less cramped in the case. Either way, it's an attractive looking SFF option that can grow as needed with future upgrades.

Formula One cars are some of the fastest things with wheels on the planet, so you'd expect the ASRock Radeon RX 6900 XT Formula OC to set some records — and it does. The AMD Navi 21 GPU beating at the heart of the Formula currently ranks third in our GPU benchmarks hierarchy while powering some of the best graphics cards. As an interesting change of pace, you can actually find the ASRock RX 6900 XT Formula in stock at Newegg right now… currently "on sale" for $1,700. That's basically scalper pricing, and we've seen it at $1,900 as well, but we're not sure if that's ASRock's MSRP or Newegg just inflating the price.

We originally thought the RX 6900 XT felt too expensive at $1,000 when it launched in late 2020, but the past year of GPU shortages requires us to reevaluate things. When RTX 2060 cards sell for $600 or more, an RX 6900 XT for $1,700 doesn't seem out of the question. Seriously, though, it's still a terrible time to buy a graphics card, and our GPU price index shows just how overpriced most models are.

Back to the matter at hand, the ASRock RX 6900 XT Formula OC takes AMD's fastest GPU and then adds a massive air cooler and overclocks the boost speed up to 2475 MHz — 10% higher than the reference model's 2250 MHz. It also cranks the power limits to 357W when you load up ASRock Tweak and select the OC mode, which increases the power limit by 10% compared to factory stock and sets the maximum boost clock to 2659 MHz. Yeah, the way AMD reports things with maximum boost clocks is a bit odd, but it's basically like Nvidia where the boost clock is more of a conservative estimate, while the maximum boost clock is only listed in software. Basically, the ASRock Formula has a very high clock speed, and the card also draws way more than 357W, even without the OC mode.

Regardless, we used the OC mode for our testing, though the extra OC only made for about a 1–2% improvement over stock performance. Here are the specs, compared to the reference AMD RX 6900 XT:

Better cooling and a 10% overclock, combined with a 10% boost to the power limit, should all result in a decent bump to performance. We're still looking at the same Navi 21 GPU, and real-world clocks matter more than the theoretical numbers on the specs sheet. Take the power rating as an example. In our in-line power testing, we measured 383W while running Metro Exodus at the default clocks, and 415W with OC mode enabled. The GPU clocks were also about 50MHz higher than the 2475MHz figure, and OC mode increased clocks an additional 60MHz.

Of course, the same was true of the RX 6900 XT at launch. It averaged over 2300MHz during our gaming tests, despite a 2250MHz "maximum" boost clock that's not actually the maximum. But we prefer having AMD underpromising and overdelivering on clocks to the old way of doing things. ASRock also recommends at least a 1000W power supply for the RX 6900 XT Formula OC, and while a good 850W PSU should still suffice, we'd take the recommendation to heart.

The one thing we'd really have liked to get with a card like the Formula is something we rarely see: a good factory overclock on the memory. AMD released the special liquid-cooled version of the RX 6900 XT to system builders, and it comes with 18Gbps GDDR6 memory. We did poke around at overclocking on the ASRock card and were able to boost the base clock by about 140MHz (17.12Gbps effective speed), but that's basically the same speed we hit with every other Navi 21 card we've tested — even the lowly RX 6800.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

The ASRock RX 6900 XT Formula isn't quite the largest graphics card we've ever tested, but it does come close. It's slightly longer than the reference RTX 3090 Founders Edition, a card that still surprises me with its heft every time I pick it up. But the ASRock Formula isn't quite as heavy, weighing 'only' 1831g — the 3090 FE weighs 2189g. The ASRock card does use every bit of its triple-slot thickness as well, with dimensions of 331 x 124 x 59mm.

ASRock sticks with traditional axial fans, though it notes the ones on the Formula are "striped" — presumably to reduce air turbulence. The highest performing fans these days use an integrated rim, increasing the static pressure and helping to direct more airflow over the radiator fins. We've seen such fans on Asus and MSI cards, for example, but not from ASRock. The three 95mm fans should still provide sufficient cooling, but for a premium card we'd like to see a fan upgrade — these are the same fans as you'll find on ASRock's lesser Phantom Gaming line.

Another interesting twist is that compared to the Phantom Gaming cards, there's very little RGB lighting on the Formula. Some will appreciate the lack of bling and generally subdued looks, particularly anyone using a case that doesn't have a window, but if you like colorful lights you'll be better served by other options. There's just a single RGB strip that wraps around the back and top sides of the card, and it's not particularly visible unless you look from the right angle. Naturally, you can use ASRock's Polychrome SYNC software to control the color of the strip, or turn it off completely.

As mentioned above, we tested the Formula using OC Mode, which you can access through the ASRock Tweak software. You can also opt for the Quiet Mode, which cuts clocks and drops the power limit, though the card was relatively quiet even in OC mode. Manual overclocking is also possible, either via ASRock Tweak or just using AMD's Radeon Settings software, and the AMD software exposes more fine-grained control than ASRock Tweak so there's no real need to install the extra software unless you just want the one-click overclock button.

ASRock does have a rather wide range of overclocking speeds exposed on the Formula, but unless you're planning to replace the cooler with a liquid nitrogen pot, you're going to end up only using a fraction of the available range. You can set the maximum GPU clock as high as 4000 MHz, which is definitely not a thing you should do, and you can try to push the memory clock as high as 2624 MHz (21 Gbps effective). While there's a bit of headroom on both, in practice you'll want to stick to about a 150MHz maximum increase on either one, and then test to ensure stability. The minor gains in performance you'll get from such an overclock likely aren't worth the added power use, heat, and potential instability, which is why we opted to stick with the OC Mode for testing purposes.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

Our test configuration for the hardware and software remains unchanged from other recent reviews. We're using an 8-core/16-thread Core i9-9900K running stock clocks, but with DDR4-3600 memory and the XMP profile enabled. The CPU generally runs at 4.7GHz during our gaming benchmarks, though the slightly older Coffee Lake architecture can be a bit of a bottleneck at lower resolutions.

We're also still running Windows 10 21H1; we plan to update to Windows 11 in the near future and see how that impacts performance — which will also require retesting a bunch of GPUs, so we're not looking to jump on the preview builds just yet.

We limit testing on third-party cards to our standard 13-game test suite, running at 4K, 1440p, and 1080p at ultra (or equivalent) settings. We compare the review card, ASRock Formula in this case, against reference design GPUs from both AMD and Nvidia. Since we just reviewed the Asus ROG Strix LC RTX 3080 Ti and it goes after a similar extreme performance market, we've included results from that card as well.

Each test setting gets run multiple times, to ensure the consistency of our results. We're also skipping ray tracing benchmarks, though you can see a full suite of tests for the reference models in the AMD Radeon RX 6600 XT review. We expect to see similar performance scaling relative to the reference RX 6900 XT, with or without ray tracing, which means the ASRock RX 6900 XT Formula should still fall behind the RTX 3080 in ray tracing performance, particularly in games that use multiple RT effects.

The supply of AMD's Big Navi GPUs has been pretty bad since the launch last year, and the RX 6900 XT tends to be even more difficult to find. This is the first custom RX 6900 XT card we've received for review, and based on what we've seen in our GPU price index as well as the Steam Hardware Survey, there simply aren't that many cards to go around. At least according to Steam, RX 6900 XT currently makes up 0.08% of all surveyed PCs from last month — slightly more than the RX 6800's 0.05% but less than the RX 6800 XT's 0.10% and the RX 6700 XT's 0.13%. It's also less than every RTX 30-series GPU, including the RTX 3080 Ti that only launched in June and already accounts for 0.16% of surveyed PCs. 

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

With a large cooler and a healthy factory overclock, we expected the ASRock RX 6900 XT Formula to easily beat the reference model across our test suite. It did, but the last testing of the RX 6900 XT is now about four months old and a few of the games actually ran slower, likely due to driver and/or game updates. There were also a few larger jumps in performance, especially at 4K, which may also be partly thanks to updated drivers.

Across our test suite, the ASRock Formula was 6% faster than the reference RX 6900 XT. Only Borderlands 3 showed a slight 2% drop in performance, while the remaining dozen games improved by anywhere from 3% to 12%. Final Fantasy XIV, The Division 2, and Red Dead Redemption 2 all showed double digit percentage gains of 10–12%.

Looking elsewhere, the ASRock Formula also came relatively close to matching the performance of the Asus ROG Strix LC RTX 3080 Ti, trailing by 5% overall. A big part of the reason for ASRock's good showing is our inclusion of Assassin's Creed Valhalla, which heavily favors AMD's latest GPUs (and is an AMD promoted game), though Forza Horizon 4 also favors ASRock's card by 17% — the largest win at 4K. Meanwhile, Horizon Zero Dawn, Strange Brigade, The Division 2, and Shadow of the Tomb Raider all have the Asus RTX 3080 Ti card leading by 12% or more.

AMD's Big Navi architecture generally does better than Nvidia's Ampere architecture at lower resolutions. That's because the large 128MB Infinity Cache can hold proportionately more data, as 4K mostly ends up using the cache for the various buffers and doesn't have as much extra space for things like textures. Where the ASRock card was slightly slower than the Asus RTX 3080 Ti at 4K ultra, at 1440p ultra it swaps places and now leads by about 1% overall. The games that favor AMD GPUs do so even more at 1440p, with Assassin's Creed Valhalla giving the ASRock card a 34% lead over the Asus card — very much the exception rather than the rule. Remove that one game from the test suite and Nvidia's GPU would still hold onto a slight lead.

The ASRock card also sees less of an advantage over the reference RX 6900 XT at lower resolutions, and is now only 4% faster. There are also two games now (Borderlands 3 and Dirt 5) where the results from the reference card were slightly ahead of the ASRock card, but again we suspect that's due to drivers and game updates that have occurred in the past four months.

At 1080p, the Infinity Cache really starts to flex its muscle and the ASRock Formula now beats the Asus ROG Strix LC by 5%. Valhalla remains an outlier, now showing a 46% lead for the AMD GPU, but Horizon Zero Dawn and Watch Dogs Legion also have ASRock leading Asus by 15% and 13%, respectively. CPU limitations also play a role, particularly in Far Cry 5 and Forza Horizon 4, with the 1080p results only being slightly better than the 1440p results.

Given the price and the target market, we're not nearly as concerned with 1080p performance as we are with 1440p and 4K performance, though esports fans might want to look for benchmarks of the various GPUs at 1080p and see which ones can max out a 240 Hz or even 360 Hz monitor. Most of the games we test don't come anywhere close to 240 fps, never mind 360 fps — Strange Brigade being the exception, though it's still not an esports game. 

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

Performance isn't the only important metric when it comes to graphics cards. We also test power consumption using in-line monitoring tools and Powenetics software. We log power, clock speeds, temperatures, and fan speeds. We loop the Metro Exodus benchmark five times at 1440p ultra settings, and then run FurMark stress test at 1600x900 for over 10 minutes. We also measure noise levels using an SPL meter. We collected the data using the default profile, as well as the OC mode.

ASRock clearly threw caution to the wind on the RX 6900 XT Formula. Even in the default mode, it used more power in our gaming test than any other GPU in our charts. Turn on the OC mode and power use jumps an additional 30W in Metro Exodus and 40W in FurMark. The triple 8-pin power connectors are definitely required to deliver this level of power, and the PCIe power draw maxed out at under 40W even in OC mode. Compared to the reference 6900 XT, the ASRock Formula uses 35% more power in gaming and nearly 40% more power in FurMark — all for about 6% more performance on average.

We've known for some time that AMD's Big Navi / RDNA 2 architecture was really tuned to hit higher frequencies. The official boost clock on the reference card is 2250 MHz, but even that card averaged 2327 MHz in our gaming test. The ASRock matches the clocks we've seen on the RX 6600 XT and RX 6700 XT, breaking 2.5 GHz in average clocks, and OC mode adds another 60 MHz.  Clock speeds are also quite high in the FurMark stress test, with the ASRock in OC mode basically doubling the clock speed of the reference RTX 3090 card. Again there was a 60 MHz bump in clocks with the OC mode, which honestly isn't much for the extra 30-40W of power that's required. It's a classic case of the diminishing returns we see when moving up the voltage and frequency curve.

Last, we have fan speeds and temperatures, which are very much interlinked. Some cards shoot for lower temperatures and ramp up fan speeds more quickly while others are okay with temperatures in the 75C range and take a more relaxed approach to fan RPMs. Given the power draw, it's not too surprising that ASRock needs both high speeds and relatively high temperatures.

The RTX 3080 Ti and RTX 3070 Ti are the only GPUs that ran hotter, and the reference cards definitely came near the limits of their cooling potential. ASRock only needed relatively modest fan speeds in the default mode, but OC mode needed a big bump in RPMs to keep thermals in check. Compare that to the Asus RTX 3080 Ti card that employs liquid cooling, and you can see that the combination yields lower temperatures and fan speeds.

We also measured peak noise levels using an SPL meter at a distance of ~10cm from the GPU fans, pointed at the center fan. Keep that distance in mind, because it's intended to show more of a difference between the graphics cards we test. At a normal distance of around one meter (where your head is relative to the case), noise drops about 10 dB(A).

At stock settings, the ASRock card generated 48.1 dB(A) of noise running Metro (just sitting in the game rather than looping, so there's no periodic dip when the benchmark loops). That corresponded to a fan speed of just 52%, so there's plenty of room for more noise and better cooling. Using the OC mode kicked the fans up to 62% and 52.3 dB(A), while setting a static fan speed of 75% resulted in noise levels of 60.0 dB(A). 

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

As with other graphics cards that we've looked at during the past year or so, pricing and availability are generally completely messed up. ASRock's official MSRP on the RX 6900 XT Formula OC is apparently 'only' $1,700, though we've seen Newegg list it for as much as $1,900 in just the past two weeks. Still, the fact that it's available at all without dealing with the Newegg Shuffle or other shenanigans is at least some consolation. But then $1,700 is basically as much as you'd pay for an RX 6900 XT card off of eBay.

Is the ASRock RX 6900 XT Formula worth $1,700? Normally, this would be a hard pass, but in the present climate there will undoubtedly be some takers. Plus you can get Far Cry 6 and Resident Evil Village for free for the time being, potentially offsetting the cost a bit (assuming you want both games). Of course, part of the reason for the inflated GPU prices is Ethereum mining, and the RX 6900 XT only manages about 65 MH/s — not enough to command the sort of prices that we still see on RTX 3090, 3080 Ti, and 3080. For pure gaming performance, the RX 6900 XT manages to trade blows with Nvidia's top GPUs and might be worth the cost… unless you care about ray tracing, in which case the RTX 3080 Ti easily comes out ahead.

Ten months after the official launch, AMD's RX 6900 XT still occupies an interesting position. Skip ray tracing and run at 1440p or 1080p and performance basically matches and sometimes exceeds the RTX 3090, at lower prices than you'll find on Nvidia's halo card. If you're after 4K and maxed out settings, and you want to be able to enable ray tracing in the games that support it, AMD's Big Navi GPUs are decidedly less enticing. Even if you want an AMD card, there's also a case to be made for skipping the RX 6900 XT and stepping down to the RX 6800 XT — or there would be if either one were regularly in stock at prices closer to AMD's MSRP.

If you have $1,700 that needs to be spent on a graphics card upgrade, the ASRock RX 6900 XT Formula might be what you're after. It's also telling that RX 6900 XT cards are at present the only GPUs currently available at Newegg without going through the Shuffaluffagus shenanigans. That suggests people aren't champing at the bit waiting to buy AMD's top cards. The unfortunate state of affairs is that waiting for prices to come down and return to normal could mean delaying a GPU upgrade for another year or more.

If you opt for the ASRock 6900 XT, you'll get AMD's top offering, boosted to even higher levels of performance, but mind the power draw and case ventilation. Those who prefer less extreme prices and performance, or anyone who wants more RGB joy or ray tracing in their next PC build, should look elsewhere.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

The Asus ROG Strix LC RTX 3080 Ti sets a new record for out-of-box performance, no doubt helped along by the copious RGB lighting — you all know RGB makes your PC parts go faster, right? Despite having just half the VRAM of the RTX 3090, a healthy factory overclock combined with excellent cooling make this the fastest graphics card we've ever tested, at least until the next heavily factory overclocked card shows up. This would potentially be one of the best graphics cards you could buy right now, if you could actually go out and buy it.

The formula for breaking records isn't much of a secret: deliver more power and better cooling to the GPU, allowing for higher clock speeds. That's precisely what Asus has done with the ROG Strix LC line. We've seen the same core design before, in the ROG Strix LC RX 6800 XT specifically, but the RTX 3080 Ti model has to kick things up a notch. Asus added a third 8-pin PEG power connector, giving a total theoretical maximum power delivery of 525W. Here are the specs, compared to the other RTX 3080 Ti cards we've reviewed: 

Fundamentally, these are all the same GPU, so the only real difference is in clock speeds and power. Asus doesn't specify a card TGP, but does recommend at least an 850W power supply. Based on our testing, which we'll get to later, we've listed an approximate TDP of 380W. Enabling the OC mode increased the power limit by an additional 10%, meaning the card could potentially draw as much as 420W, but in practice other limits (like clock speed) come into play, which we'll see in the power testing on page four.

Besides increasing the clocks, Asus provides some other extras. For example, you get two HDMI 2.1 ports, along with the usual gamut of three DisplayPort 1.4 outputs. While DisplayPort generally remains the preferred solution among gamers for the time being, HDMI 2.1 technically allows for higher resolutions and bandwidths. We haven't seen any DisplayPort 2.0 hardware yet, and HDMI 2.1 provides for up to 8K 60Hz over a single cable, which should prove more than sufficient for many years to come.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

The Asus ROG Strix LC cards require a spacious case. Even with most of the cooling moved over to the 240mm radiator, the card remains beefy. It measures 295x133x51mm for the card itself, while the radiator measures 274x120x50mm (both are our measurements, give or take 1mm). The combined weight tips the scales at 2488g, but about half of that is in the radiator, so at least your PCIe x16 slot won't have to shoulder the load alone. Officially Asus calls this a 2.6-slot width, but it's basically a triple-slot card, plus a chunky radiator, all in pursuit of minimum temperatures and maximum performance.

As with other ROG Strix cards, there's a toggle on the card to switch between P-mode (performance) and Q-mode (quiet), but if you want maximum performance, you still need to install the Asus GPU Tweak III software. That provides access to the OC mode, which boosts the power limit by 10% and also increases the boost clock an additional 30MHz. That's what we've used for our performance testing, since we figure anyone forking over the money for this sort of card will want the maximum performance possible.

The only real omission here is in the memory department. The RTX 3080 Ti comes standard with 12GB GDDR6X, but at the current prices we still question the need for such a card. Why not give us more RTX 3090 models with double the VRAM? But that's its own answer. Double the memory would potentially limit the number of cards that can be produced, though it's not clear if the limiting factor right now is GPUs from Samsung Foundry, GDDR6X memory from Micron, or the substrate and other components needed to manufacture a complete graphics card. Still, it's a bit weird that Asus offers ROG Strix LC models on the RX 6800 XT and RX 6900 XT, but only the RTX 3080 Ti on the Nvidia side.

One nice benefit of all the extra cooling on the ROG Strix LC model is that GDDR6X temperatures remain relatively sane for a change. We've seen Founders Edition cards hit 102C or more playing games, and that goes for many other models as well. Run something like cryptocurrency mining and you'll often need fan speeds of 80% or more just to keep the memory below the maximum 110C! But the ROG Strix LC topped out at 98C on its memory, even after an hour of mining, and gaming workloads landed in the 90C range.

Not that we'd recommend buying an RTX 3080 Ti for mining purposes, as the potential profits are lower thanks to Nvidia's LHR limiter. You can now get about 70–75% of 'normal' non-LHR performance, but that still puts the RTX 3080 Ti at around 75MH/s for Ethereum in our testing. Mining of Ethereum will also go away in the coming months, once the cryptocoin shifts to proof of stake, though plenty of other memory-intensive coins still exist and might take Ethereum's place among GPU miners. Regardless, if you need some excitement, we can think of other better uses of your time and money.

You can also manually overclock the ROG Strix LC, and with a bit of tweaking and tuning you can get an extra 100MHz or so and bump the memory clocks to 20 or even 21 Gbps. For purposes of our graphics card reviews, however, we're focusing on the out-of-box experience (using the OC mode).

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

Test Setup for Asus ROG Strix LC GeForce RTX 3080 Ti

Our test configuration for the hardware and software remains unchanged from other recent reviews. We're using an 8-core/16-thread Core i9-9900K running stock clocks, but with DDR4-3600 memory and the XMP profile enabled. The CPU generally runs at 4.7GHz during our gaming benchmarks, though the slightly older Coffee Lake architecture can be a bit of a bottleneck at lower resolutions.

We're also still running Windows 10 21H1. We plan to update to Windows 11 in the near future and see how that impacts performance — which will also require retesting a bunch of GPUs, so we're not looking to jump on the preview builds just yet.

We already tested ray tracing and DLSS performance using the reference RTX 3080 Ti Founders Edition, so we're limiting testing on third-party cards to our standard 13 game test suite, running at 4K, 1440p, and 1080p at ultra (or equivalent) settings. Each test setting gets run multiple times, to ensure the consistency of our results. Now let's see how much the liquid cooling and factory overclock help with performance. 

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

We were more than a little underwhelmed with the Zotac RTX 3080 Ti Amp Holo Black, which basically tied the Founders Edition in terms of performance. The Asus ROG Strix LC scoffs at such paltry improvements and zips past the other 3080 Ti cards, then keeps right on going as it waves to the RTX 3090. Currently, the Asus ROG Strix LC RTX 3080 Ti is the fastest graphics card — out of the box — that we've ever tested.

Overall, the Asus card was 7% faster than the Zotac and Founders Edition cards at 4K ultra, and 4% faster than the RTX 3090 Founders Edition. Okay, that's not a massive difference, but it's at least measurable and consistent. We even saw up to a 12% lead over the other 3080 Ti cards in Watch Dogs Legion and The Division 2, and the Asus was up to 7% faster than the 3090 FE in individual game results at 4K.

We start to encounter some GPU bottlenecks once we drop to 1440p, and it's only going to get worse at 1080p. AMD's Big Navi architecture also tends to do better at lower resolutions, as the Infinity Cache provides proportionately larger improvements in effective bandwidth. There were several games in our suite where AMD's RX 6900 XT beats even the Asus ROG Strix LC, and at 1440p Horizon Zero Dawn joins Assassin's Creed, Borderlands 3, and Forza Horizon 4 in favoring Team Red. Still, overall the Asus card was 5–6% faster than the Zotac and Nvidia cards, and 2% faster than the RTX 3090. It was also 5% faster than the RX 6900 XT.

Cards like the RTX 3080 Ti aren't really intended to run at 1080p, at least not without other stuff like ray tracing. The Core i9-9900K becomes more of a limiting factor, and the 128MB Infinity Cache on AMD's Navi 21 GPUs helps push the RX 6900 XT to the top of the overall chart. Of course, that's also thanks to our use of Assassin's Creed Valhalla, which is laughably biased toward AMD's latest GPUs. Then again, even Watch Dogs Legion has the RX 6900 XT and RX 6800 XT beating the fastest Nvidia GPUs.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

Performance isn't the only important metric when it comes to graphics cards. We also test power consumption using in-line monitoring tools and Powenetics software. We log power, clock speeds, temperatures, and fan speeds (note that the fan speed data wasn't properly collected for the Zotac card — we're working to fix that). We loop the Metro Exodus benchmark five times at 1440p ultra settings, and then run FurMark stress test at 1600x900 for over 10 minutes.

For the Asus card, we're also collecting these power metrics in both the default Gaming mode as well as using the OC mode. Again, all of our benchmarks were done with OC mode enabled, for maximum performance. Here we'll get to see how much that affects power consumption, temperatures, and other aspects of the card. Note that in limited testing, the OC mode improved performance just 1–2%.

Power consumption for the Asus 3080 Ti was about 10W higher than the Founders Edition in Metro Exodus and 20W higher in FurMark. The Zotac card actually used less power than the Founders Edition, but it also didn't perform any better. Enabling OC mode adds another 15–20W of power, still far lower than the theoretical limit you'd expect from a 10% increase. The other charts will help explain the power difference, as all of these factors are interrelated.

The Asus card managed to deliver the highest GPU clocks we've seen from an Nvidia card to date, which isn't surprising considering the added cooling. It averaged nearly 2GHz in Metro Exodus, beating the Founders Edition by over 175MHz and the Zotac card by 145MHz. It's still a far cry from the 2.3–2.5GHz we're seeing on AMD's RX 6000 GPUs, but those are very different architectures. The gap in FurMark is even larger, where the Asus card averaged 1575MHz compared to just 1278MHz on the Zotac card, and that 300MHz advantage more than explains the 25W power gap. As for the overclocking mode, it helped a lot more in FurMark, adding 115MHz to the average clocks, but in Metro it only increased the GPU clock by about 20MHz.

Fan speeds and temperatures are the final part of the power and performance story. Asus seems to play it pretty conservatively, with fan speeds quickly ramping up to a steady 1750 RPM, give or take. The flip-side of that is that the GPU temperatures on the Asus card during the Metro and FurMark tests pretty much flatlined at 54C max, over 15C cooler than the Zotac card and 20C lower than the Founders Edition. There's definitely plenty of cooling headroom left in the tank — and enabling the OC mode doesn't really change things.

We also measure peak noise levels using an SPL meter at a distance of 15cm from the GPU fans. We had to shift the radiator around a bit so that both it and the single blower fan on the card were equidistant from the SPL meter, and the result was a noise level of 51.8 dB(A). Interestingly, that corresponded to a rather high fan speed of 80%, so setting a static 75% fan speed dropped the noise level to 48.8 dB(A). The Zotac card was a bit quieter, but also ran at higher temperatures, while the Founders Edition was slightly louder (52.0 dB(A)) but ran significantly hotter.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

It's difficult to determine the actual price of the Asus ROG Strix LC GeForce RTX 3080 Ti. Officially, it's supposed to be $2,199, but even then it's basically impossible to find the card anywhere. Amazon has a listing for the card, with two marketplace sellers asking $3,900 and $4,000. Yeah, thanks but no thanks. Most reputable places don't have any inventory, so for the time being this is basically vaporware.

When will the madness end? I don't know, but I think there's a good chance we'll see Lovelace and RDNA3 GPUs before we see RTX 30-series graphics cards on shelves selling at Nvidia's MSRPs — meaning, some time in late 2022 probably.

We get that this is an extreme card, and maybe some people could even justify paying $2,000 for it. Nvidia sold plenty of Titan graphics cards over the past five years, and this is basically as good as most of those (minus the 'professional' drivers). But when normal RTX 3080 Ti cards go for an average price of around $1,900 on eBay, and the air-cooled Asus ROG Strix RTX 3080 Ti has an average price of $2,250 on eBay during the past month, the liquid-cooled version will be out of reach of any 'normal' gamer.

That's basically the story of the past year. Every new graphics card we review feels like we're testing and writing about a product most people will never see or use. We keep hoping things will improve, but so far that hope has been all in vain. Hope can't change the realities of limited supply and extreme demand.

If a product like the Asus ROG Strix LC RTX 3080 Ti had launched a few years back, with a price of $1,200, we could have made a case for it being an extreme card built for enthusiasts. At potentially more than double that price, all we can do is give a sad shake of the head and look forward to better days.

Theoretically, Asus has an awesomely powerful and cool-running GPU for extreme enthusiasts. We'd love to be able to recommend buying one. Practically speaking, unfortunately, all we really have is a product page and a review sample; we don't even know where to point people to actually buy one, at any price. If you can find one in stock at a reasonable price, by all means, go for it. Given current market conditions, we continue to suggest figuring out what settings to use in order to play games on your current hardware while we wait (and wiat, and wait) this GPU shortage out. 

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

The AMD Radeon RX 6600 XT launched on August 11, 2021, and today we have the Gigabyte RX 6600 XT Eagle. My mom had a saying that she used to quote to me whenever I stayed up late playing computer games: You can't expect to soar with the eagles if you run around all night with the turkeys. Never mind that turkeys sleep at night, and I usually responded with a joke about aspiring to be more like an owl than an eagle. Gigabyte's Eagle GPUs on the other hand are oddly placed at the value end of the spectrum, which is not what I normally think of when people reference the majestic birds, but let's see how it compares to the competition. Will it be one of the best graphics cards, or will it hang out with the turkeys?

The Gigabyte RX 6600 XT Eagle comes with AMD's reference clocks, which I find rather surprising for a triple-fan cooler. If you want the factory overclocked models with some extra bling, that would be the Gaming and Gaming Pro line, which have substantially higher MSRPs. I picked up the base model Gigabyte card at Micro Center on launch day for $380 (plus tax), opting to save my pennies rather than splurging on the Gigabyte Gaming OC ($470 MSRP) or Gigabyte Gaming Pro OC ($500). Initial stock was good, at least by 2021 metrics, but most places are now back to being sold out and waiting for additional shipments, which often amounts to a few hundred cards a week for a larger chain. (That's what a Micro Center employee suggested, at least, saying they typically got about two dozen cards per week — and there are 25 locations nationwide.) 

Here are the specs for the three different RX 6600 XT cards we've tested. The Gigabyte Eagle sticks with AMD's reference clocks, the Sapphire RX 6600 XT Pulse comes with a minor 23 MHz factory overclock, and the ASRock RX 6600 XT Phantom Gaming OC (which AMD provided for the launch review) has a slightly higher 69 MHz factory overclock. That should mean at best about a 3% difference in performance, though there's more to a graphics card than just pure performance — at least for some people. If you want a quieter card, or something with a bit more flash, the Eagle falls short.

Getting those extras can cost quite a bit, however. The Phantom Gaming is priced over 30% higher than the Eagle and Pulse cards, and blinged out models from Gigabyte and Sapphire (e.g., the Gaming Pro OC and Nitro+) carry similarly high prices. Honestly, I find it difficult to justify the expense on what is ostensibly a mainstream card. In a normal market, we'd be looking at RX 6700 XT cards for $500 — something the RX 6600 XT can't hope to compete with in performance. But of course, the market remains anything but normal right now.

Over the past two weeks since it launched, the Radeon RX 6600 XT has followed the well trodden path of other recent GPUs in our GPU price index, selling out quickly and then showing up on eBay at inflated prices. While there are a few retail models that start at AMD's base $380 asking price, most are listed at $450–$550. Checking sold listings at eBay, the average price for the RX 6600 XT right now sits at $637, and that includes plenty of minimum spec models like the Gigabyte Eagle. More desirable GPUs like the RX 6700 XT and RTX 3060 Ti average $835 and $935, respectively. The theoretically direct competition comes in the form of Nvidia's RTX 3060, which has an average price on eBay of $715. That's to AMD's advantage right now, as the RX 6600 XT generally beats the RTX 3060 (even though it has less VRAM and memory bandwidth), though Nvidia takes the lead if you factor in ray tracing and DLSS performance. 

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

Gigabyte's packaging for the Eagle doesn't really convey the fact that you're getting a barebones reference card. The card itself looks like the Eagle logo should light up in blue, but it doesn't. It's basically the same PCB design as the higher spec Gigabyte cards, but without any RGB or factory overclock. The three fans measure about 78mm in diameter (Gigabyte says 80mm, but our measuring tape told a different story), and unlike higher end fans, there's no integrated rim. There's certainly room for a slightly larger fan, and the impact of using smaller fans can be seen in both the cooling and noise levels we'll get to later.

The Eagle measures 289 x 112 x 38mm and tips the scales at 674g (those are both our measurements), which is longer but shorter and slightly thinner than the Sapphire Pulse. The ASRock Phantom by comparison weighs 898g and measures 306 x 131 x 47mm, easily the largest of the three RX 6600 XT cards. If you have a sufficiently-long mini-ITX case, the Gigabyte Eagle might be a good fit, but again, check the cooling and noise results later in the review as you'll see it doesn't stack up that well against the other cards.

About the only real differentiating factor between the Eagle and the other cards is that it includes two HDMI 2.1 ports. While nearly all gaming monitors still tend to prefer DisplayPort 1.4 — it can do 4K at 144 Hz with DSC, and FreeSync / adaptive sync is more commonly supported over DP — HDMI 2.1 technically offers superior specs. You can read more about DisplayPort vs. HDMI, but either one should be sufficient for typical gaming use, particularly on a 1080p display, AMD's target for the RX 6600 XT. If you want to use the card with a TV, dual HDMI 2.1 ports might be more useful, but then you'd really need to run a multi-monitor setup with TVs to need both ports, as finding any gaming monitors with HDMI 2.1 support remains difficult. 

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

As with the Sapphire RX 6600 XT Pulse review, our test hardware remains unchanged other than using the public 21.8.1 AMD drivers. We have the same 13 games, tested at 1080p, 1440p, and 4K using 'ultra' settings. Technically, we use the ultra preset on most of the games, but a few (like Final Fantasy XIV, Horizon Zero Dawn, Red Dead Redemption 2, and Shadow of the Tomb Raider) use a different name or don't have presets. The important bit is that we use the same settings in each game, then we test each resolution three times, discard the first result, and take the higher of the second two runs (after checking for anomalous results). Here are the galleries of all three resolutions, with comments on the results below.

1080p Gaming Performance

1440p Gaming Performance

4K Gaming Performance

There's very little to discuss here. Technically, the Gigabyte Eagle came in as the slowest of the three RX 6600 XT cards, which makes sense considering it has the lowest boost clock. The individual test results do vary a bit between runs, though, and so it didn't place last in every single test of every single game. Similarly, the ASRock Phantom delivered the best performance overall, but the actual cumulative advantage over the other cards was miniscule — it was 0.3% faster than the Eagle at 1080p, 0.6% faster at 1440p, and 0.5% faster at 4K. The largest margin of victory was 3%, which definitely isn't something you'd notice in normal use.

When you consider that the ASRock card has 3% higher clocks and uses about 10% more power (that's on the next page), the tightly clustered results are even more telling. Eschewing bling and larger fans likely keeps power requirements a bit lower on the Gigabyte and Sapphire cards, and the RX 6600 XT runs cool enough that it really doesn't require a massive heatsink. In short, you're better off shopping for an RX 6600 XT card based on price, features (i.e., RGB lighting), and availability rather than worrying about how much faster one card might be compared to another.

Of course there's always overclocking, and better cooling might help eke out a few extra MHz and fps if you're pushing the GPU to its limit. We haven't done in-depth testing of overclocking on these GPUs, largely because it requires quite a bit of time for a relatively small payoff in terms of performance. Most GPUs can run about 5–10% faster than factory stock with a bit of effort, but that also means pushing up power use and fan speeds. We'd rather focus on the out-of-box experience instead of quibbling about final performance figures that are all still within 2% of each other.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

We've run our normal suite of Powenetics testing to check the GPU power consumption and other aspects of the cards. We used Metro Exodus at 1440p ultra and FurMark stress test at 900p for these results. The difference between the Gigabyte Eagle and the other two RX 6600 XT cards are more pronounced here than in our gaming benchmarks, and anyone concerned with noise levels will want to pay attention. Let's start with power.

The Gigabyte Eagle occupied the middle ground between the Sapphire Pulse and ASRock Phantom. It's pretty close to AMD's official 160W TDP, but ended up using just a bit more power than that — 166W in FurMark and 162W in Metro Exodus, not that 5W or less really matters. Most likely, the VRMs and power circuitry differences on the cards account for the minor differences in power use.

Average clock speeds on the Gigabyte card are basically tied with the Sapphire Pulse in Metro but slightly lower in FurMark. We put more weight on the gaming result, however, as FurMark can cause cards to behave a bit differently — some throttle more based on firmware settings, others ramp up fan speeds more to cope with the added power, etc.

This is where things become more problematic for the Gigabyte Eagle. While the core temperature was lower than the Sapphire card in FurMark, by 2C, and none of the RX 6600 XT cards ran particularly hot at a maximum of 70C (21C ambient), the Gigabyte card ran 2C hotter in the gaming test. But that's only the first part of the story, as temperatures and fan speeds are directly linked.

And here's the real problem. The smaller fans on the Gigabyte Eagle, combined with a relatively lightweight heatsink, ultimately required substantially higher RPMs to keep the card from overheating. Where the ASRock and Sapphire cards both land around 1300 RPMs in Metro and FurMark (give or take), the Gigabyte card runs the fans at over 2100 RPM. As you'd expect, that results in more noise, and the difference is clearly audible when using the cards — though even the Gigabyte card doesn't get particularly loud compared to extreme GPUs that can draw 350W or more in power.

We measured noise levels using an SPL meter, at a distance of 15cm. That focuses the microphone on the GPU fan noise rather than other sources like case or CPU fans. Ambient noise levels for testing were 33 dBA, and like most other modern GPUs, the fans on the Gigabyte card will stop spinning when GPU temperature falls below about 55C. Interestingly, where the ASRock card fully halts the fans between loops in the Metro benchmark, and the Sapphire card dips down to under 1000 RPM, the Gigabyte card takes longer to return to cooler GPU temperatures and as a result the fan only dipped slightly below 2000 RPMs. While actively playing games, the fans stabilize at around 2200–2300 RPM.

Fan noise for the Gigabyte Eagle was 44.4 dBA while gaming, about 7dB louder than the ASRock Phantom and Sapphire Pulse. Again, that's not incredibly loud — RTX 3070 Ti and above tend to land at 50 dB or more — but it's clearly more audible than the other two cards. If you're hoping to build a near-silent gaming PC, we definitely recommend the Sapphire as a better option, which also generally cools better and runs just as fast, for the same nominal price.

The fan speed as a percentage was 54% on the Gigabyte card. Just to see how bad things might get (e.g., in a hotter testing environment), we set a static fan speed of 75% as well. Once we did that, the Eagle noise level was 62.1 dB, compared to 60.0 dB on the ASRock Phantom and 54.1 dB on the Sapphire Pulse. We don't expect these cards to normally get hot enough to require such fan speeds, but it's yet another point in favor of the Sapphire card.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

Superficially, the Gigabyte Eagle looks like it should be just as capable as any other midrange GPU. It has three fans and a dual-slot configuration, and with a TDP of only 160W we would expect it to run cool and quiet. Performance ended up being nearly identical to the other two cards, though technically it did come in last place out of the three RX 6600 XT models we've tested, but the fans tell a different story. They work harder, and as a result generate more noise.

If you can't find any other RX 6600 XT cards in stock, the Gigabyte Eagle should still do fine. We also haven't tested every one of the dozen or more RX 6600 XT cards that are presently launched in the U.S. market, and some might be even worse than the Eagle when it comes to fan noise. But if you have the choice between something like the Sapphire Pulse and the Eagle, we'd pick the Pulse every time.

It feels like Gigabyte just cut a few too many corners on the RX 6600 XT Eagle. Slightly larger fans could move more air and generate less noise. Integrating a rim like the Sapphire Pulse would also improve static pressure and cooling. Gigabyte makes a point of having the middle fan spin clockwise while the two outside fans spin counterclockwise, saying this reduces turbulence, but the net effect on cooling wasn't better than competing designs. What's more, Gigabyte appears to use the exact same "3x80mm Windforce 3X cooling system" on its higher-end Gaming and Gaming Pro models. Those do add a bit of RGB bling to the package, but I can't help but question using these same fans with higher GPU clocks on "premium" models.

The Gigabyte Eagle isn't a bad card, but it does seem uninspired. It will get the job done, just like other budget RX 6600 XT models. Will it do so in a better fashion than the competition? Probably not. It's time to update the fans and heatsink to get back to parity with the competition, because right now, the dual-fan Sapphire Pulse comes out ahead and doesn't cost anything extra.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

The AMD Radeon RX 6600 XT officially launched on August 11, 2021, and we're looking at the Sapphire Radeon RX 6600 XT Pulse today. Initial supply, at least for overseas markets and some brick and mortar stores like Micro Center, was better than we've seen for other new GPU launches during the past year. If supply could keep up with demand, even at a higher starting price of $380, it could be one of the best graphics cards. A card in hand is worth two in the virtual shopping cart, after all. But indications are that supply isn't keeping up with demand, even on these 'lesser' GPUs.

That's despite the fact that initial demand was quite a bit lower for the RX 6600 XT. After all, it's only about the same level of performance as the previous generation RX 5700 XT, for about the same price as well — in theory, anyway; the RX 5700 XT generally sells for about twice its launch price right now. That's because it can do over 50MH/s in Ethereum mining, while the RX 6600 XT can only do about 32MH/s, and while they both hit similar levels of mining efficiency after optimizing mining settings, deploying fewer cards at higher hashrates is generally preferred by miners.

Anyway, on to the story at hand, the Sapphire Radeon RX 6600 XT Pulse. This is, for all intents and purposes, a reference model RX 6600 XT. It does come with a modest factory overclock, but the power use does adhere to AMD's 160W TDP, unlike the ASRock card we received for the initial launch review. Here's a quick look at how the specs compare, and we'll also have a Gigabyte RX 6600 XT Eagle in the charts (with a separate review coming soon).

The Sapphire Pulse comes with a very modest 23MHz factory overclock, compared to the reference specs. That's a one percent overclock that should basically end up as noise in the benchmarks. The Gigabyte card actually uses the reference clock, so we can at least make some comparisons there, though differences in cooler design and other elements come into play.

Overall, though, we're looking at a potential 3% factory overclock on the ASRock card, while the other two RX 6600 XT samples are lower. There are also higher clocked RX 6600 XT cards, but they cost quite a bit more — like the Phantom Gaming. The Sapphire and Gigabyte 6600 XT cards we have for testing and review actually come with a theoretical MSRP of $380, while the ASRock Phantom costs over 30% more. The only place you're likely to see such a price is at Micro Center, at least in our experience, but the Pulse at least looks a lot better than the $550 MSRPs we've seen on a few other models, especially since end-user overclocking can mostly close the gap.

Something else worth noting is that, now that we have almost two weeks of data to draw on, the average selling price of the RX 6600 XT on eBay right now sits right around $640. If you're willing to deal with eBay, you could get lucky on an auction and pay less than that, but in general, we're looking at a 50% or higher markup on the RX 6600 XT. About 15% of that goes to eBay, so the scalpers aren't the only ones taking their pound of flesh. Our advice, as usual, is to avoid eBay if at all possible.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

There's no question this is more of a 'budget' RX 6600 XT when you look at the Sapphire design and box. The packaging is about as no-frills as you can get, with a plain brown cardboard box inside the outer sleeve and a relatively small box to begin with. That's not a bad thing, and if you're interested in building a smaller PC, the dual-fan cooler and relatively compact design are just what you'll need.

The Pulse measures 240x118x43mm (our measurements — the official size is 240x119.85x44.75mm) and weighs just 611g. The ASRock Phantom by comparison weighs 898g and measures 306x131x47mm, while the Gigabyte Eagle measures 289x112x38mm and tips the scales at 674g. The Sapphire takes up just a bit more than a standard 2-slot width, but can probably be classified as such still, though we recommend users avoid putting any expansion card in the adjacent slot regardless, as that can impede airflow and lead to substantially higher temperatures.

Sapphire uses two custom-sized 88mm fans, with integrated rims that help improve the static air pressure and cooling. We'll see the effects of that design choice when we get to the power and cooling tests later, but potential buyers shouldn't have anything to worry about. Like most other RX 6600 XT cards, it also includes a single 8-pin power connector, and unless you plan on radical overclocks with LN2, that should be more than sufficient.

Aesthetically, there's zero lighting on the Pulse, RGB or otherwise. Some people will appreciate that, as it means you can put the card in a PC in your bedroom and not have to deal with the technicolor light show. Of course, you'd still need a case and motherboard that don't have glowing lights everywhere, but we'll leave that as an exercise for the PC builder. On the other hand, if you're a fan of RGB lighting, you'll probably want to look elsewhere.

The one noteworthy extra Sapphire includes with its graphics cards is Sapphire Trixx. Most graphics card manufacturers have software of some form, either for overclocking or hardware monitoring, or both. Sapphire doesn't provide overclocking, but it does have HW monitoring if you want it. More importantly, it has Trixx Boost, which can use AMD's Radeon Image Sharpening to let you select a lower resolution and then have everything scaled to the normal resolution. The default 85% setting is a reasonable option, and as an example, it renders at 2176x1224 and upscales to 2560x1440. We included some benchmarks at 1440p with Trixx Boost enabled, and you get about 25% better performance than native 1440p.

What about image quality? There's a bit of a tradeoff there, mostly noticeable on high contrast text. Subjectively, though, most people likely wouldn't even notice the difference unless they're specifically told to look for it. I gave it a shot, having someone else set the in-game resolution to 2560x1440 or 2176x1224 while I was out of the room, then coming back and trying to determine whether Trixx Boost was enabled or not. I managed about 75% accuracy, but considering random guessing would get me 50%, that's not too bad.

Certainly, Sapphire makes a case for Trixx Boost being more practical than end-user overclocking. If you custom-tune your GPU, adjusting clocks and increasing the fan speed, you might get a 10% increase in overall performance. However, that generally comes after an hour or so of tweaking and tuning, and it's still not guaranteed to be 100% stable. On the other hand, Trixx Boost can easily get you 20% more performance, in about 30 seconds, for a minor drop in image quality but zero reduction (that I noticed) in stability. It's sort of like AMD's FidelityFX Super Resolution (AMD FSR) only it was already available a year ago, and it works in all games. We have to wonder if Sapphire will look into updating Trixx Boost to incorporate AMD FSR instead of RIS, given both are open source, but AMD FSR has better overall image quality supposedly.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

Our basic test hardware remains unchanged from the initial RX 6600 XT launch, except we're using the public 21.8.1 AMD drivers now. You can refer to the launch review for additional details, but we're using the same 13 game test suite. We've also enabled Trixx Boost and tested it at 1440p (2176x1224) as an extra data point. All of the games worked with Trixx Boost, though Dirt 5 required setting the desktop resolution to 2176x1224 and then running at "100% Native" scaling. We'll start with 1080p and then look at 1440p and 4K ultra results.

Pretty exciting stuff, right? At 1080p, the three RX 6600 XT cards all land within 0.5% of each other. The only real difference came from our Ryzen 9 5900X testing, where performance improved by about 2% overall, thanks mostly to a couple of the games that were really hitting CPU limits on the i9-9900K. There are a few cases where the Sapphire card came out slightly ahead, but mostly it's just margin of error differences. Let's move on.

The results at 1440p are again super close. This time, however, we've got Trixx Boost enabled. That improved performance by 24% on average, though a few games showed smaller or larger gains. Red Dead Redemption 2 was the outlier, with 52% higher performance. It's not clear if the card was running out of memory, or memory bandwidth (maybe both), but the lower resolution clearly helped a lot. Forza Horizon 4 showed the lowest gain of 13% because it was already hitting CPU limits. Everything else landed in the 20–32% range of improvement.

Trixx Boost is a nice extra feature, though it's possible to get similar results by manually creating a lower resolution as well — just without Radeon Image Sharpening doing the upscaling. We wouldn't necessarily spend a lot of extra money on a Sapphire card just to get access to Trixx Boost, but it's there if you happen to encounter a game that can't quite hit a steady 60 fps or more at your desired resolution.

Last, we have 4K, a resolution that's generally a better fit for faster GPUs — unless you don't mind 30 fps gaming. Forza Horizon 4 was the only game to clear 60 fps at 4K, and several games are right at the 30 fps threshold. We also saw the biggest difference between the ASRock and Sapphire cards at 4K: just 3%, in Red Dead Redemption 2. Despite its substantially larger cooler and third fan, plus a higher TDP and modest overclock, the Sapphire Pulse is only a hair slower than the ASRock Phantom Gaming. 

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

We're running our normal suite of Powenetics testing to check the GPU power consumption and other aspects of the cards, using Metro Exodus at 1440p ultra and FurMark stress test at 900p. Things are perhaps a bit more interesting here, as there are more noticeable differences between the three RX 6600 XT cards. Let's start with power.

The Sapphire Pulse came in slightly below AMD's official 160W TBP rating in Metro Exodus and FurMark and used 17–23W less power than the ASRock Phantom. Meanwhile, the Gigabyte card landed in between the two, though it's slightly closer to the Sapphire card. Interestingly, power use was measurably lower despite having nearly identical performance, though a few watts at least likely goes to the RGB lighting and extra fan on the Phantom card.

Average clock speeds on the Sapphire Pulse were lower than the Phantom by 100MHz in Metro, which was a bit surprising. Clocks were also about 90MHz lower in FurMark, though that hits power limits and cards often behave quite differently in that sort of workload than in gaming. We noticed some oddities with Metro Exodus after our initial benchmarking, however, which may account for some of the differences here.

As far as temperatures go, the ASRock Phantom came out with a clear lead. It ran at higher clocks, drew more power, and still had temperatures 8–10C lower than the Sapphire Pulse and Gigabyte Eagle. But temperatures on their own don't tell the full story.

Higher fan speeds can improve temperatures at the cost of making more noise. That didn't happen with the Sapphire Pulse, where the fans had slightly lower RPMs than on the ASRock Phantom card — and the Gigabyte card didn't do so well. But even fan speeds don't tell you everything, as different fans have different acoustics.

We measured noise levels using an SPL meter at a distance of around 15cm. That helps it to focus on the GPU noise and not system or CPU fans. Ambient noise levels for testing were 33 dBA, and like most other modern GPUs, the fans stop spinning when GPU temperature falls below about 55C. You can see how quickly that happened in the fan speed over time chart, where the RX 6600 XT cards are the only ones to actually halt the fan in between loops of the Metro benchmark. While gaming, noise from the Sapphire Pulse was only 37 dBA, tying the ASRock Phantom, so not only does it run cool, but it's also nearly silent. (The Gigabyte Eagle was noticeably louder, if you're wondering — about 44 dBA.)

During gaming, the fan was spinning at around 30%, which was all that was needed to keep temperatures in check. We also set the fan speed to a static 75% as a secondary test, and it generated 54.1 dB of noise — slightly quieter than the Phantom, which measured 60 dB. Admittedly, that's not as quiet as some of the other GPUs we've tested, but those mostly have larger fans and bigger heatsinks and cost quite a bit more than the Pulse.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

Some people want an in-your-face graphics card with laser beams coming out the ports; others prefer something a little less ostentatious. Sapphire's Pulse line caters to the latter, with no fancy lights but with a competent cooling setup that runs just as quiet as some larger cards. If you don't have a case window, or you're looking for a smaller card that can fit a mini-ITX build, this might be the RX 6600 XT for you.

Performance comparisons are largely meaningless with the sort of factory overclocks we've seen so far. Theoretically, the ASRock Phantom Gaming has a 2% advantage in clock speed, plus a 10% higher power limit, but it didn't really matter in actual testing. Across our entire test suite, we only saw an average performance difference of 0.5%. Of course, no one would notice that in real-world gaming, and even in benchmarks, it's not really significant. Plus, Sapphire offers Trixx Boost as a way to potentially hit much higher frame rates if you want to use upscaling on pretty much any game out there.

What it really boils down to is the type of graphics card you're looking to buy. Actually, what it really comes down to is availability — everything continues to sell out, even at retail stores, so unless you're willing to wait quite a while, your options are going to be limited to whatever you can find. But if you care about looks and bling, there are better cards out there — including Sapphire's Nitro+.

AMD's RX 6600 XT performs fine in nearly any game out there, especially if you're only looking to run at 1080p. The only real exception is games with ray tracing, particularly games that use a lot of RT effects (e.g., Cyberpunk 2077, Control, or Minecraft). If you want to play games with ray tracing on, Nvidia's GPUs remain the better choice, and you can typically find the RTX 3060 for roughly the same price as the RX 6600 XT. If all you want is a new GPU that's efficient and hopefully won't cost much more than $400, the RX 6600 XT is one of only a few options. Until the GPU and component shortages ease up, that's about the best we can get.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

A few months ago, when I was visiting Phanteks to pick up a case for review, I spotted something in the corner of my eye: A fan that I hadn’t seen before. I asked about it, but was quickly told that I wasn’t supposed to see that and swiftly hushed out the door. Fast forward to today, and I have four Phanteks T30-120 fans on the test bench, along with a wind-tunnel to put it through its paces.

All Phanteks told me back then was that this fan would be better than Noctua’s best spinners, triumphing its performance by a healthy margin. Of course, I was skeptical, but it’s now here, and lo and behold, the T30-120 is actually impressively good. Because Phanteks claims that the T30-120 is the ultimate fan, I’m mercilessly pitting it against the Noctua NF-A12x25 – the benchmark for fan excellence in today’s market. 

Let’s dig in.

What is the T30-120’s secret?

Much like Noctua’s flagship fan, the NF-A12x25, the T30-120 is built using forward-swept fan blades made from a liquid crystal polymer (LCP) material, edging extremely close to the frame at just 0.5mm for maximum static pressure. But whereas Noctua’s spinner is built with nine blades, Phanteks’ T30-120 only has seven. Unlike Noctua’s fan, the T30-120’s frame is also made from the LCP material and comes with rubber anti-vibration pads glued on at the corners, so there is no radiator gasket.

At the center, you’ll spot the fan’s motor, which is a SUNON-made 3-phase hub with a magnetic, levitated dual-vapo bearing. This is an extremely quiet and smooth motor, with a long rated lifetime. It’s also an extremely powerful motor, as the fan is capable of spinning at up to 3000 RPM.

But there is another reason why the hub has to be quite powerful, and this is the real secret of how the T30-120 outperforms Noctua: Phanteks is kind of...cheating. Okay, I kid. But the T30-120, unlike the vast majority of 120mm fans, doesn’t restrict itself to the 25mm standard for frame thickness. Rather, the T30 is, as its name implies, 30mm thick, giving room for 25% larger fan blades and thereby more performance. Of course, this trick nuances the performance uplift somewhat as it’s no longer a 120x120x25mm fan, but performance is still performance. And lots of fan applications have room for a thicker fan.

One SKU to rule them all

From the factory, the T30-120 comes configured to spin at ‘just’ 2000 RPM, responding to the PWM signal in a linear fashion. To get to 3000 RPM, you’ll have to flip a switch at the back of the fan to pop it into ‘Advanced’ mode. Do be careful with this mode though, as it’s so fast that you will hurt your fingers in the fan if you touch it during operation.

For silence enthusiasts too lazy to set their own PWM curves, there’s another mode: Hybrid. This one limits the maximum rotational speed to just 1200 RPM, and at PWM signals below 50%, completely shuts the fan off. 

Having all these different modes at the flip of a switch is actually a great way of keeping the product line simple and giving the customer easy flexibility – there are no separate SKUs for the different flavors, nor do you need to mess about with messy low-noise adapters. Just buy the fan and use it as you intend, and if you ever feel you want a different fan mode, just flip the switch at the back to take the fan from whisper quiet to brutally powerful.

Other than that, there really isn’t much to the T30-120. It does come with a daisy-chained cable to spare you the pain of splitters, and because it’s a little thicker than most, it comes with longer screws so that you can still fix it to your AIO or radiator of choice. Heck, it even comes with a second set of even longer screws than the already-extended ones, because Phanteks also sells the digital Halos (which we used in the liquid-cooled Mirror Maze PC build), for when you want to turn this ultimate fan into the ultimate RGB fan.

Synthetic Testing: Phanteks T30-120 vs Noctua’s Finest

For the first test, I simply placed the fan in an open space, chucked the dB meter in front of it at 30cm, and measured the noise levels emitted throughout the RPM range with intervals of 250 RPM. Both the Phanteks and Noctua fans will happily turn at just 250 RPM at the lowest PWM signal, but whereas Noctua’s spinner wraps things up at about 2070 RPM, Phanteks’ T30-120 will soar well to 3000 RPM. In fact, it’s so powerful at this speed that the fan started pushing itself along my test bench. Hah.

The T30-120 came out louder than the Noctua NF-A12x25 at identical RPMs, but that’s no real surprise. With its thicker design, it should push more air at lower RPMs, and therefore achieve better acoustic efficiency. Hence the need for the next test.

Noise-Normalized Testing with a Wind Tunnel

For the next test, I decided to focus on noise-normalized performance using a wind tunnel. When I spotted the fan way back, I hinted to Phanteks that I was looking into proper fan testing, so they set up a full kit that I could use to test not only the T30-120, but also a ton of other fans in the future (stay tuned for more). This wind tunnel is extendable, can fit 120mm and 140mm fans, has air straighteners throughout it to (mostly) resolve any turbulence, and measures the air speed exiting the tunnel with an anemometer. Pretty simple.

For this series of noise-normalized tests, I wanted to know the fan’s performance at a handful of noise levels, both in open-air configurations and when restricted by a radiator. The lowest noise level I tested at was 34 dBA, as that is the noise floor in my office, at night. That’s a bit high for proper scientific tests, but still very quiet and most rooms will have a similar noise floor anyway. And let’s be fair, none of us live in an anechoic chamber – so the first value was the RPM figure the fans achieved right before breaking through the noise floor. After that, I measured from 35 dBA and upwards, noting down the rotational speed of each fan at 2.5 dBA increments.

Then, I chucked the fans onto the wind tunnel, set them at the rotational speeds I had recorded for each noise level (noise testing cannot be done on the tunnel as it echoes and resonates), and recorded the resulting air speeds for both fans, with and without a simple radiator as restriction, then using that value to calculate the airflow. This methodology isn’t real-world, think of it more as a ‘synthetic benchmark’ if you will, but it’s consistent, and the results don’t vary when the temperature in my room changes.

The results speak for themselves. Phanteks’ T30-120 merrily beats the Noctua NF-A12x25 with the highest airflow for each noise level. When restricted by a radiator, the difference isn’t quite as drastic as in open air, but the T30-120 still wins by a healthy margin.

A Word on Future Fan Testing

Of course, these are only synthetic tests – I’m sure you’re wondering where my real-world tests are. At this time, I’m still working on building a full fan testing setup that combines both synthetic and real-world testing on a PC. I would have liked to include a real-world result in this hands-on piece, but the setup isn’t complete yet. Sorry, you’ll have to wait until next time, but we’ll also be including more fans in the future.

Conclusion

Phanteks’ T30-120 might be ‘cheating’ a little by extending the frame to 30 mm thickness as opposed to the industry standard of 25 mm, but the company didn’t skimp on the rest of the spinner’s engineering. Sure, it’s not a Noctua level of refinement that’s going on here, but the results to speak for themselves: the 5-mm thicker T30-120 beats Noctua’s famous NF-A12x25 – a fan that is considered to be the best PC fan in the world – at least from a noise and acoustics point of view.

With this in mind, I’m not really upset that Phanteks decided to make the fan a little thicker. In fact, I’m actually surprised it took this long for anybody to do it at all – it’s such an obvious opportunity, and I’m curious whether this is the start of a new era of thicker PC fans. I’m curious how a 30-mm thick 120mm Noctua fan would perform, but I don’t expect to find that out anytime soon. 

The T30-120 might not be the best-looking fan with its rough, industrial design (though it’s Noctua’s fans are visual stunners, either), and you’ll have to double-check whether it fits in the configuration you want to use due to the added thickness. But the thickness shouldn’t be an issue in the vast majority of builds, and if you want it with RGB, you can chuck on a Phanteks Halo light ring.

All things considered, between the exceptional performance, construction quality, speed switch, 6-year warranty, and daisy-chaining cables, the Phanteks T30-120 is an easy recommendation, even at its $29.99 price tag (which is also roughly the going rate for the competing NF-A12x25). I’m sorry Noctua, but first place is no longer yours – unless for some reason you need to stick to 25mm.

The Galax GeForce RTX 3080 Ti SG — the SG is for "Serious Gamer," if you're wondering — looks to compete with the best graphics cards by offering a unique extra: a fourth clip-on fan to boost cooling performance. That's double the fans of Nvidia's reference GeForce RTX 3080 Ti Founders Edition, if you're keeping track. The 3080 Ti currently ranks as the second fastest GPU overall in our GPU benchmarks hierarchy, just a couple percent behind the RTX 3090, and less than one percent ahead of AMD's competing RX 6900 XT — though those results don't include ray tracing, DLSS, or FSR performance comparisons. How does Galax fare against the competition, and more importantly, will you be able to find one in stock anywhere? The answer to the latter, unfortunately, is generally no, though readers over in Europe might have better luck.

Galax products are also sold under the KFA2 brand overseas, but other than slight differences in the box art, the cards are the same. The parent company for both brands is Palit, which by volume is one of the largest graphics card companies in the world. Despite the name, the Serious Gamer edition is basically equivalent to Nvidia's reference model. Out of the box, it sports a boost clock of 1695MHz, just 30MHz higher than the Founders Edition. If you install Galax's Xtreme Tuner Plus Software and use the 1-Click OC option, you'll get a mostly meaningless bump to a 1710MHz boost clock — that's less than a 1% difference. Still, we've used the factory overclock for testing, and we'll use that for the specs table below.

We've now tested three different RTX 3080 Ti cards, with the two custom cards having identical specs. But specs don't tell the whole store, as card design and cooling can also factor into the equation. The Zotac Amp Holo is the largest of the three cards, at least by certain metrics. It measures 317.8mm x 131.8mm x 64.6mm and weighs just over 1500g. The Galax SG card measures 317mm x 113mm x 61mm by comparison, so it's a bit longer but not as tall or as thick.

However, the Galax card also includes a support bracket that adds 12mm to the length and 17mm to the height of the card, and then there's the extra clip-on fan. That's 25mm thick, though it's only at the end of the card and sits on the back. The Galax RTX 3080 Ti SG weighs 1424g, including the extra fan but not the support bracket. In other words, it's a rather large card and definitely not something you'd want to try to cram into a smaller case.

As far as clock speeds go, even compared to the reference 1665MHz, the theoretical difference in performance is only 2.7%. In practice, all of the Nvidia Ampere GPUs we've tested tend to run well above the official boost clocks, and it's not uncommon to see cards with lower official boost clocks actually run at higher speeds in practice. As with the Zotac 3080 Ti, most differences between the GPUs can be chalked up to margin of error in testing, or differences between driver revisions and/or game patches.

The biggest problems with all of the RTX 3080 Ti cards are price and availability. Given current demand, one could argue that the RTX 3080 official launch price of $699 was simply far too low. That's mostly because cryptocurrency mining profitability basically quadrupled after the GPU launched, which meant every miner and their dog wanted to buy as many 3080 cards as possible. But absent cryptomining (see best mining GPUs) and component shortages, $699 was still the right price. Asking $1,199 for the RTX 3080 Ti for an extra 5–10% performance was too much, and third-party cards are taking things even further. Galax informed us there's a lot of variability in pricing right now, but that it expects the retail price for the RTX 3080 Ti SG to be around $1,500 — or £1,150 or €1,350. That's too high in our book, but unfortunately there are apparently plenty of people willing to pay such prices still, as the GPU price index indicates.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

Galax packs in a few extras with the SG model of their card, specifically the fourth clip-on fan and a mounting bracket that doubles as extra RGB lighting. The design is otherwise pretty standard, with triple 92mm fans and a metal plate that covers the back of the card. The fans will stop completely when GPU temperatures are below about 50C, at which point the card is entirely silent.

The end of the plate has cutouts to let air flow through, which we've seen on many of the Ampere cards, and the fourth clip-on fan attaches over the cutouts, operating in a push-pull configuration. In testing, the extra fan does improve temperatures a bit, and GDDR6X temperatures in particular were much better on the Galax card than on Nvidia's Founder's Edition.

Above is a shot of mining performance with the Galax card, which tends to be the best way to stress test the memory. Here we've let NiceHashMiner run for over 15 minutes mining Octopus, after doing the initial 'precise' benchmarking of the algorithms. Octopus at the time was the most profitable algorithm, and it also generally hit the highest memory temperatures — 96C in this case. Kawpow also tends to be more profitable than Ethereum mining on the 3080 Ti, depending on the day, and memory temperatures were 8C lower in that case.

The RTX 3080 Ti Founders Edition by comparison quickly hit 110C on the memory before fan speeds went to 100% and GPU clocks dropped. GDDR6X temperatures while gaming were about 10C lower than the worst mining temps we recorded, however, so outside of mining, the cards tend to do okay. Still, we'd much rather have the Galax cooling and fans than those of the Founders Edition. Not only does it keep the memory temps in check, but it also ran quieter — both desirable traits.

While we're on the subject of mining, the best case right now — even after the recent bump in Bitcoin and Ethereum prices — was around $5 to $6 per day in net profits. That might sound decent, but for a GPU that costs at least $1,200 and often goes for closer to $2,000, that means a minimum of 200 days to break even, and potentially more than a year. The Ethash limiter Nvidia has imposed on all RTX 30-series GPUs other than the RTX 3090 at least seems to be working so far, but there are almost certainly miners still buying these cards — either in hope of a hack to remove the limiter, or simply because other hashing algorithms still work fine.

The Galax card includes two 8-pin power connectors, which is pretty typical of high-end RTX 30-series GPUs — only a few models have opted to go with triple 8-pin connectors. It also features three DisplayPort 1.4 outputs and a single HDMI 2.1 port. The TDP sticks with Nvidia's reference 350W value, and in practice, the Galax card tended to come in slightly below that mark.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

Our test configuration for the hardware and software remains unchanged from other recent reviews. We're using an 8-core/16-thread Core i9-9900K running stock clocks, but with DDR4-3600 memory and the XMP profile enabled. The CPU generally runs at 4.7 GHz during our gaming benchmarks, though the slightly older Coffee Lake architecture can be a bit of a bottleneck at lower resolutions.

We already tested ray tracing and DLSS performance in the reference RTX 3080 Ti Founders Edition review, and for third-party cards, we're only going to look at our standard 13 game test suite, running at 4K, 1440p, and 1080p and ultra settings. Each test setting gets run multiple times, to ensure the consistency of our results. Considering the similar specs on the Galax card and other 3080 Ti models, we should see very little difference between the offerings.

Performance from the Galax card was right where we'd expect, though a few games show a slight drop or gain in performance that we believe comes from game updates (drivers could also be a factor). The Galax card came in less than 1% behind the other two 3080 Ti cards we've tested in our overall metric, and all of the individual results are within margin of error. We tested the Galax with Nvidia's latest 471.41 drivers, while the other two cards were tested with the launch drivers.

In short, we didn't see anything particularly worrisome with the Galax card. Slight variations between benchmarks are the norm, particularly for cards that were tested two months apart from each other. Anything less than a 5% difference in performance certainly isn't noticeable when actually playing games, and we'd say even 10% differences are right at the threshold of perceptibility. That doesn't mean one card isn't slightly faster, but small differences in performance aren't nearly as important as other aspects, like aesthetics or price — or just finding a GPU in stock.

At 1440p, the overall results from the three RTX 3080 Ti cards are even closer, with the usual variations in the individual games. Sometimes the Galax card is the 'slowest' of the three 3080 Ti cards; other times it's the 'fastest' option. Practically speaking, the cards are all basically tied. Do note that Red Dead Redemption 2 got a recent update that enables DLSS support, and that may have dropped the non-DLSS results a bit.

Finally, 1080p ultra puts more of a load on the Core i9-9900K, and performance ended up as a tie between the three 3080 Ti cards. Also note that AMD's RX 6800 XT now places ahead of the 3080 Ti, though just barely — another tie. We'd generally skip buying a card like the RTX 3080 Ti for 1080p gaming, as 'slower' GPUs end up performing just as well. Even if you have a high refresh rate eSports monitor, usually the game matters more than the GPU. CSGO and LOL don't need a 3080 Ti any more than a 3080 or even a 3060, for example.

You can of course make the argument that maxed out ray tracing effects at 1080p could benefit, but the most demanding ray tracing games — like Cyberpunk 2077 as an example — also tend to support DLSS 2.0, which means 1440p and 4K are still within reach of GPUs like the 3080 Ti. There's nothing stopping you from using an extreme GPU at 1080p, but overall the 3080 Ti was only 6% faster than the RTX 3080, and plenty of games run into CPU bottlenecks. 

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

Performance isn't the only important metric when it comes to graphics cards. We also test power consumption using in-line monitoring tools and Powenetics software. We log power, clock speeds, temperatures, and fan speeds. We loop the Metro Exodus benchmark five times at 1440p ultra settings, and then run FurMark at 1600x900 for over 10 minutes.

In a bit of a surprise, power consumption for the Galax 3080 Ti came in slightly lower than the other two RTX 3080 Ti cards. It's only 3W less than the Zotac card in Metro Exodus, but it's also about 10W less than the 3080 Ti Founders Edition. FurMark is more prone to kicking in varying levels of throttling, so while the Galax card used 20W less than the Zotac in that test, we have to look at clocks as well to determine what exactly was going on.

All of the 3080 Ti cards have the same 350W TDP, but the clocks and performance can be quite different. The Galax card averaged slightly lower clocks than the Zotac and Founders Edition cards in Metro Exodus, while in FurMark it landed between the Zotac and Founders Edition. The GPUs all exceed their stated boost clocks in Metro, but not by the same margins. The Founders Edition ran at 136 MHz over its nominal boost clock and the Zotac was 123 MHz above boost, but the Galax card only ran about 80 MHz higher than its boost clock.

We had a problem collecting fan speeds on the 3080 Ti and 3070 Ti cards initially, but we've since sorted that out. Unfortunately, the Zotac card has moved on to other pastures (it will now be used for case testing), so it's not part of the fan speed results. Anyway, the temperature and fan speed results go hand in hand: Higher fan speeds yield lower temperatures, and vice versa. The Zotac and Galax cards are both substantially larger than the Founders Edition, and end up with similar temperatures, with lower fan speeds as well. In Metro, the Galax fans ran about 350 rpm lower than the Founders Edition, while the gap was 800 rpm in Furmark — though with lower clocks on the Galax card, so there are other factors at play.

The other part of the equation is VRAM temperatures. The larger coolers on the Galax and Zotac cards reduced temperatures a few degrees overall, but more importantly, peak GDDR6X temperatures were in the 90–92C range on the Galax card. The Zotac by comparison hovered at around 96C–98C, and the Founders Edition consistently hit 100C–104C.

We noise tested all three RTX 3080 Ti cards in the same test PC, with the SPL meter sitting about 10cm from the GPU fans. (Note that the testing environment has changed, so noise comparisons aren't necessarily identical to previous testing.) This was done to focus on the sound coming from just the GPU, rather than picking up other noise sources like that of the CPU cooler and case fans.

The Galax RTX 3080 Ti SG while running Metro Exodus settled down at 48.5 dB, while the Founders Edition ran at 52.0 dB (the Zotac, as already noted, was unavailable for retesting). Fan speeds for the two cards were 55% on the Galax and 66% on the Founders Edition — as we noted in the initial GeForce RTX 3080 Ti review, the Founders Edition definitely seems to struggle a bit with balancing cooling and fan speeds. As another point of reference, we used MSI Afterburner to set the fan speed to 75% on the cards, which is perhaps similar to what you'd need if you were doing 24/7 cryptomining. That gave noise levels of 56.3 dB on the Galax and 56.5 dB on the Founders Edition — though the Galax temperatures, particularly on the GDDR6X, would be lower based on our testing. 

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

Finding any GeForce RTX 3080 Ti card in stock, at least in the U.S., continues to be an exercise in futility — unless you want to pay eBay prices, in which case you can find RTX 3080 Ti cards for around $1,800. Yeah, thanks but no thanks. At present, there's only one Galax card listed on eBay, and it ships from overseas for $2,200, which we hopefully don't need to point out is an even worse deal. You can try to find a card for sale on the Galax Store, but right now the RTX 3080 Ti doesn't even show up as an option there. U.K. readers might have better luck, though, as there are currently three Palit / KFA2 RTX 3080 Ti cards in stock at Scan — at higher-than-MSRP prices, naturally.

If you're in the market for an RTX 3080 Ti, can find the Galax card in stock, and are willing to pay the high asking price, the Galax card works quite well and might even be able to justify its MSRP of $1,500. That's a lot of qualifiers, and at that price we'd be more inclined to shop around for an RTX 3090 — or, you know, just give the whole Ampere family a pass and wait for Lovelace, and pray that pricing and availability will be better by then.

As we said in the Zotac RTX 3080 Ti review, it's unfortunately still difficult to find any of the top graphics cards in stock, even at inflated prices. It's not clear whether the cards are still going into the hands of miners (unlikely, given the hash rate limiter), or if there simply aren't enough to meet demand. It's probably the latter, and given the widespread reports of continuing shortages — even Intel now says it expects to not be able to manufacture as many chips as it would like this quarter — that's not likely to change until 2022.

For PC gamers, it's been a terrible year for graphics card upgrades. Console gamers aren't doing that hot either, with PS5 and Xbox Series X still routinely out of stock and overpriced, but at least Sony says it has now sold over 10 million PlayStation 5 consoles, and Microsoft says the latest Xbox is its fastest selling console ever — which means maybe 5.6 million Xbox Series S or X consoles sold since launch. We don't know for sure how many RTX 30-series cards have been sold, but it's almost certainly far less than 10 million, across all models.

The one thing Galax appears to get right with the GeForce RTX 3080 Ti SG is memory cooling. We haven't tested that many other 3080 Ti cards, but the Galax card kept the GDDR6X memory running at less than 100C, regardless of what we threw at it. That might not sound like much, but it's better than 110C while mining on the Founders Edition. Meanwhile, we recommend gamers continue to wait for GPU prices to drop and inventory to improve. What goes up must come down, eventually.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

The Nvidia RTX 3060 brings a new level of performance to the mainstream market--sort of. Officially, the RTX 3060 launches today with prices starting at just $329. Realistically? You're as likely to find one at that price as you are to find an RTX 3060 Ti at $399, RTX 3070 at $499, or RTX 3080 at $699 — not entirely impossible, perhaps, but highly unlikely. Nvidia's Ampere architecture now powers many of the best graphics cards, and they're all seeing massive levels of demand from both gamers and cryptocurrency miners. Nvidia has added firmware and driver code to detect Ethereum mining, which should help a bit, but when people are willing to pay extreme scalper pricing on eBay, even for cards like the GTX 1660 Super and RTX 2060, everything in our GPU benchmarks hierarchy is pretty much sold out right now. Nvidia is even working with partners to bring back previous generation Turing and Pascal cards.

None of that makes this a bad GPU, but we expect the RTX 3060 to be just as difficult to acquire as any other modern GPU. Eventually, the current Ethereum mining boom will fade away, but it could take a year or more before we see the end of chip shortages. That shouldn't surprise anyone at this point, but if you've been hoping for a reasonably priced gaming PC upgrade, it's a depressing state of affairs. 

Unlike the previous Ampere GPUs, Nvidia won't offer an RTX 3060 Founders Edition, so we're looking at a third-party card. Nvidia shipped us the EVGA GeForce RTX 3060 XC for this launch review, a reasonably compact and relatively unassuming card. There's no metal (or even plastic) backplate, no RGB lighting, and two custom-sized 87mm fans for cooling with a 2.0-slot form factor. The card measures 202x110x38mm and weighs 653g, which is quite the change of pace compared to the other third-party Ampere cards we've reviewed so far.

There are reasons for that, of course. Creating a mainstream card and decking it out with all the bells and whistles costs money. And we think most gamers shopping for a good value are far better served by modest designs with good performance. There will certainly be extreme variants of the RTX 3060, and some of them will be priced higher than the budget RTX 3060 Ti options. Let's be clear: Even the fastest RTX 3060 won't beat a 3060 Ti in most situations — yes, even with 12GB VRAM. That's because memory capacity isn't a huge factor once you go above 8GB, and having more memory bandwidth, thanks to its wider memory bus, gives the 3060 Ti a big advantage. Also, the 3060 Ti has 35% more GPU cores.

Here's how things break down, comparing the RTX 3060 with its closest Ampere sibling and Turing predecessors. The RTX 2060 and 2060 Super show how much things have changed for the -60 suffix cards between Turing and Ampere. Ampere gives you a lot more shader cores, which means potentially much higher computational performance, and a minor improvement in memory bandwidth for the 12GB card. It also doubles VRAM capacity (at least until the anticipated RTX 3060 6GB shows up, though perhaps maybe Nvidia will just leave that for the RTX 3050 line) and boasts improvements in the RT and Tensor cores, as well as the memory subsystem, all leading to better performance. Power use remains similar, with a 170W TGP (Total Graphics Power), a decent step down from the RTX 3060 Ti's 220W TGP.

One interesting tidbit is that this is the first time Nvidia has used 15Gbps GDDR6 memory. The RTX 20-series cards all used 14Gbps memory, except for the RTX 2080 Super that came equipped with 15.5Gbps VRAM. That narrows the bandwidth gap between the 3060 and 3060 Ti a bit, though the extra 64-bits of interface width still gives the GA104 cards a clear advantage. And GA106 doesn't have an advantage is in ROPs, Render Outputs, as it only has 48 — the same as the RTX 2060.

However, the differences between Turing and Ampere GPUs don't always show up in specs tables like the above. Theoretically, the RTX 3060 has up to 95% more FP32 performance and 97% more FP16 Tensor core performance than the RTX 2060. In practice, the actual performance difference is much less, as half of the FP32 pipelines share processing resources with INT32 pipelines. The 3060 shouldn't ever be slower for gaming purposes, but most of the time, it will only be around 20-25 percent faster.

This is the first desktop card to use Nvidia's GA106 processor. At a high level, there are three GPCs (Graphics Processing Clusters), each with up to 10 SMs and 16 ROPs (the two blocks of eight blue rectangles each at the bottom of the GPC). The full chip has 30 SMs while the 3060 disables two and ends up with 28 SMs, but everything else is left alone. (Note that the mobile RTX 3060 has all 30 SMs enabled, though it only comes with 6GB of memory, which is also clocked lower than on the desktop card.)

Each SM contains 64 dedicated FP32 CUDA cores, plus 64 more FP32+INT32 CUDA cores — only FP32 or INT32 can be used for each cycle. The SMs also contain one second-gen RT core and four third-gen Tensor cores, each of which is up to twice the performance as the previous generation cores, and with sparsity the Tensor cores are potentially four times as fast as on Turing. Finally, there are six 32-bit memory interfaces, each one linked to a single 8Gb or 16Gb GDDR6 module — the latter is reserved for desktops at present, with the 8Gb modules used on laptops.

The full GA106 chip has 12 billion transistors, down from 17.4 billion in GA104. That shrinks the die size from 393mm square to just 276mm square, which not only helps to reduce the cost of the chip, but also increases the number of chips Nvidia can get from a single wafer — and if you're wondering, GA106 is less than half the size of GA102, which measures 628.4mm square and has 28.3 billion transistors. At an estimate, Nvidia can get around 130 dies per wafer with GA104 (some of which are defective, most of which end up as partially disabled chips), while the smaller size of GA106 allows for around 200 dies per wafer. More dies mean better yields and more graphics cards to go around. That's the hope.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

As mentioned above, Nvidia isn't doing a Founders Edition, AKA reference card, for the RTX 3060. That leaves us with third party cards, and Nvidia did ship us an RTX 3060 sporting a reference boost clock of 1777 MHz. This is as close as we'll get to a baseline level of performance for the RTX 3060, in other words, and factory overclocked cards should be numerous and should all perform at least as well as the EVGA RTX 3060 XC. Let's go ahead and quickly dissect the card to see what makes it tick.

There are eight screws on the back of the card holding the heatsink and shroud in place. Four of those surround the GPU, with the other four near the back of the card. As usual, one screw is covered by an EVGA sticker to detect tampering. Remove those, unplug the two fan connections near the bottom-left area of the card (when looking at the back), and with a little bit of effort, you can pull it apart.

The PCB is interesting, as it looks to be the same as some RTX 3060 Ti and RTX 3070 models. There are clearly two memory locations that go unused, except we're pretty confident that Nvidia only has a 192-bit interface on the GA106 chip, so we don't expect to see 8GB or 16GB graphics cards using GA106.

The six GDDR6 chips are from Samsung, model K4ZAF325BM-HC16, and they're rated for 16Gbps operation. Not surprisingly, we could push the memory well beyond the stock 15Gbps clocks (see below). The memory is also actively cooled via thermal pads connecting it with the heatsink. And as you'd expect with a 170W card, the power delivery system isn't nearly as robust as what we've seen on RTX 3080 and 3090 cards, because it doesn't need to be.

EVGA GeForce RTX 3060 XC Overclocking 

While this particular card doesn't come with a factory overclock, we did some manual overclocking to see what we might expect from higher-tier custom cards. Spoiler: You're not going to come anywhere near RTX 3060 Ti levels of performance. That's because the 3060 Ti has 35% more processing cores and 24% more memory bandwidth. We can make up some of the difference, but overclocking on graphics cards usually gets around 10% more performance at best.

After testing and tuning, we ended up with a maximum stable overclock of +200 MHz on the GPU core, and +1000 MHz (17Gbps effective speed) on the memory. We also pushed the power limit up by 11%, which provides at least some of the performance gains, and typical clocks during testing were in the 2.0–2.1GHz range. The cooling solution on the EVGA card is decent but not extravagant, so fan speeds ended up at around 80% with these settings, but GPU temperatures were fine with a maximum of 69C. Unfortunately, there aren't any temperature sensors for the GDDR6 memory, so we don't know if the chips were getting hot or not. But GDDR6 temperatures are usually quite a bit lower than GDDR6X.

We'll include the results for our overclocked settings in the charts below. On average, the overclock improved performance by 8% at 1080p, 9% at 1440p, and 10% at 4K. That's pretty much par for the course when it comes to GPU overclocking.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

Our gaming benchmarks are all run using the same Core i9-9900K PC, which includes various high-end components that you can see in the boxout to the right. We enable the XMP memory profile, but the CPU runs at stock speeds, which basically means 4.7GHz for multi-threaded workloads. We're using the same 13 game test suite as in recent reviews, with all games tested at ultra / highest / maximum settings (the label varies by game, but basically close to max settings without things like supersampling enabled). Each resolution was tested multiple times, discarding the first run and checking for consistency of performance.

As a mainstream card, the RTX 3060 primarily targets 1080p and 1440p gaming. Some lighter games may also run fine at 4K, or in some cases, you could shoot for 4K at medium settings. But despite having more VRAM than even the RTX 3080, frame rates definitely take a hit at the highest resolutions. We'll start with our 1080p benchmarks, then move on to 1440p and 4K, but we won't spend a lot of time on that last one as it's mostly academic.

The RTX 3060 sits near the bottom of our charts, as expected considering the remaining GPUs are all higher-priced and higher-performance models. We've omitted some of the top-tier options like the 3090 and 6900 XT, as they don't remotely compete for the same audience, as well as some previous generation cards like the 2080 Super and 2080 Ti — again, different targets, plus those overlap with the modern 3060 Ti and 3070. We also left off all of AMD's earlier GPUs, since none of them can run the two games we tested with ray tracing enabled (Dirt 5 and Watch Dogs Legion) — not that you really need or should plan on using ray tracing with a mainstream GPU.

Overall, the 3060 ended up 20% slower than the RTX 3060 Ti, and 27% slower than the RTX 3070. That's a bigger drop as you go down the GPU hierarchy, giving you proportionally less performance for your money — not a good thing. The RTX 3060 is also just 20% faster than the RTX 2060, and 7% faster than the RTX 2060 Super. Of course, 1080p is a low hurdle to clear and CPU bottlenecks come into play. Let's quickly go through the individual 1080p game charts.

Of the 13 games tested, the largest differences in performance between the 3060 and 3060 Ti are around 25% — as in, the 3060 is 25% slower than the 3060 Ti, or if you prefer, the 3060 Ti is 32–35% faster than the 3060. Borderlands 3, The Division 2, Dirt 5, Metro Exodus, Strange Brigade, and Watch Dogs Legion all fall in that category, with a spread of 24–26%. That difference corresponds directly with the difference in shader computational performance. Other games show less of a gap, mostly due to CPU bottlenecks, with Far Cry 5 (12% slower) being the closest result.

The story doesn't change much when looking at other GPUs. The 3060 delivered a 43% advantage over the RTX 2060 in Watch Dogs Legion, but that's probably due to the increased memory requirements of ray tracing. In most other games, it was only 20–30% faster, and in Final Fantasy XIV it was only 6% faster. Against the RTX 2060 Super, the gap narrows considerably, with a tie in FFXIV and leads ranging from 3–13% elsewhere. So, about 18 months after the 2060 Super launched, Nvidia has a card that's theoretically 5–10% faster and costs 18% less. That's definitely progress, but it's not a massive improvement.

Incidentally, we don't have the full suite of results for older cards like the GTX 1060 6GB (again, due to including some ray tracing games), but in our limited test suite (which excludes Assassin's Creed, Dirt 5, Horizon Zero Dawn, and Watch Dogs Legion), the 3060 is over double the performance. It's also basically tied overall with the RX 5700 XT, but with ray tracing support and with lower power use.

Increasing the resolution to 1440p doesn't radically alter the overall relative standing of the RTX 3060. Actual performance drops by 25%, and it's now 22% slower than the 3060 Ti on average, but it's also 24% faster than the RTX 2060 and 8% faster than the 2060 Super. Factory overclocked cards will do a bit better, but even though the card technically has 3584 GPU cores, compared to Turing it behaves more like a card with about 35% fewer cores (due to the shared FP32/INT32 cores).

About half of our test suite still averages over 60 fps, but Assassin's Creed Valhalla, Borderlands 3, Dirt 5, Metro Exodus, Red Dead Redemption 2, and of course Watch Dogs Legion (with ray tracing enabled) all come up short. Actually, even if we enabled DLSS Quality mode in Watch Dogs Legion, it can't get to 60 fps at ultra quality with ray tracing enabled — even at 1080p.

That's a good example of how far the RTX 3060 can go. If you're running at medium to high settings and with DLSS, ray tracing at 1080p and maybe 1440p is viable. Maxed out quality in many of the games that support ray tracing is going to prove too much for the GPU. It's slightly faster than the RTX 2060 Super, but that card was also at the bottom end of what we'd recommend if you want to run with all the eye candy enabled.

4K gaming proved to be a bit much for the RTX 3060, at least if you're hoping for 60 fps gameplay. Some games can get there, especially if you're willing to drop the settings a notch or two, but at our Ultra settings only two games in our suite managed smooth framerates: Forza Horizon 4 and Strange Brigade — neither of which are particularly demanding. A few others came close to 60 and should be fine with a G-Sync compatible monitor, but several games struggled to stay above 30 fps, and both games with ray tracing enabled (but no DLSS) dropped well below 30.

The 3060 fell 23% behind the 3060 Ti, but it also beat the 2060 by 32% — a bit of a pyrrhic victory, considering the actual fps, but having double the VRAM certainly helped. If you're hoping to game at 4K native, we recommend stepping up to a more potent graphics card. Or you could do what the latest consoles do and shoot for 30 fps with dynamic resolution scaling. Basically, the RTX 3060 looks to be roughly the same level of performance as the PS5 GPU.

GeForce RTX 3060 Mining Performance 

Nvidia made a big deal about crippling the RTX 3060's mining performance. That doesn't mean you can't mine on the card, however; it's just not as profitable as it otherwise could have been. We did a short test of mining performance, using NiceHashMiner, to see how the card managed with a variety of different mining algorithms. Here are the results.

The good news is that DaggerHashimoto, aka Ethash, which is the algorithm used by Ethereum, was definitely a lot lower than expected. It would start at 45-50MH/s and quickly drop to about half that speed. This happened with every piece of miner software we tried: Excavator, NBMiner, PhoenixMiner, and T-Rex. Of course, that doesn't cover every possible miner, but it's a fair start.

The bad news is that there are a lot of other mining algorithms, and in some of these — like Octopus — the RTX 3060 is still reasonably profitable. With the recent drop in Ethereum pricing, and the increased difficulty, at 45-50MH/s the RTX 3060 would have netted around $5.50 per day. In Octopus, it can still do around $4 per day, which means it could hit the break-even point in just 82 days at current rates.

Obviously, the rates are prone to wild swings, but given the potential profits, there's no way miners don't buy these cards. Plus, there's no guarantee that some future miner update doesn't figure out a way around Nvidia's driver and firmware protections, and miners are probably counting on that. It would technically be more profitable to buy a different GPU — even the RTX 2060 can beat the 3060 in mining profitability right now — but the current shortages mean that miners will try to buy up any reasonably priced graphics card.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

Nvidia rates the GeForce RTX 3060 TGP (Total Graphics Power) at 170W, making it the lowest power Ampere GPU by 50W. It has a single 8-pin power connector, and combined with the x16 PCIe slot that's enough for up to 225W of power — more than sufficient headroom for overclocking. We use Powenetics in-line power monitoring hardware and software to collect the real power use of the graphics card, which is relatively close to what software reports for Ampere GPUs. Powenetics also links up with GPU-Z to record the GPU temperature, fan speed, and GPU clock speed.

You can read more about our approach to GPU power testing, but the takeaway is that we're not dependent on AMD, Nvidia or any other GPU vendor to accurately report how much power a GPU uses. We run FurMark as a worst-case stress test, and we also run five loops of the Metro Exodus benchmark at 1440p ultra (without ray tracing or DLSS) for the following charts. Because Metro loops every two minutes or so, you'll see a saw pattern, but maximum power and temperature are pretty consistent.

On average, the EVGA RTX 3060 used 171W in our Metro Exodus test, matching up pretty much exactly with the TGP. FurMark reached a slightly higher power draw of 175W, but that's not particularly worrisome. Our manual overclock bumped that up to 182W in Metro and 193W in FurMark, which is pretty much right in line with the power limit increase for the latter. For gaming purposes, it looks like something else was the limiting factor, at least on the EVGA card.

Temperatures were the same for both our stock and overclocked tests, coming in at 64C on the GPU core in Metro and just one degree higher in FurMark. There's not a huge spread in thermals, mostly because the fan speeds automatically adjust to help keep the GPUs cool, which is where we start to see a lot of separation.

The EVGA RTX 3060 ended up with some of the highest fan speeds we've seen on recent cards. Granted, we've been looking at much larger cards, often with triple fan cooling solutions, but the fan RPMs here are higher than any of the other reference cards we've looked at. Part of that is likely due to the use of old-style fans, rather than the new higher static pressure models with an integrated rim — and the fan selection was likely done to keep costs as low as possible.

That makes EVGA RTX 3060 a less-than-stellar example of a quiet graphics card. It's not horribly loud, but neither is it silent. We measured noise from 15cm away using an SPL meter, with dB(A) weighting and a noise floor of 34 dB. The card's fans turn off at idle (provided the GPU is below 50C), but after gaming for a few minutes the fan speed ramped up to 1940 RPM, or about 71% of maximum. That resulted in 43.1 dB(A) of noise for gaming, and FurMark took that slightly higher to 72% and 43.9 dB(A). Overclocking naturally required even higher fan speeds, around 80% or 2350 RPM, with a very noticeable 49.3 dB result. If you're looking for a quiet RTX 3060, you'll be better off with a larger card that has a bigger heatsink and improved cooling.

Last, we have GPU clocks. With a boost clock of 1777 MHz for the reference cards, the RTX 3060 is the highest clocked Nvidia GPU from the Ampere family so far. As usual, the real-world clocks are even higher than the boost clocks (except in FurMark). The EVGA card averaged 1876MHz in the Metro test, and 1609MHz in FurMark. Overclocking allowed for average GPU clocks of 2085MHz. GA106 doesn't clock quite as high as AMD's Big Navi chips, but we expect we'll see factory overclocked models running at well over 2GHz.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

The launch of mainstream graphics cards always feels anticlimactic, particularly when it comes about half a year after the big splash made by the new halo parts. Fundamentally, there's nothing wrong with the RTX 3060, and it actually has a lot going for it. Performance is basically at the level of the RTX 2070 from 2018, which means it's also as fast as AMD's RX 5700 XT (give or take), plus it has ray tracing and DLSS. Power use is slightly lower, and the price has dropped by at least $170 — $270 if you include the RTX 2070 Founders Edition that launched at $599. It also includes 12GB of VRAM, 50 percent more than most of the 20-series cards, double the VRAM of the 2060, and more than the 3060 Ti, 3070, and 3080. These are all good things.

There are two main concerns. First, we think the $329 asking price is basically fantasy land right now. Remember how the RTX 2080 Ti launched at $1,199, but there were supposed to be $999 models, which pretty much never existed in any real quantity? This will be like that, only so much worse with the current market trends. Even if we forget about miners for a moment, it's clear there are a lot of other people wanting to buy mid-range and high-end graphics cards. There were GPU shortages back in September when the 3080 launched, and things have only gotten worse since then. Plus, let's be real: AMD's Ryzen 5000 parts are still in short supply, as are the best webcams, and it's very unlikely all of those are being bought up by miners. Miners do make things worse in the GPU arena, but Ampere and Big Navi would still be selling out even without their 'help.'

The other concern is with performance and features. If you look at the results, out of our current gaming suite, there were only a few instances where the 12GB VRAM had much of an impact (compared to the 8GB 3060 Ti and 2060 Super), and that was mostly at framerates that were already lower than most gamers would like. It's nice that Nvidia equipped the card with 12GB, but more VRAM can't make up for the reduction in memory bandwidth and GPU computational prowess. Slightly better performance than an RTX 2060 Super at a moderately lower price is the least we could expect, and that's pretty much what we got.

If you can't afford a higher performance graphics card, and you can find an RTX 3060 available for anything close to MSRP, in today's world, that's a great deal. Or wait until next year, and hopefully it will actually drop down to MSRP. Yeah, the prospect of waiting another 10 months to save some money doesn't thrill us either. Anyone that already has an RTX 20-series card, particularly one of the higher-end models, has been able to enjoy this level of performance for quite a while.

There's the other way of looking at things as well, which is that this is the current generation console equivalent. There are differences, of course, but an RTX 3060 should deliver pretty comparable performance to the Xbox Series X and PlayStation 5 — and even better performance in games that support DLSS. It's likely better for ray tracing as well, but it's far enough down the totem pole that we feel ray tracing shouldn't be high on your list of priorities if you're shopping for this card.

We can't predict the future, but all we have to do is look at current resellers and places like eBay to know that, no matter what Nvidia does to hinder mining performance, the GeForce RTX 3060 is going to sell out. You basically can't buy any RTX 30-series or 20-series card on Newegg right now at less than double the original launch prices. The same goes for AMD's RX 5000 and RX 6000 series cards, and even the GTX 16-series GPUs. If you can find an RTX 3060 on sale today for less than $400, grab it fast — heck, buying one for $500 would still be a better deal than the $800 RTX 2060 Supers we've seen floating around.

Sadly, this is likely to be yet another product that won't be able to meet the demand for it any time soon. We hope we're wrong, we really do. But hope doesn't get you a shiny new graphics card.

MORE: Best Graphics Cards

MORE: GPU Benchmarks and Hierarchy

MORE: All Graphics Content

Unlike regular mid-towers, ITX cases generally take the board’s compact dimensions and use that to the designer’s benefit to craft some really funky cases. But sometimes you just want a classic layout but in a smaller form factor. For that audience, Phanteks just announced the P200A, which is really just a fairly standard-layout chassis for ITX systems.

Indeed, the P200A isn’t the most exciting ITX case, but where it underwhelms in its layout, it excels in functional simplicity, RGB, and most importantly: cost. This chassis carries an MSRP of just $50 for the standard variant or $70 if you want the RGB bells and whistles. We have the latter on the test bench today, together with the optional riser cable, so let’s find out if it’s good enough for our Best PC Cases list.

Specifications 

Features 

Touring around the chassis, we spot that it’s made from painted steel, has a plastic front with mesh that’s made from perforated, painted steel, and a glass side panel with a light tint -- but that’s not all. Unlike almost every other case that came out in the last decade, the PSU sits at the top here, which is a little strange but not a deal-breaker. 

You’ll also spot an intake on the side of the chassis to aid with airflow, though no fans are mounted here from the factory.

Front IO, mounted at the bottom, consists of two USB 3.0 ports, a USB Type-C port, mic/headphone combo jack and two buttons to control the RGB mode and color – indeed, this RGB variant comes with an RGB controller built in, and it’s quite a good controller with smooth effects and plenty of mode and color options. 

The power button resides at the top of the chassis, and what’s nice about Phanteks’ approach is that it actually has two RGB LEDs for illumination. This means that the button is part of the RGB goodness, which is a nice touch on a budget case like this. There is no HDD activity LED, nor is there a reset switch, which makes connecting the IO pins on the inside a breeze. 

Inside the case you’ll find room for up to ATX power supplies, an ITX motherboard, large graphics cards, a 240mm radiator, and four 2.5-inch SSD mounts – two behind the motherboard and two on the side intake covers. There are no mounts for 3.5-inch drives, but you can buy optional brackets to install them – though in all fairness, most people probably aren’t going to use desktop drives in their next build anyway.  

The chassis comes with two 120mm RGB fans pre-installed at the front intake location.

So without further ado, lets install a system inside the P200A.

We are using the following system for today’s build: 

Step 1: CPU Block & Motherboard Installation 

We’re starting off with the CPU, memory, and SSD pre-installed onto the motherboard. We then chuck the CPU block of our AIO on the board, simply because it’s generally easier to do this beforehand in case the cutout at the back of the motherboard tray is too small to access the socket afterward.

One of the pre-installed standoffs has an anchor, meaning the board was held in place with the case upright before tightening it in – a little detail that makes life just a little easier.

Step 2: Finishing AIO installation 

Then, I installed the remainder of the all-in-one liquid cooler. I made sure to use the side vent as an intake, with the fans pushing air through the radiator. This would give the AIO access to cold fresh air for optimal CPU temperatures, which paired with the front intake should lead to impressive temperatures in the thermal tests.  

Step 3: PSU Installation 

After that, I installed the power supply at the top of the chassis, ensuring the fan was pointed down as there is no intake at the top of the case. I routed all the cables to the back of the chassis. You can slide it in from the inside of the chassis if the PSU isn’t too big, but if you’re having trouble, there’s also a mechanism to slide in from the back of the case. 

Step 4: Switching to Vertical GPU Layout 

As Phanteks supplied us with a GPU riser cable, I figured it would be fun to put it to use. To do this, I first swapped two plates at the back of the chassis around to switch the expansion slots to the vertical layout.  

Then, I installed the metal bracket that came with the chassis and the optional riser cable. I used position A, which sits a little deeper inside the case for better cooling, as otherwise the GPU gets too close to the glass panel. 

Step 5: Cable Management 

At this point, I could install the GPU, but it would sit in the way of all the cables I still had to manage. With RGB, an RGB controller, four fans and the rest of the system, there were quite a few cables to tidy up, so I figured it would be easier to do this before putting in the GPU. 

As far as cable management goes, the P200A is surprisingly good for an ITX chassis. Of course, it’s still an ITX case – by which I mean that it’s still small and you’ll have to strain a bit to get everything where you want it to be, but there are cutouts in all the right places, along with plenty of Velcro straps and brackets to hold everything in place. In total, I spent about 30 minutes on cable management, though this could have been less if I hadn’t removed the Velcro strap at the top PSU area, as it was difficult to get back into its anchor point.

Step 6: GPU Installation 

After all that, dropping our Founder’s Edition RTX 2070 Super was simple. Nothing more to be said here. 

Build Complete 

Despite being a $70 chassis, the Phanteks Eclipse P200A certainly doesn’t feel like it. Sure, the materials aren’t anything to write home about, but the design overall is quite well thought-out and it was a joy to build a system in this case. The end results were pretty, too.  

For thermal and acoustic testing, other than fan speeds, we’re running the hardware we have entirely at stock settings to mimic the results you can achieve as closely as possible.  

Fan Characteristics 

The two included fans are, despite the budget nature of the case and RGB implementation, actually quite good spinners. They have PWM support and are surprisingly efficient with regards to noise levels vs. cooling performance. They can spin at a minimum speed of just 400 RPM and will ramp up all the way to about 1400 RPM at full duty. Without an extra layer of filtration behind the front mesh, this means the chassis gets plenty of fresh air.   

Acoustic Results 

Acoustically, the Phanteks P200A offers very respectable results, achieving an optimized idle noise level of just 33.9 dBA, jumping up to 39.5 dBA with the case fans at full blast. Max out the 240mm AIO with its fans spinning at 2200 RPM, and the RTX 2070 Super FE at 75% and noise levels quickly become obtrusively loud, but that’s to be expected -- you shouldn’t run your system with fan speeds this high anyway. As with any mesh system, try to find parts that run quiet and preferably find a GPU with a Zero-RPM mode so that it’s quiet most of the time when you aren’t gaming. 

Thermal Results 

For the thermal tests, all case and CPU fan speeds are set to 100 percent, and the GPU fan is set to run at 75 percent fan speed.  

Thermally, the Phanteks Eclipse P200A is equally impressive. Funnily enough, the CPU and GPU temperatures were matched, with a reading of just 59 degrees Celcius when normalized to a 20-degree room. Remove the intake panels (front and side), and we witness a temperature decrease of only 2 degrees, which goes to show just how well optimized Phanteks’ mesh is -- it barely impedes airflow.

These figures are very impressive, and clearly there is plenty of thermal headroom to get those noise levels down a fair bit.

Conclusion 

Phanteks’ P200A is essentially a tiny version of their larger Eclipse P-series cases, and all things considered it’s a brilliant little chassis. Okay, it’s a little big by ITX terms, and the top-mounted PSU gives off major 2005 vibes, but it’s a very affordable chassis that offers excellent thermal performance paired with a good amount of quality RGB. 

It won’t blow you away with materials quality, nor is the P200A’s layout very exciting, but it’s well put together, easy to work in and has no major flaws that stop me from giving it a recommendation. If you’re on a budget and just looking for a simple case to house a small gaming system, and all you care about is good performance, a great price, and some attractive RGB, the Phanteks P200A is certainly one to shortlist. 

Whenever a new case comes out with a price point that’s below $100, I get a little excited. When its $25 below that, it gets me pretty excited. So just imagine my joy when Phanteks sent me the P360A, which carries an MSRP of just $70 but is available for a few dollars less if you shop around.

Add to that an absolutely brilliant, vibrant, and lavish RGB implementation with two fans, an LED strip and a genuinely effective controller. My day has been made, and if you’re in the market for a case for a budget-oriented system but still want it to look good, this case is for you.

Of course, clear shortcuts have been taken to bring this kind of feature set to the table at this price. The sheet metal is quite thin, the paint job on the interior doesn’t ooze quality like more expensive cases, and the case doesn’t even include basic essentials such as a power LED, HDD LED, or a reset switch. There’s no USB Type-C to be found here, either.

However, none of those things really matter at this price – when finished, a system in the P360A genuinely looks good, cools well, and it’s easy to assemble. In case you haven’t guessed yet, the Phanteks P360A is easily deserving of a spot on our Best PC Cases list as Best Budget ATX case. Let’s find out why, shall we?

Specifications

Features

Touring around the exterior of the Phanteks P360A, it is immediately clear that this is a budget case. The sheet metal and glass panel are thin and the latter doesn’t quite line up with at the back of the case. And the PSU shroud is external, allowing for a smaller glass panel. Regardless, these shortcuts don’t take much away from the final look of the product, as the paint job on the outside is nice. The mesh intake is also well-manufactured, and as you’ll see later, there is a very nice RGB strip along the bottom of the side panel.

Phanteks doesn’t bother with a power LED, HDD LED or reset switches – all of that costs money and is rarely important anyway, so why bother when the funds can go to RGB instead? Front IO is covered by two USB 3.0 ports and separate headphone and mic jacks. The power button is at the center, and the two buttons you see below the USB ports are the RGB mode and RGB color selectors.

Internal Layout

Inside the P360A there is space for up to an ATX motherboard and long graphics cards up to 400mm (15.75 inches), so you’ll be able to fit very powerful gaming systems in here. Storage options aren’t lavish, but then again that isn’t the focus of this chassis. There’s room for two 3.5-inch drives that slide in from the front, and two 2.5-inch drives behind the motherboard tray. You can buy an optional third tray to add a third 2.5-inch drive.

You’ll notice that on the interior, the paint job isn’t quite as nice as the outside, but it’s nothing to be concerned about.

The connectivity for the RGB is also found inside. The fans and LED strip all connect through Phanteks’ own 3-pin headers, but a ‘standard’ header is present to connect additional RGB devices not from Phanteks, and you can opt to connect the RGB to your motherboard and override the built-in controller. But in all fairness, what’s here is quite a good controller – there’s no need to shop for a board that has an A-RGB header, which can help you save some money.

Cooling

Another spot where Phanteks cut costs is in the filtration – or rather, lack thereof. Of course, the marketing lingo is all about how the front panel filters without the need of a filter, but we both know that’s to cut the cost of a real air filter. The front has a 1mm perforation, which should take care of the worst offenders, but stuff will still get through more than a dedicated filter. Of course, the lack of removable filtration will be good for cooling.

That being said, you have to evaluate these things on a per-use-case basis. If you’re like me, and you run your PC for 12+ hours a day, you obviously want to pass on the P360A because you’ll be cleaning out the interior far too often. But if you only play games on it for a couple hours a day or so, the amount of dust that accumulates in the system will be far lower, so won’t be as much of an issue. Fortunately, there is a filter at the top above the radiator exhaust to protect from falling dust and debris, which will protect the system from passive pollution. The power supply also has its own filter to spare cleanup jobs there.

However, cooling is another area where the P360A differs from the P300A. It comes with two RGB fans instead of one non-RGB spinner, and the top radiator mount supports up to 240mm radiators instead of only a 120mm unit – something many buyers may appreciate. Theoretically, the case should also be able to fit a 280mm cooler at the top, but you’ll undoubtedly run into the VRM coolers or tall memory with the vast majority of motherboards, so I’d recommend you stick with a 240mm AIO at the largest.

If you’re dead-set on a 280mm radiator though, you can mount it at the front, but I don’t see why you would want to get rid of the pretty fans that are included.

We are using the following system for our case test bed:

Building a system in the Phanteks P360A is simple. All the standoffs come pre-installed, so the motherboard drops right in -- the GPU too -- and there was even space for our big Corsair HX750 power supply. Space is a little tight around the top of the chassis, but as long as you have someone there that can help you with the connectors in tight spots it shouldn’t be an issue.

Cable Management

Surprisingly, cable management is also incredibly simple in the P360A. A simple shroud keeps things looking tidy in the main motherboard compartment, and gives a nice cable routing channel on the other side of the chassis. Two Velcro strips help to keep everything in place.

You’ll want to connect the two fans to your motherboard and use a SATA power connector to power the RGB controller. There’s no HDD LED, power LED, or reset switch headers to fumble around with as this case doesn’t have these, but I doubt hardly anyone will miss them.

If there was something I’d have to complain about, it would be the lack of cutouts around the bottom of the motherboard. But of course, because the PSU shroud isn’t internal and part of the case’s structure, Phanteks couldn’t cut out so much or it would compromise the case’s structural rigidity.

Build Complete

Isn’t she pretty? For a $65 case, I’m honestly impressed with how this system turned out. Click through the pictures to see a few of the different RGB modes.

For thermal and acoustic testing, we are using the following software and settings:

Test Results

The fans Phanteks includes are two 120mm RGB spinners, and they run at speeds between about 350 and 1350 RPM. This is an excellent range, allowing you to run the system very quietly at the low end and with plenty of cooling at the top. The fans are voltage controlled.

Acoustic Results

For our acoustic tests, we run three scenarios: CPU full load, CPU and GPU full load, and an optimized idle. The CPU Full Load test runs the CPU and case fans at their maximum speed. For the CPU and GPU full load acoustic test we add the Nvidia RTX 2070 Super FE at 75% fan speed, because in practice it never runs at 100 percent and is far too loud when it does.

For the optimized idle, we run the GPU fan speed at 40 percent (the 2070 Super FE GPU does not have a Zero-RPM mode), and run the CPU and included case fans at the lowest speed they will spin at.

Comparing the chassis to similar cases in or near its class, the Phanteks P360A performs in the middle as far as acoustics are concerned. At full load, it’s not a very loud case, and with the fans at full idle, the GPU produces more noise than the case’s fans do, which should give you a good idea of what kind of noise levels to expect – quiet.

Thermal Results

For the thermal tests, all case and CPU fan speeds are set to 100 percent. The i9-9900K is pegged at a 4.6GHz clock at 1.1v on all cores to ensure consistent power consumption across test scenarios, and letting the GPU run at 75 percent fan speed enables it to maintain its power target while maintaining one set reasonable fan speed, so that the temperature is the only variable.

Thermally, the Phanteks P360A pleasantly surprised us, jumping straight to the top of our leaderboards. It performs nearly identical to the Lian Li Lancool II Mesh, which we awarded the best ATX case for $100.

Considering the P360A has one less fan, is quieter overall, has RGB, and is cheaper, this is a very impressive performance indeed. Obviously, it doesn’t have proper air filtration, so it’s clear where its performance benefit is coming from, but it’s impressive nevertheless.

Bottom Line

Phanteks’ P360A is one of those budget ATX cases that just brings a smile to your face. Of course, it’s not going to blow you away with quality, but it has all the basic features needed for simple ATX gaming systems. While it takes clear shortcuts to reach its low street price of about $66, all of its negatives are minor points that don’t really matter in the grand scheme of things and can easily be forgiven at this price.

The only major concern with this chassis is the lack of proper filtration on the intake. The 1mm mesh front panel will surely catch the worst offenders, but if you’re someone that works on their PC and runs it for 8+ hours a day, it may be worth looking into a chassis with better filtration to spare yourself the regular internal cleanup job.

But if you’re only building a gaming PC in here that’s going to run a couple of hours a day, the P360A will offer a lovely foundation to build a system into. You can add an extra fan to the back to complete the look, along with an AIO, but as shown by the thermal results, you clearly don’t need to. This really is a case to just throw some parts at and call it a day.

Featuring all the basic essentials, excellent thermals and packing an awesome, software-free digital-RGB controller with two RGB fans and one stylish LED strip, all at $66, I have no choice but to make this the new Budget ATX Gaming case top pick and chuck it onto our Best PC Cases List. If you don’t want to spend a lot, like RGB, and won’t be running your system all day and into the night, the P360A is highly recommended.

AMD launched the Radeon RX 6800 XT and RX 6800 on November 23, 2020. The first of the new RDNA2 architecture graphics cards had plenty to offer, ranking near the top of our GPU benchmarks hierarchy and earning a place on our list of the best graphics cards. AMD does particularly well with games that don't support ray tracing. In such cases, there are quite a few games where the 6800 XT leads the (theoretically) more expensive RTX 3080, though enabling ray tracing or DLSS quickly turns the tables. The biggest problem, as we've seen with all of the recent GPU launches, is actually finding one in stock. Now we're looking at three third-party custom cards, from ASRock, Asus, and Sapphire, to see what they bring to the table.

The core features and RDNA2 architecture are all unchanged, so the main differences between the cards will be in clock speeds, cooler designs, and aesthetics. There are also a few third-party add-ons, in the way of software, that might sway your purchasing decision. But let's be real: Finding any of these cards in stock can be an exercise in futility, and with the recent surge in cryptocurrency mining, it could be months before supply is anywhere close to matching demand. In other words, if you want an RX 6800 XT as soon as possible, the brand and model of card will be far less of a consideration than whatever you can actually lay your grubby little mitts on.

The good news is that performance across all of the tested RX 6800 XT cards is very close. At factory stock settings, the speediest of the cards we've tested is only 2-3 percent faster than the reference RX 6800 XT. Between the three custom cards, the performance deltas are even smaller, to the point of being effectively non-existent. But that doesn't mean the cards are all equal, as the cooling designs and other elements come into play. Here's a quick overview of the specs before we get into the individual card analysis and benchmark results. 

Let's first address the elephant in the room: The pricing is either fantasy land or, in the case of actual 'street' pricing, egregious. Theoretically, the Sapphire Nitro+ initially launched at $769, but of course, it was sold out — just like every other GPU. Newegg currently lists it at $999, and it’s still out of stock. The Asus Strix LC is a similar story, with a launch price of $899 but a current Asus store price of $1,080, and it's also sold out. ASRock gave a launch price of $829, but retail prices are much higher than that, and naturally, the reference AMD RX 6800 XT can't be had for anything close to $649.

Beyond price, the only difference in specs is the TDP. Sapphire lists 350W, while Asus and ASRock don't give any value. We put in 350W with a question mark based on our testing. All three AIB cards have the same 2360 MHz Boost Clock, which they can exceed in some cases. That's where the cooling solutions come into play.

That's it for the introduction. Let's get to the individual cards, and we'll dig into the finer points of each one, including any extra features that can help it stand out. We're going to dispense with actual scores on these cards, mostly because they all feel like ghost launches. Yes, they technically went on sale, but both pricing and availability are so limited that we don't know where they'll really land. They're all more or less equal, depending on your wants and needs.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

ASRock's Taichi brand is generally used for the company’s top-of-the-line products, and the RX 6800 XT Taichi is no exception. It's big and bold, with one of the largest coolers we've seen on a third party card, plus a copious amount of bling. Meanwhile, performance is as you'd expect: right in line with the other top-tier solutions from competing cards.

The 6800 XT Taichi is a massive card, dwarfing the reference model 6800 XT and tipping the scales at 1.75kg (3.85 pounds). ASRock actually lists the weight at 1815g, but my scale disagreed by about 65g. The dimensions are 330x140x56mm, and the cooler occupies 2.8-slots, which in today's single GPU market isn't much of a problem. The Taichi is also one of the longest cards we've seen — 13 inches long — so you'll definitely need a spacious case if you want this card to fit.

For a while, most high-end GPUs tried to stay close to a 2-slot thickness so that you could add a second (or even third) card for CrossFire or SLI, but multi-GPU support in games has seriously declined in recent years, so it's now less of a consideration. The most common use case for multi-GPU these days is cryptocurrency mining, but since coin miners just build custom mining chassis with PCIe extension cables, size isn't much of a factor there either. That means manufacturers are more willing to create cards that effectively block off the two expansion slots adjacent to the GPU.

ASRock's Taichi brand is generally used for the top-of-the-line products from the company, and the RX 6800 XT Taichi is no exception. It's big and bold, with one of the largest coolers we've seen on a third party card, plus a copious amount of bling. Performance meanwhile is as you'd expect: right in line with the other top-tier solutions from competing cards.

The 6800 XT Taichi is a massive card, dwarfing the reference model 6800 XT and tipping the scales at 1.75kg (3.85 pounds). ASRock actually lists the weight at 1815g, but my scale disagreed by about 65g. The dimensions are 330x140x56mm and the cooler occupies 2.8-slots, which in today's single GPU market isn't much of a problem. The Taichi is also one of the longest cards we've seen — 13 inches long — so you'll definitely need a spacious case if you want this card to fit.

For a while, most high-end GPUs tried to stay close to a 2-slot thickness so that you could add a second (or even third) card for CrossFire or SLI, but multi-GPU support in games has seriously declined in recent years and so it's now less of a consideration. The most common use case for multi-GPU these days is cryptocurrency mining, but sice coin miners just build custom mining chassis with PCIe extension cables, size isn't much of a factor there either. That means manufacturers are more willing to create cards that effectively block off the two expansion slots adjacent to the GPU.

ASRock provides plenty of RGB lighting, with the center fan lighting up along with the top Taichi logo and the surrounding light strip, and there's another Taichi icon on the back of the card as a final RGB option. I sort of wish the company had gone whole hog and used RGB on the other two fans, but that would make for a very bright card. Also of note is that in a traditional PC case, the 'front' fans on the graphics card will often end up facing the bottom of the case. That means you won't even see the fans unless you use a case that supports a vertically mounted GPU (which would also need to be able to handle a triple slot thickness, or you'd obstruct the fan intakes).

There are some Taichi elements that haven't been updated to match the competitive landscape, and here I'm looking specifically at the fans. Most of the latest generation RTX 30-series and RX 6000-series GPUs now have fans with an integrated 'barrier rim' around the outside of the fan blades. This ring helps improve static pressure and airflow, improving cooling capabilities while potentially reducing fan noise. Meanwhile, ASRock has traditional fan blades like what we've seen for many years. It's a small thing, but we'd like to see the latest technology utilized on a premium card.

ASRock doesn't specify a TDP for the Taichi, though it does recommend a PSU wattage of 800W or more. The card also requires three 8-pin PEG power connectors, which in theory can deliver 450W of power, giving the card a peak power delivery of 525W when combined with the 75W of the x16 PCIe slot. (It didn't come close to hitting that mark, though maybe an adventurous overclocker with LN2 could do so.) In our testing, the Taichi used a bit less power than the other custom cards. It averaged 332W in Metro Exodus (still 10 percent more than the reference 6800 XT) and 352W in FurMark. Overclocking pushed power use up to 344W in Metro and 400W in FurMark. That means there's still potential for higher overclocks, but we generally seem to hit a similar limit with all of the 6800 XT cards of around 2.5GHz.

Digging into the overclocking specifics, for now, we're still somewhat restricted in what utilities we can use on RX 6000 cards, as MSI Afterburner doesn't fully support the new GPUs yet. AMD's own Radeon Software seems the best option, and we used it to increase the power limit by 15 percent, set the GPU to a maximum clock of 2580 MHz, and bumped the GDDR6 speed up by 150 MHz (to 17.2 Gbps). Combined with a more aggressive fan curve, we reached a stable OC where clocks in Metro Exodus averaged 2529 MHz, compared to 2391 MHz at factory stock settings. That's a 6 percent increase in GPU clocks and 7.5 percent on VRAM clocks, which improved performance in our test suite by 4-5 percent overall. 

ASRock offers its Tweak software for tuning, which is sufficient for modest overclocking but doesn't really do much more than AMD's own Radeon Settings. You can see the default power limits (289W, or 317W in OC mode), but that's just for the GPU power — the VRAM, VRMs, and other components will also use power. Ultimately, we didn't see much reason to install the custom ASRock Tweak software, since the core functionality is already present with AMD's drivers.

Technically, the ASRock Taichi ended up as the slowest of the three custom cards, but that's very much splitting hairs. It’s within 0.5 percent of the other two cards at stock, and it's within 1.5 percent of the faster Asus card when overclocked. Performance when overclocked is also basically tied with the stock RX 6900 XT. It's also up to 2 percent faster than the reference 6800 XT.

Frankly, performance and even pricing are sort of a non-issue for now, as none of the cards are readily available for purchase. If you want a 6800 XT and can find the Taichi in stock somewhere, and you're willing to pay $900 or more for it, be our guest. Our general advice is to wait for supply to improve and hopefully for prices to end up closer to MSRP. With the recent tariffs on graphics cards further impacting pricing, however, it could be a long and painful wait.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Asus has traditionally stayed away from the AIO graphics card market, but that changes with the ROG Strix LC. Not only does it come with an AIO (all-in-one) liquid cooler, but it includes a relatively large 240mm radiator. Some people will love that option, but it comes at a higher price and can be cumbersome to install — for someone like me that swaps GPUs regularly, AIO cards are more trouble than they're worth.

The external radiator adds a lot of bulk to the package, with the card plus radiator weighing in at 2166g. That's about as much as the RTX 3090 Founders Edition, but the good news is the x16 motherboard slot only has to deal with a 1340g card, with the rest of the bulk secured to the PC chassis. Thanks to the large radiator, the Strix LC can also get by with a traditional 2-slot thickness and a single blower fan on the main card, and the blower doesn't really have to work that hard.

The main graphics card measures 277x131x43.6mm, but the radiator is an additional 276x120x51.7mm. Make sure your case has a good mounting location before taking the plunge. If you're also using a 240mm or larger AIO for CPU cooling, speaking from experience, you'll definitely want to verify both coolers will fit properly before buying.

The ROG Strix line is known for being Asus's top offering, and the Strix LC takes that one step further. Besides the RGB lighting on the graphics card, the two radiator fans also have RGB lighting. The above photos don't really show the RGB properly, but if you have a reasonably dark room and the radiator mounted in a case, there's plenty of RGB to go around. As with many other graphics cards, however, the RGB on the front of the card will end up facing the bottom of your PC in most builds, which sort of defeats the purpose.

Asus lists a relatively tame PSU requirement of 750W, which is technically sufficient based on our testing, but we recommend at least 850W. Actually, with most of the high-end graphics cards, you can make a legitimate argument for 1000W power supplies. PSUs often hit peak efficiency with a load of around 50 percent, and in a high-end PC build, that means the entire PC will consume close to 500W (depending on CPU and other factors) while gaming. If you do plan on using a lower wattage PSU, make sure it's a high-quality offering because the Asus card can hit power draws of over 400W when overclocked.

The big selling point with the Strix LC is, of course, the liquid cooling, and what that does for thermals. Despite having a similar power use and clock speed relative to the other two cards, thermals are far lower even with a modest fan speed. At stock, the Asus card has an excellent temperature of just 53C in FurMark, and that didn't change with our overclocked settings (though fan speeds did have to go up). Temps and fan speeds are even better when playing games, which are generally not as demanding as FurMark when it comes to power. Overall, the Asus card runs 15-20C cooler than the other two custom cards, with the same level of performance, and it delivers a bit more headroom for overclocking.

Speaking of overclocking, you can use Asus's GPU Tweak II (and GPU Tweak III) software suite in addition to the built-in Radeon Software options. By default, like many of Asus's graphics cards, the 6800 XT ROG Strix LC will run at 'gaming' clocks, and you need to install GPU Tweak II and select the OC profile to unlock the full potential. We did this for our testing, though the difference between 'gaming' and 'OC' modes is pretty minimal.

Is the Strix LC worth the price, though? Originally released with a $900 MSRP, Asus is one of several companies that has recently increased prices due to US tariffs and other factors. The Strix LC 6800 XT now shows a price of $1,080 at the Asus store, and it's still out of stock. It's hard to recommend spending that much money on a GPU that has a nominal price of $650, but then again, that nominal price is nowhere to be found. Maybe we'll eventually see RX 6800 XT cards selling at $700 or less, but for now, they're more commonly in the $1,000+ range. By the time supply improves, we might have other GPUs that represent a better overall value. 

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

The Sapphire RX 6800 XT Nitro+ is perhaps the most traditional of the three custom cards we're looking at today. It's not quite as large and certainly not as heavy as the ASRock card, but the cooling still gets the job done and the card runs cool and quiet. It's actually lighter than AMD's reference card, despite being slightly larger physically. The Nitro+ measures 310x134x55.3mm, so it's a bit shorter than the ASRock card but still occupies 2.7 slots. However, it only weighs 1232g, over 500g less than the ASRock, and 350g less than the AMD reference design. (AMD's card measures 267x120x49mm and weighs 1500g, if you're wondering.)

Sapphire often has three different models of each GPU. The base model is the Pulse, the Nitro+ is a step up and offers a higher factory overclock and more RGB lighting, and sometimes a Nitro+ SE kicks things up another notch. The top GPUs may also get a Toxic variant, which boasts the highest overclocks and pulls out all the stops, often with extreme pricing. Due to the limited quantity of Navi 21 GPUs currently available, Sapphire doesn't have a Pulse 6800 XT right now, or a Toxic. Maybe those will come in the future, but currently all of Sapphire's 6800 XT cards are from the Nitro+ line.

The main difference between the Nitro+ we're looking at and the Nitro+ SE is that the SE has RGB fans. If you want extra bling, the SE might be what you're after. Our take is that Sapphire actually has a smarter RGB setup than many other GPUs, as the fans typically face the bottom of the PC case and aren't even visible. The Nitro+ has all of its lighting on the 'top' of the card (which would face the side of your PC case, right where the window would be) and also puts an RGB icon on the back of the card (which would face upward in your typical case). There's a light strip on the top along with the Sapphire logo, and while it's not quite as in-your-face as other cards, it gets the job done.

Sapphire is also the only company to specify a higher 350W TDP on its Nitro+, though all three custom cards use similar amounts of power in practice. No surprise there, since they're all rated at 2360 MHz boost clocks (and can often exceed that speed in games). Sapphire also recommends an 850W power supply. With a pair of 8-pin power connectors, plus the PCIe slot, the Nitro+ has access to 375W of power and uses nearly all of it. Overclocking does push the card beyond 375W, but the extra power comes from the 8-pin PEG rather than the x16 slot, which is good. Even overclocked, the highest PCIe slot power we measured for the Nitro+ was only 45W. Sapphire also has new fans with an integrated rim to improve static pressure, and the thermals and noise levels are very good.

One extra that Sapphire offers is its TriXX software suite, which now has a new trick that's perhaps work checking out. TriXX Boost leverages AMD's own RIS (Radeon Image Sharpening) and supports upscaling of content. In most games (specifically, games that don't detect your monitor's native resolution and then internally scale from that), TriXX Boost lets you create a custom resolution that will then get scaled to the normal native resolution, and the sharpening helps to avoid the normal blurriness.

How does it perform? At 4K, upscaling from 85 percent resolution (3264x1836) to 4K, we measured performance that was anywhere from 20 to 30 percent higher than native. Did it look worse? Using screenshots and comparing, sure, there was a slight loss in fidelity, but it’s not something you'd really notice in motion. 20-30 percent higher fps, though? Yeah, that was very noticeable. It's perhaps not as sophisticated as Nvidia's DLSS technology, but it's worth considering for relatively minor upscaling of around 10-15 percent.

Interestingly, because of TriXX Boost, Sapphire specifically discourages end-user overclocking and doesn't support it via TriXX. You can check the fans and tweak the lighting, but overclocking requires some other utility. Considering overclocking can void your warranty and cause instability in pursuit of usually 5 percent more performance, toying with resolution scaling generally delivers far more noticeable performance improvements.

Thanks to the traditional design, Sapphire officially has the lowest price of the custom cards we've looked at so far. Whether it will remain at $770, or jump to a higher price segment (we're seeing a 'suggested' price of $1000 at Newegg right now), however, isn't clear. Is the Sapphire Nitro+ 6800 XT a good value? That depends on actual prices. For now, good luck finding one in stock. In fact, supply of AMD's Big Navi chips appears to be even worse than Nvidia's Ampere GPUs, which is why the retailers are charging extra. If you can find the Sapphire Nitro+ at a more reasonable price, we have no qualms recommending it. At $1,000 or more, though, we'd suggest biding your time. 

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

We’ll shift away from the specific cards for the next several pages and look at gaming performance and other test results. We've tested all of the RX 6800 XT GPUs using AMD's most recent 20.12.1 drivers, and we'll include a few other GPUs as reference points in the charts. Our standard test bed hardware is listed to the right, and it includes a stock clocked (4.7 GHz all-core) Core i9-9900K processor with overclocked DDR4-3600 CL16 memory.

While our CPU reviews focus on performance with the CPUs at full stock, meaning DDR4-2666 memory on the 9900K, my take is that enthusiasts will at least enable the memory XMP profile, and DDR4-3600 kits are readily available. We looked at performance scaling with similar memory kits (all running at DDR4-3600) with Core i9-10900K and Ryzen 9 5900X, and while there are instances where the newer CPUs do better — particularly at 1080p — in general the top three current CPUs are all within spitting distance of each other.

Our testing for the RX 6800 XT cards will use our expanded test suite from recent reviews, which includes the same nine games that we've used for the past year, plus four newer releases. We've tested all of the games at 'ultra' settings (or whatever the highest preset is called — or in the case of Red Dead Redemption 2, we set all of the advanced settings to the minimum option and then set the basic settings to maximum).

Only two of the games we've tested have ray tracing enabled, and one of those (Dirt 5) is an AMD-promoted game using a still-in-beta DXR patch. We didn’t enable DLSS on any of the games, though it can boost performance quite a bit, especially at 4K. We also want to note that our testing suite is decidedly slanted toward AMD right now, with half of the games as AMD-promoted titles, and two of the newer additions (Assassin's Creed Valhalla and Dirt 5) relatively strongly favoring AMD. If you want to view the larger picture of how the various GPUs stack up, we recommend looking at the full set of results, including DXR testing that we've done in other articles, including the RX 6800 XT and 6800 launch review. Since we're mostly focused on the custom 6800 XT cards here, any inherent bias in our game selection is less of a factor.

We'll start with 1080p results and then move up to 1440p and 4K on the following pages, and wrap up with power and thermal testing. We have both stock and overclocked results for the four RX 6800 XT cards in the charts as well, which generally clump together. We'll focus our commentary on the main results, with the individual gaming charts as background detail.

Given the mix of games and APIs used, it's not too surprising that the top four GPUs (RTX 3090, RX 6900 XT, RX 6800 XT, and RX 3080) end up relatively close in performance. The overclocked 6800 XT cards end up tying the 3090 and 6900 XT, but we're also hitting CPU bottlenecks in many of the games. While you can make the argument that 1080p is still the most popular resolution, and it's also the only place where you can get 240 Hz and 360 Hz displays right now, the latter is really only beneficial for a few select esports games. Out of the 13 games we tested, only Strange Brigade (which isn't exactly a popular game) breaks 240 fps at 1080p, and five of the games don't even get above 144 fps at the settings we've used.

As far as the four 6800 XT cards go, the results are pretty close to margin of error differences. The Asus Strix LC does take first place at both stock and overclocked settings, and the AMD reference card does take fourth place, but in general, you wouldn't be able to tell the cards apart when actually playing games.

Looking at the individual games, there's a bit of variability in some instances, like The Division 2, but nothing that gives us any real indication that any of the custom cards are clearly superior. Aesthetics, availability, features, and pricing are going to be much greater factors than pure performance. 

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

We still feel like 1440p gaming is the sweet spot for most people. Reasonably high frame rates are possible on quite a few GPUs, high refresh rate displays with adaptive sync technology can be had for under $300, and even 144Hz IPS and VA panels are commonplace. In our 1080p testing, we noted that a lot of games couldn't break 144 fps even at the lower resolution, but G-Sync and FreeSync make that less of a concern. CPU bottlenecks also become less of a factor at the higher resolution. Which doesn't actually change the overall standings much. 

At stock settings, the three custom 6800 XT cards are within 0.2 fps of each other, emphasizing once again how little difference there is among cards with the same GPU. The reference card does fall a bit off the pace, largely because it's using less power we'd wager, though even with our overclocked settings it can't quite close the gap. Interestingly, despite having three 8-pin PEG connectors and a large cooler, the ASRock card doesn't do any better than the Sapphire card — actually, it's a bit worse. It could be the luck of the draw, or just variance between benchmark runs, but our results consistently put the ASRock a bit below the other two AIB cards.

In the larger scheme of things, the overclocked 6800 XT cards continue to basically match the reference 6900 XT. The RTX 3090 starts to pull away from the other GPUs a bit, now that CPU bottlenecks are reduced, while the 6800 XT still leads the RTX 3080. It will be interesting to see what happens if Nvidia actually does release a 20GB 3080 (or 3080 Ti) card in the future, though the pricing on such a GPU would be higher than the current 3080. Well, theoretical 3080 prices, since most cards are selling for quite a bit more than MSRP and supplies are still very limited.

There's a bit more variation in performance this time, but nothing particularly noteworthy. The Sapphire Nitro+ takes the top spot in both stock and overclocked modes in several games, but most of the results are within 1-2 percent of each other at most (among the 6800 XT custom cards).

Looking at frame rates, even at stock clocks all of the games run at well over 60 fps, often into the 100+ fps range. Only a few break 144 fps, but again, that's why G-Sync and FreeSync are useful. There's one exception, naturally: Watch Dogs Legion with ray tracing enabled is a beast to run, as are most DXR games without DLSS. The 6800 XT can only manage high-30s performance, and it's worth mentioning that the DXR reflections on AMD and Nvidia still aren't the same. Check out the extra images in the gallery below to see what we're talking about.

The above are with the latest patches in place, and we're still getting a lot of differences between AMD and Nvidia DXR (DirectX Raytracing) quality. It's only used for reflections in Watch Dogs Legion, but you can see the differences in puddles, shiny metal surfaces, and various windows. It's as though AMD has greatly reduced the range of testing for ray triangle intersections, which could improve performance but at the cost of image quality.

Whether you want or need ray tracing is debatable, but with consoles now supporting the tech, we expect to see additional DXR games. Will they also support DLSS, or AMD's upcoming FidelityFX Super Resolution? And will they be 'dumbed down' for the console GPUs? Those are good questions, and considering the RX 6800 is faster than the Xbox Series X, we suspect limited RT effects on console games will be more common than more robust implementations like we've seen in Cyberpunk 2077. 

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

The next-gen consoles are supposed to tackle 4K gaming, but there's 4K gaming and then there's 4K gaming. 4K via upscaling and ray tracing effects targeting 30 fps is most likely what we'll see from the PS5 and XSX. 4K native at 60 fps, even without ray tracing, is going to be more difficult, even on high-end PC hardware. 

The good news is that many games easily reach 60+ fps at 4K on the RX 6800 XT cards, along with other extreme performance GPUs. The custom AIB models all end up around 2 percent faster than the reference 6800 XT at stock, which isn't actually that great considering they also use about 10 percent more power. Overclocking pushes the cards into 370W and higher territory, again for relatively minor performance improvements.

It's important to remember the mix of games and settings used when looking at this overall chart as well. If we toss out the newer games that strongly favor one GPU brand, like Assassin's Creed Valhalla, Dirt 5, and Watch Dogs Legion, the relative performance can shift around quite a bit. But we figure we at least capture a relatively large cross-section of the overall market with 13 games. The 3080 ends up just slightly ahead of the 6800 XT cards at stock, mostly thanks to its significant advantage in WDL. Do you need DXR to make a game fun? Absolutely not. However, it's also disingenuous to suggest that DXR doesn't matter at all; it's just going to take longer to get to the point where DXR and RT become the norm rather than an option for extreme hardware. 

Individual gaming charts at 4K mostly echo what we've seen already, just with lower frame rates. Every game outside of Watch Dogs Legion easily surpassed 60 fps at 1440p ultra, but there are a few more exceptions at 4K. Assassin's Creed Valhalla and Dirt 5 join WDL in the sub-60 group, though they're both relatively close to 60 fps on the 6800 XT. On the other hand, Watch Dogs isn't even close — it falls below 20 fps on the custom 6800 XT cards we're focusing on. Radeon gamers will definitely need to disable DXR reflections or run at a lower resolution if they want to play that game at acceptable frame rates. 

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Wrapping up our performance testing, the power and thermal results are actually more interesting than the raw performance. At least there's more variation between the cards in these areas, though all three custom cards stay reasonably cool and should run fine in any PC large enough to accommodate the hardware.

For these tests, we're using Powenetics software to gather the actual power draw of the graphics cards. GPU-Z is used to collect thermal, clock speed, and fan speed data. We also have an SPL meter that we set up 15cm from the side of the cards to capture noise. However, we don't have an anechoic chamber or anything particularly fancy. External noise (e.g., from traffic) means we have to simply eyeball the meter rather than logging data and generating a chart. We test all of these metrics using Metro Exodus, set to loop five times, and FurMark running at 1600x900 in stress test mode.

Let's quickly talk about the noise levels first. Recent changes to my office mean the new setup isn't identical to the old one, but the noise floor (when traffic isn't driving by) is now 34 dB. The SPL meter is kept close to the GPUs in order to isolate the noise from the card and not pick up as much fan noise from the CPU cooler, though differences in card design can be a factor. All of the latest GPUs support 0 dB fan technology, which means idle noise levels are all the same: 34 dB. Literally any other noise, like typing or someone walking around, registers at much higher levels than that, so it's pretty quiet.

Under load, the GPUs start to show their differences. For example, the Asus has two large 120mm fans on the radiator that move quite a bit of air, while the third fan on the card itself is basically silent. Asus ended up being the loudest of the three GPUs in the gaming test, measuring 45.6 dB peak and generally hovering in the 43-45 dB range, but the actual fan noise tends to be lower and, to my ears at least, not as noticeable as smaller fans. ASRock peaked at 44.5 dB in gaming, again with a range of around 41-44 dB. Sapphire showed the most variation, running as quiet as 37 dB but peaking at 42.1 dB.

Results under FurMark are more consistent, but the cards deal with the extreme power draw in different ways. As a result, ASRock was the loudest card in FurMark, peaking at 45.6 dB and basically staying in the 45+ dB range. Asus came in second at 44.1 dB, while the Sapphire card was the quietest of the three and only ran at 39.5 dB. Again, we didn't make charts because it's just the three cards, but none of the cards are particularly loud. Pay attention to the actual GPU clocks and thermals below when looking at noise levels, however, as all of these metrics are interrelated.

As mentioned earlier, the three custom AIB cards all have higher power use than the reference 6800 XT. In the Metro test, the ASRock uses the least power at around 332W, 30W more than the reference model. The Asus card uses 341W, though it has a pump and two larger fans on the radiator that undoubtedly use more power than the three smaller fans used on the other two cards. The Sapphire card ends up with the highest power use of 344W, which is as much as some of the other cards use even when overclocked. Overclocking isn't even that bad when you consider that performance improves by 4-7 percent (in Metro), while power use increases 1-6 percent. (Except for the reference card, which used 13 percent more power.)

FurMark takes power use to even higher values, ranging from 352W for the ASRock to 358W on the Asus and Sapphire cards. Overclocking pushes the three cards into the 400W and higher range, again with Sapphire using the most power of the 6800 XT models.

At factory stock settings, all of the RX 6800 XT cards meet or exceed the official boost clock, though in-game clocks will vary by game. In Metro, the custom cards reach 2386-2394 MHz average clocks, with overclocking pushing all three cards into the 2500 MHz and higher range. FurMark is a different story, with the GPUs reigning in clocks in order to keep thermals and power use in check, but 2.1GHz while running FurMark is still quite impressive, and overclocking takes that up to around 2.25GHz. 

Temperatures have a direct correlation with fan speed and noise, so they can't be considered on their own. As noted earlier, the Asus card had the highest noise levels, and while the fan speeds are quite low, two large 120mm fans still move a lot of air and can create more noise and turbulence than three 90mm fans. Still, the Asus runs over 15C cooler than the other custom cards in Metro, and over 17C cooler in FurMark, all while hitting similar clocks. Somewhat surprisingly, the Sapphire card has lower fan speeds than the ASRock but tends to be just as loud while running a bit hotter. Still, 72-75C isn't particularly high for a high-end GPU playing a demanding game.

You can also see how the significant ramp in fan speed to help with overclocking affects things. Temperatures are actually lower than stock, but that's because overclocking puts enough of a strain on the GPU that we didn't want to risk instability in pursuit of lower noise levels. If you want low thermals with a decent overclock, the Asus liquid cooler ends up being far superior to traditional air cooling.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

The performance, features, power, and aesthetics for these RX 6800 XT cards doesn't do much good if you can't actually buy the product, unfortunately. That's where we find ourselves right now. Theoretically, the Radeon RX 6800 XT is a good product, particularly if you're not as concerned about ray tracing performance. It's also theoretically slightly less expensive than the competing GeForce RTX 3080. In practice, however, it's all for naught.

As bad as the supply is on GeForce RTX 30-series cards, it might be even worse on AMD's alternatives. It's difficult to ascertain exactly how many of any of the cards are actually being shipped and sold, though, and both feel like ghost products right now. Tell someone you actually saw any of the latest GPUs in stock and they're liable to put you away. Which makes reviewing these cards a bit weird, but such is the way of the tech world right now.

The problem is that it's not just one thing causing the shortages. COVID plays a big role, and that impacted worldwide shipping as well. More people working and schooling from home, or just wanting to play games, means more demand. And in the case of the AMD RX 6800 XT, AMD uses TSMC's N7 process, which is in high demand from other companies as well. In the latest TSMC news, you can see that TSMC only does about 55-60 thousand wafers per month. How many of those go to Apple and Nvidia, and how many are for AMD? And of the AMD wafers, there will be Zen 3 CPUs, PS5 and XSX console APUs, and RDNA2 GPUs.

Suppose AMD gets one-third of TSMC's N7 production. That's perhaps 20K wafers. However, the PS5 and XSX consoles probably use up 75 percent of the allotment, meaning only 5K wafers for GPUs and CPUs. And right now, the Zen 3 CPUs are more lucrative (smaller chips means more per wafer, and higher profit margins). It's entirely possible that AMD is only doing a few thousand (or less) Navi 21 wafers per month. With a 520mm square die size, that's at best around 100 GPUs per wafer. Even that might be optimistic, as based on what we've seen it seems like less than 100K RX 6800 XT cards exist. Or maybe demand is simply so much higher than the supply that even with hundreds of thousands of cards, it wouldn't be enough.

If you want an RX 6800 XT and can find one of these cards in stock at a price you're willing to pay, have at it. The same applies to most of the other recent GPU launches. What everyone really wants to know is when these cards will be readily available at anything close to the launch price of $650-$700. Sadly, we don't know. Maybe in a few months, but we heard multiple companies at CES 2021 suggest that graphics card shortages are likely to continue until June at least. And if Bitcoin and Ethereum prices stay high, that would only make the situation worse, plus tariffs are also impacting prices in the United States.

It's annoying, and that's putting it nicely. In July, I wrote that it was a terrible time to buy a graphics card and updated that article after the new cards launched and immediately sold out. In retrospect, if you bought a previous-gen RTX 20-series or RX 5700-series GPU at MSRP or below right before the new cards launched, that was a smart buy. An even better buy would have been purchasing an RTX 20-series card back in 2018 because you'd then have over two years of enjoyment from it, and you could probably still sell it at close to the original price.

The winner of the current GPU battle will be whichever company can produce the most GPUs first and ship them at reasonable prices, with features, performance, and all the other aspects being secondary concerns. If we take the RX 5700 XT and the RTX 2060 Super as $400 graphics cards for our baseline, the RX 6800 XT is around 75 percent faster than the 5700 XT and 90 percent faster than the 2060 Super. That means we could reasonably accept prices of $700-$800. Anything more than that and we recommend waiting and searching for a better deal.

We know for certain that, just as the 2017 GPU shortages eventually came to an end, the current shortages will also pass into history at some point. Hopefully, that happens sooner rather than later.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

The Evolv Shift 2 is a case you buy because you love its towering, small footprint design – and those oh-so-beautiful anodized aluminum panels. Priced at $100 for the mesh version and $110 for the variant with TG and an addressable-RGB fan, lets find out if the Evolv Shift 2 deserves a spot on our Best PC Cases list.

I’ve been on the verge of purchasing Phanteks’ Evolv Shift ITX chassis a handful of times throughout the years, but I never did. Why? Because I didn’t ‘need’ another chassis. If you’ve ever felt this way about a case, welcome to my life. Of course, when Phanteks told me roughly half a year ago that a version 2 was on its way, it got me all giddy, but then like so many other things in 2020, it got delayed and delayed and ... you get the idea.

But now the new Evolv Shift is here, and I’m telling you, the wait was worth it for one simple reason: It appears Phanteks waited out the release of the newest GPUs to make sure they all fit in here.

The Evolv Shift 2 takes the first iteration and merely chisels away at its design to bring it up to 2020’s standards. There are now things like addressable-RGB with a built-in controller, better filtration, a slightly altered layout, and most importantly: room for today’s monolithic graphics cards.

These are all good developments, but it’s important to keep in mind that this is still a challenging case to build in, and performance enthusiasts aiming for the lowest possible temps won’t feel welcome here due to the limited cooling potential on the CPU.

Specifications

Features

The first thing you’ll notice about the Evolv Shift 2 is that it’s extremely tall at 19 inches (490mm), but that it has a very small footprint, meaning it doesn’t actually occupy a lot of space on your desk. Its footprint measures just 6.7 x 10.8 inches (170 x 274 mm), which including its feet give the case a volume of 22.9 liters. This isn’t small in the world of ITX cases, but because of its small footprint, the chassis gets away with its large format.

The outside of the chassis is tiled, with beautiful anthracite-gray aluminum panels at the front and back, while the sides are clad with tempered glass.

You might therefore wonder how to get to the case’s IO, but everything you do goes through the top of the chassis. Press down on the top vented panel to pop it open, and you’ll reveal the ‘rear’ IO and thumbscrews to loosen and remove the case’s paneling.

Being at the top, the rear IO is easy to access like this, and the cables coming out the back can be managed tidy as a single bundle. If you don’t like the cables coming out the top of the rear of the case, there are also cable guides behind the panel so that you can route them down, out of sight.

With the top vent closed, you have access to three buttons at the top that control power, RGB mode, and RGB color. It’s all very premium feeling around here, despite the $110 price point for this tempered glass version, though these three buttons do wobble around a bit– not that you’ll interact with them much if you just keep your system in sleep mode.

Around the bottom of the chassis are two USB 3.0 ports.

You’ll notice that there are no dedicated headphone or mic jacks to be found, but this is for good reason: the ‘rear’ IO is so easily accessible at the top that they’re simply not needed, and these dedicated ports always introduce a ton of interference to the audio signal anyway, so it’s no major loss.

Internal Layout

The interior of the Evolv Shift 2 has one of the most unusual layouts I’ve ever seen, though ‘unusual’ is basically the norm in the world of Mini-ITX. The entire lot is divided into three main sections.

At the bottom of the chassis, in the lower cubicle, there is room for up to SFX-L power supplies and a 120mm radiator. There is more space left here, but long GPUs will eat it up, and anything that remains will get filled with cable clutter.

The top main cubicle houses the motherboard on the right of the chassis, with the graphics card sandwiched on the other side via a PCIe riser cable. Note that the riser cable only does PCIe 3.0, as at this time, many case makers seem to be struggling getting a PCIe 4.0 signal working over riser cables.

At the front of the case you’ll spot a few thumbscrews to attach a 3.5-inch drive to, along with a tray that can house up to two 2.5-inch SSDs.

Cooling

To cool the system, Phanteks only includes a single 140mm spinner with addressable-RGB. If you’re buying the ‘air’ variant of this case with vented side panels instead of glass, you’ll get a fan without RGB instead as you won’t see it anyway, but you will still get the RGB controller, mainly because in the top vent is a power light that has RGB.

To cool the CPU, the most common method I expect to see here is 120mm AIOs, which will fit at the bottom of the case. This pushes the pump above the radiator, and yes, that’s normally a style worth scolding at, but the Evolv Shift 2 means for you to mount the outlets on the side, meaning the air bubble can move to a spot on the radiator’s reservoirs where it won’t rise to the pump and break it.

By removing the feet at the bottom of the chassis, you can add another 140mm spinner.

The top and bottom of the case are filtered. The bottom intake filter is easily removable after removing the right glass panel, and the top will catch falling dust.

Planning Your Build: Look at the manual before ordering

If you’re planning a build for the Evolv Shift 2, Phanteks has already published the manual, and it’s surprisingly good. You can find it here, and in it you’ll find detailed information on exactly which hardware combinations fit and which don’t, so do give it a glance if you’re planning a build. Or just order the case before any other parts.

Does it fit an RTX 3080?

Yes, the Evolv Shift 2 has been updated to fit larger 2.9 slot graphics cards. They can be up to 13.2 inches in length (335mm), 5.9 inches wide (150mm), and up to 2.9 slots thick (59mm). Note that this also goes for the IO: the bracket only has room for dual expansion slots, and GPUs with an IO shield that’s three expansion slots thick won’t fit. Fortunately, those aren’t very common anymore.

One thing to keep in mind, however, is that the tempered glass version of the case is only suitable for use with up to an RTX 3070. This isn’t because the RTX 3080 doesn’t fit, but because the TG side panel takes a heavy hit on thermals, and I reckon you wouldn’t want a GPU with a TDP over 250 W in that version.

We are using the following system for today’s system:

Step 1: Installing Lower Fan

Before anything else, I installed the intake fan at the bottom of the chassis. as getting to its mounting holes would be challenging after installing the power supply and AIO.

Step 2: Installing the PSU

I installed the Corsair SF750 power supply with the fan facing inwards, as facing outwards would be too close to the glass panel. If you are using the ventilated side panels, feel free to point the fan whichever way you prefer.

Step 3: Installing the Motherboard

Of course, I had already installed the CPU, SSD, and memory into the motherboard before installation, so all I needed to do here was pop the IO shield into place, line up the motherboard and screw it in. While here, I also took the liberty of attaching a few cables. As you can see, the mass of the cable clutter underneath was a job for later.

Step 4: Installing the AIO

Because we can’t see the RGB of the fan behind the radiator anyway, I opted to use a Corsair ML120 fan instead of the fans that came with the Cooler Master AIO. Technically there is space here for a push-pull setup, but with the sheer number of cables, a run-in with the fan would be inevitable if you don’t spend at least an hour managing cables and tying matters up. So I went with just a pull setup to save time. Note that SFX-L PSUs do fit here, but you’ll probably want to stick just to SFX to keep the space reserved for thicker radiators.

Note that the pump is at the top of the loop. I ran the tubing over the pump so that any air bubble wouldn’t get trapped in the pump, and I followed Phanteks’ recommendation to mount the radiator sideways so that the bubble would get trapped at the top of its reservoirs, out of harms way.

I now also took some time to connect a handful of cables and tidy up.

Step 5: Installing the GPU

To install the graphics card, I removed the included bracket, installed the GPU into it, and placed it back into the system. The beauty of this bracket system is that it’s easy to shift it left or right, allowing you to perfectly center your card behind the tempered glass.

Step 6: Crying over Cable management

Now that all the main parts are installed, it was time to connect the remaining cables and get everything tidied up. This wasn’t the easiest job with the limited space, but as I had done some of it along the way, I only needed about 30-45 minutes before I was satisfied. If you’re building this system to keep it running for longer than I will, you’ll probably want to spend some more time on this.

Step 7: Bask in the Glory of Top ‘Rear’ IO

With the rear IO at the top of the chassis, I now got to enjoy the beauty of all of the rear IO available at the top of the case. This is truly unlike anything I’ve seen before, and I can see the appeal of it – no more leaning over the desk, awkwardly feeling where the heck that Ethernet port is, and then realizing you need to flip the connector – everything is in sight and easy to access. Note, though, that if you have the case on a desk and you aren’t very tall, access here can also be frustrating.

Step 8: Build Complete

After a test run without the panels on and verifying that everything worked, I slapped the paneling on and had a moment to rest and relax. In the end, it took me about three hours to build this system. including photographing the most important steps along the way, which is significantly longer than it normally takes me to build a system in a ‘normal’ ATX chassis.

The Evolv Shift ITX isn’t an easy case to work in. It’s cramped inside, and if you have big hands, do make sure you have a small-handed building partner with you for plugging cables into tight spaces and fishing out dropped screws.

But effort and frustration along the way is the price you pay for a Mini-ITX system, and the end result you’re left is well worth the effort. I’ve always found working on Mini-ITX systems to be much more rewarding, despite the cramped frustrations along the way, and the Evolv Shift 2 is no exception.

MORE: Best Cases

MORE: All Case Content

Because this system is different from any standardized test bench, we’re keeping testing simple this time around. We’re keeping all the hardware at stock settings, only setting the fan speeds at a constant level so that we can easily tell if there is a difference between the ventilated ‘air’ panels and the tempered-glass side panels.

The end goal here is not to compare this chassis to other cases, but rather to decide whether the provided thermal capabilities are adequate – because let’s be honest, if you’re buying this case, you’re buying it because you like its style, and not because it runs a couple degrees hotter or colder than another – so our test is just to see whether it’s sufficient so that we can give you the go-ahead on buying.

So without further ado, the charts:

Phanteks was kind enough to provide us with the panels that go on the Air version of the Evolv Shift 2, and as you can tell from these tables, there is a stark difference in performance from the glass paneled version. CPU temperatures didn’t vary much between the two, but swapping the mesh panels onto the case dropped the GPU temperature by a whopping 10 degrees, which is substantial.

Of course, that thermal improvement came at the cost of acoustics, and I can tell you that the chassis ran noticeably louder with the mesh leaking sound all over the place. With the GPU at full blast we recorded the biggest difference in acoustics with as much as 2 dB of a penalty. In the other scenarios, the chassis also ran louder and subjectively, much less dampened – the sound level might only have been slightly lower, but the glass panels do a great job of keeping the harshness of fan bearing and pump noise in, resulting in a smoother sound.

So, I’d put it this way: If you’re running an RTX 2070 Super, or RTX 3070, you ‘can’ get away with using the glass panels. Yes, you’ll sacrifice thermals and a tiny amount of performance, but it’s within safe margins and no cause for concern. But, any GPU with a TDP above 250 W, and you should seriously consider the Evolv Shift 2 Air over the variant with tempered glass – unless you’re okay with sacrificing GPU Boost performance in favor of pretty looks and quieter operation.

For CPU cooling, consider a thicker 120mm radiator instead of the thin radiator we used, and a Noctua NF-A12x25 fan wouldn’t be a crazy consideration either if you want to keep noise levels down, especially if you’re using a recent high-end Intel processor. Dual-fan will help, but you’ll need a grille for the intake fan as otherwise the cables will jam into the blades.

Conclusion

The Evolv Shift 2 is one of the prettiest cases I’ve had the honor of working on all year. Building in it was tight, and came with the typical frustrations associated with Mini-ITX systems, but I still managed a build within about 3 hours, and the end result was well worth the effort.

That’s not to say that it’s a difficult system to build in – there’s only one way to install your hardware, so there’s not a lot of thought that needs to go into it: Mini-ITX motherboard? 120mm radiator? No more than one 3.5-inch drive? SFX or SFX-L PSU? If you’ve checked those boxes, you’re good to go. And yes, you can fit big GPUs in here. An extra 140mm spinner as bottom intake helps CPU temperatures and overall cooling.

With a small footprint and beautiful finish in both the tempered-glass and mesh variants, the Evolv Shift 2 is perfect as an SFF PC for use in the living room, moving around the house wherever you need it or taking to LAN parties. The easily accessible top IO makes plugging devices in a breeze too.

Priced at $100 for the air variant, or $110 for the glass version of the Evolv Shift 2, it’s not expensive either and therefore easy to recommend. This one deserves a spot on our Best PC Cases list as Best Mini-ITX case, at least until something better comes along. 

MORE: Best Cases

MORE: All Case Content

If you're planning to build a new PC (perhaps one of the best PC builds), you'll need a case for it. With early Prime Day hardware deals already upon us, we've found a great deal on an affordable chassis.

With clean and simple looks, the Phanteks Eclipse P300 tempered glass case is a great case for your build. Now for only $49.99 after rebate, and with free shipping via Newegg.

A high airflow ATX Mid Tower case with plenty of space for even the RTX 3090. The Phanteks P300 has a tempered glass side, to show off your RGB dreams. Front I/O is a little sparse, with just two USB 3.0 ports, headphone and microphone jacks, but this fits with the design of the case.

Dedicated cable management space and a hidden area for the PSU provide a clean aesthetic for your build. With two SSD drive bays hidden under the motherboard, and two more pop out forward facing 3.5 inch drive bays this case has all your storage needs. For a space saving case, the Phanteks Eclipse P300 offers plenty of space for your build.

I thought that Phanteks was done building big cases after the P500A, but then the postman showed up at my doorstep with the Enthoo Pro II with that look in his eyes: another big box? Yes, Phanteks is coming out with a new Enthoo Pro, which is arguably overdue as the original Enthoo Pro has been on the market for the better part of six years now. 

The old model is outdated, with tons of optical drive bays and, believe it or not, acrylic side panels. So Phanteks took to the drawing board to conjure something new, and boy did they make it. The Phanteks Enthoo Pro II is an absolute beast of a chassis, having room for not one, but two systems, the biggest of radiators and oh so many hard drives. 

Of course, we won’t be testing the chassis to its full extent. That would result in a trilogy, but it’s safe to say from the get-go that this is all the case you’ll need for years to come -- probably far more than you’ll ever need. 

The Enthoo Pro II sports a starting price tag of $130 for the closed version, or $140 for the variant with a huge slab of tempered glass on the side and RGB – which is the one we’re looking at today. Will the Phanteks Enthoo Pro II make it to our best PC cases list? (Spoiler alert: yes it does!) 

Specifications

Features

Phanteks is back boasting about its mesh, and the Enthoo Pro II comes with a whole new mesh that’s different from what we’ve seen before. Rather than being metallic or synthetic, it’s a fabric mesh, and it actually looks really good. It leaves quite large holes, but behind it is a fine synthetic mesh for filtering smaller particles and dust. 

Other than that, the Enthoo Pro II doesn’t deliver much in the way of premium materials. The entire front facade is made from plastic, and although it does have a brushed-aluminum design, it looks and feels cheap. But of course, what do you expect from a $140 chassis that’s this big? 

The tempered-glass side panel is absolutely enormous, and comes with plastic peel on both sides for protection. It sits on rails and uses thumbscrews at the rear to lock in place, so you do get a clean appearance on the side. 

At the top of the front there’s a slot that looks quite like an optical drive bay, but it isn’t. Rather, it’s a door that opens up to reveal all the front IO. Underneath are four USB 3.0 ports, one USB Type-C port, a headphone/mic combo jack, buttons to control the RGB effects, and a reset switch. 

With all this hidden underneath the latch, the outside of the chassis keeps its elements to a minimum and it all looks nice and tidy. All there is for you to interact with is the power button at the top, which surprised me with its RGB lighting. Believe it or not, the power button comes with two LEDs behind it that take part in the addressable-RGB goodness, with an LED strip inside taking care of most of the show.

Internal Layout

Opening the chassis up, we’re welcomed with a rather unusual layout that lends itself to an immense number of hardware configurations. At the top motherboard compartment there’s room for up to E-ATX motherboards, but an ATX board doesn’t feel too small here either. Right below it is a bump in the chassis, underneath which hides the power supply and space for cable management. 

On the motherboard side of that bump you have the choice of installing one of three different things: a PSU, a bunch of hard drives, or a Mini-ITX motherboard, as you can see in the pictures below. 

We’re not sure why anyone would want to install a second power supply there as it eats away at the space for the second system, but stacking four HDD trays there or installing a Mini-ITX board makes sense – why would you buy such a big case otherwise? Of course, do note that to get the Mini-ITX board powered up together with the ATX mainboard at the top, you’ll need a power supply that is capable of driving two systems, but Phanteks has you covered with its Revolt line for that. 

When it comes to hard drives, Phanteks offers tons of possibilities. Included with the chassis are four HDD caddies, which you can also hang from the side of the case in pairs of two. These are incredibly easy to hang up here, as they just slot into place, but they rely fully on gravity to keep them in place. As such, we would only install drives here if you have absolutely no intention of moving your system. In a car, it’d only take one bump for them to drop out of place and fall down, which isn’t something you want happening with your hard drives. 

If you purchase additional HDD caddies and populate all the spots, you’ll be able to fit up to 12 3.5-inch drives. The covers by the side intake also support two 2.5-inch drives each, provided you haven’t put something else in place of the covers.

Cooling

The Phanteks Enthoo Pro II doesn’t come with any fans included, which isn’t too surprising given that you’re getting this much chassis at this price point. Every system built in this case will undoubtedly be totally different from another, and the enthusiast who dares take on this case will probably have their own ideas for how to cool their system. 

If you’re buying this case though, keep in mind that for as affordable as it may be given its size, it won’t be cheap to fill up with hardware. Say you want to max out just the dans: The rear exhaust supports one 140mm fan, top exhaust supports three 120 mm fans, front intake can take up to four 120 mm fans, the blower intake another three, and side intake yet another four 120 mm fans. Tally that up and you’re looking at a total of fourteen 120 mm fans and one 140 mm spinner.  Of course, once you start installing a second system, hard drives, or other gear you’ll quickly start eating away at potential fan slots, but nevertheless. 

Next to the filtration at the front through the fabric mesh and fine mesh behind it, the case also comes with a filter for the lower intake, a cover over the top exhaust, and filters on the inside of the side panel to filter air coming in through the side. 

We are using the following system for our case test bed: 

As with all Phanteks cases, the central motherboard standoff anchors the board during vertical installation. We appreciate this as a time-saving thing in many cases, but in the Enthoo Pro II it’s almost essential – you won’t want to be placing this case on its side just to install the motherboard – it’s too big, and if your workspace is anything like mine, it won’t even fit on its side.  

The power supply went in without any fuss, and installation of the ITX motherboard in the lower section was just as easy. Of course, we shouldn’t be all too surprised. This case is huge, and with this much space to work in it’d be odd if installation wasn’t easy.  

Per Phanteks’ usual fashion with its mid-range cases and above, the accessory box comes with tons of goodies, including a plastic box with all the screws sorted – yet another thing that makes Phanteks cases so nice to work in. The materials might be cheap here, but you still feel like you get a ton of value for your money.

Cable Management

Unsurprisingly the Enthoo Pro II comes with oodles of room for cable management, but there is one thing we have to warn you about: Get a power supply with long cables, especially for the CPU. Most PSUs do  have long cables and we had no issue with our Corsair HX750i, but the CPU power cables were stretched to the limit and had they been much shorter, we would have needed extensions. 

Other than that, there’s nothing we can complain about. Phanteks includes three Velcro straps that allow for multiple points to guide cables through, and it didn’t take us more than few moments to tidy up the mess we created behind the motherboard tray. There is even a cover to shove over the mess you leave behind the power supply, because honestly who ever really spends the time on tidying that mess up?

With rubber grommets between the main compartment and the rear, the end result was a neat, tidy system. 

For thermal and acoustic testing, we are using the following software and settings: 

As noted earlierno fans are included with the Phanteks Enthoo Pro II, so you’ll have to bring your own spinners. Therefore, we have to skip the testing with the case’s own fans and jump straight to our standardized tests, where we use three Corsair ML140 fans. Because this case supports the large fans, it gets rewarded with the 140mm spinners, though at the 50% duty the 120 mm and 140 mm spinners produce identical noise levels.

Acoustic Results

For our acoustic tests, we run three scenarios: CPU full load, CPU and GPU full load, and an optimized idle. The CPU Full Load test runs the CPU and case fans at 50% duty. For the CPU and GPU full load acoustic test we add the Nvidia RTX 2070 Super FE at 75% fan speed, because in practice it never runs at 100 percent and is far too loud when it does.

For the optimized idle, we run the GPU fan speed at 40 percent (the 2070 Super FE GPU does not have a Zero-RPM mode), and run the CPU and case fans at the lowest speed they will spin at.   

The Phanteks Enthoo Pro II is the quietest case we have tested yet. Quite how this came to be isn’t clear to us as it is, like most other cases, a mesh case with a huge open front that happily lets all the noise out, but it seems to do something right. The differences aren’t huge though, and you’ll still want to invest in a quieter graphics card than we’re using for your own system.

Thermal Results

For the thermal tests, the CPU fan is set to 100% and the Corsair case fans to 50%. The i9-9900K is pegged at a 4.6GHz clock at 1.1v on all cores to ensure consistent power consumption across test scenarios. Letting the GPU run at 75 percent fan speed enables it to maintain its power target while maintaining one set reasonable fan speed, so that the temperature is the only variable. 

Thermally, the Phanteks Enthoo Pro II also puts down the best figures we have seen to date here. When Lian Li’s Lancool II Mesh landed, that case blew us away with its thermals and it’s an absolutely stellar case. The Enthoo Pro II earns an identical CPU temperature, but somehow drops the GPU temperature down by a whopping two degrees and has the RTX 2070 Super running at just 57 degrees Celsius. Perhaps this is just down to the extra space in the case making for less thermal bleeding between components.

Of course, temperatures will undoubtedly rise a little once you decide to slam the system full with a multi-GPU setup, second system, and a ton of hard drives. But with good fans, we trust that the Enthoo Pro II will be able to keep things cool.  

Conclusion

Phanteks’ Enthoo Pro II is a very unique chassis. It offers the most seamless dual-system support we’ve ever seen, and its new fabric mesh front looks really great, especially when you get close to the case. 

Thermally and acoustically the Enthoo Pro II also performs phenomenally well. Of course, the mesh does let more noise out than a closed-front case would, but if you’re careful in your component selection and only pick quiet parts, it should all remain very tolerable. 

But this is a big case. Its price tag, $130 for the closed version, or $140 for the variant with tempered glass and RGB, is good for a case with its capabilities. But most people simply don’t need a case that’s as big as this, and the finish of the front facade also reveals the shortcuts Phanteks had to take to get the price down to this level. 

So, this isn’t a case we’d recommend for everybody. Most of us would rather spend much less on our case, or spend this much on a case and have something with a proper quality finish – which this lacks. 

Like a big, reliable truck, the Enthoo Pro II is an excellent, roomy place for your PC parts. It’s pure practical bliss, and if you buy it you can rest assured that you won’t be needing another chassis for eons to come. For the crowd that wants a chassis like this with dual-system support, it’s easy to recommend.

Having already spawned the P300A and the P400A cases into existence, it’s time for Phanteks to unleash its biggest airflow-optimized case yet: the P500A. This chassis delivers a large mid-tower design that can accommodate all sorts of elaborate systems. 

This is an ATX chassis, and you wouldn’t look at it and think you could fit two systems in it. But, with an optional $20 accessory, you’ll be able to build a Mini-ITX system into the roof of the case, allowing you to hide your media server in plain sight.

Today we’re testing the $99 variant of the P500A that comes with two 140mm fans and no RGB support. For an extra $30, Phanteks will remove the boring black spinners in favor of three D-RGB fans at the front and a D-RGB strip along the PSU shroud.

Will the Phanteks P500A make it to our best PC cases list? Read on to find out. 

Phanteks P500A Specifications

Features of Phanteks P500A

In general, the Phanteks P500A comes loaded with features, though this sub-$100 RGB-less variant cuts out the fancy lighting effects. But it keeps the same clean external design, making it a wonderful chassis for tidy workplaces. 

Phanteks isn’t holding back when boasting about its mesh. Like with the P300A and P400A, the company is eager to explain how the mesh wraps around the corners of the front panel, enabling more airflow. There is also no filter to be found behind it – the focus here is performance, and you wouldn’t want a layer of filtration to reduce airflow. But that of course means you’ll want to be extra diligent about cleaning.

The side panel on the left is made of one huge slab of tempered glass, and it swings open at just a push. The front panel has a cutout where you can grab it, making opening it a breeze. Once swung wide, you can simply lift the panel out of its hinge to put it aside during assembly. One more thoughtful detail about the panel is its rubber strip – this ensures that it smoothly and evenly closes against the system, damping potential rattles and ensuring no air is drawn through the gaps. 

The IO resides at the top of the case, and consists of two USB 3.0 ports, a mic/headphone combo jack, a USB Type-C port, and power and reset switches. A white LED lights up at the top of the front panel to indicate that your system is switched on. 

Internal Layout of Phanteks P500A

Phanteks has put great thought into the P500A’s internal layout. At first glance, it looks like every other ATX chassis, but it brings some tricks that help it stand out from the crowd. 

You won’t be able to fit any E-ATX motherboards in here, but with the addition of a $20 bracket you can install a Mini-ITX board at the top of the chassis, hanging from the roof. Throw in another $60 and Phanteks will ship you a whopping 600-mm long (two feet) PCI-Express riser cable, allowing you to connect a GPU to a Mini-ITX board in the vertically-oriented location above the PSU shroud. But before you worry, there’s also a shorter riser cable that costs much less for those who only want to vertically mount a GPU connected to their primary motherboard.

Beneath the PSU shroud there’s room for the biggest of power supplies, though you may have to sacrifice some of the storage options. Out of the box, two 3.5-inch HDD caddies are included. These are installed by sliding into place, and you can fit up to four underneath the PSU shroud. You can also install up to six of these caddies in the main compartment, in place of the cable covers.

Behind the motherboard are three brackets to install 2.5-inch drives -- so yes, if you were so inclined, you could install up to ten 3.5-inch drives and three 2.5-inch drives in this system.

Cooling

As you would imagine, being the biggest of the range the P500A comes with the most space for cooling gear, and this is also true. Naturally, you’ll have to make some sacrifices if you wanted to install tons of hard drives or the second system, but if your primary focus is cooling, you’ll be able to fit up to two 140 mm or three 120 mm fans at the top, three 140 mm fans at the front and one 140 mm fan at the rear. These figures also match the supported radiator sizes. 

As noted earlier, the front does not have an intake air filter, so you’ll have to rely on the mesh to take care of that. The system opens easily though, so more regular cleaning shouldn’t be difficult. Thankfully the PSU does have its own fine mesh filter, as cleaning the PSU would otherwise become too laborious.

GPUs can be up to 435 mm (17.2 inches) long, or 294 (12.6 inches) with HDD caddies installed. CPU coolers can tower up to 190 mm tall (7.5 inches). 

System Configuration

We are using the following system for our case test bed:

As with most cases of this kind, installation of the motherboard was a snap. The central standoff anchored the board easily during vertical installation, and the GPU went in just as easily.  

The PSU also went in without any fuss, though with our huge Corsair HX750i, it wasn’t possible to fit the HDD caddies in the closest-to-the-PSU position, limiting us to two 3.5-inch drives below the PSU shroud. We liked that the PSU slides in from behind after applying a brace with thumbscrews, as this makes it very easy to slide the PSU out to make cable changes later. 

The HDD caddies come with their rubber grommets pre-installed, so all you need to do is screw in four points of the HDD and slide the assembly into place. 

One detail we appreciated about Phanteks’ P500A is the inclusion of a plastic sorted box for its screws. This is something Phanteks typically includes in its more expensive cases, and although it’s a small detail and not really environmentally responsible, it does save time compared to awkwardly digging through a tiny plastic baggy. 

Cable Management

Managing the cables in the Phanteks P500A was a breeze. It’s clear that Phanteks has put a lot of effort into making cable management easy, as is evidenced by the sheer amount of Velcro found behind the motherboard tray. 

The PCI-Express power cables come through their own cutout in the PSU shroud, which slides open to allow the connectors through and then shuts again to keep things tidy. The same principle applies to the cable covers in the motherboard area – these slide open to easily allow all the cables through, and then slide shut, creating an almost seamless look.

MORE: Best Cases

MORE: All Case Content

For thermal and acoustic testing, we are using the following software and settings: 

Fan Characteristics

The Phanteks P500A comes with two 3-pin 140 mm fans, which perform surprisingly well. They spin at a minimum speed of about 550 RPM at 35% duty, and top out at about 1300 RPM, creating a not-unpleasant sound at higher speeds. This isn’t very fast, but should be ample for good airflow in this open case. 

Acoustic Results

For our acoustic tests, we run three scenarios: CPU full load, CPU and GPU full load, and an optimized idle. The CPU Full Load test runs the CPU and case fans at their maximum speed. For the CPU and GPU full load acoustic test we add the Nvidia RTX 2070 Super FE at 75% fan speed, because in practice it never runs at 100 percent and is far too loud when it does.

For the optimized idle, we run the GPU fan speed at 40 percent (the 2070 Super FE GPU does not have a Zero-RPM mode), and run the CPU and included case fans at the lowest speed they will spin at.

The Phanteks P500A comes in as the loudest case with the fans spinning at full speed, but don’t let this scare you. You’ll see in a moment that its thermal performance is also among the best, so you can easily choose to run the fans at lower speeds if you prefer lower noise levels. The case can run extremely quietly when optimizing the fans to run at their lowest speeds. 

Thermal Results

For the thermal tests, all case and CPU fan speeds are set to 100 percent. The i9-9900K is pegged at a 4.6GHz clock at 1.1v on all cores to ensure consistent power consumption across test scenarios, and letting the GPU run at 75 percent fan speed enables it to maintain its power target while maintaining one set reasonable fan speed, so that the temperature is the only variable. 

The Phanteks P500A puts down the best thermals we have recorded yet. This isn’t too much of a surprise given the noise levels, but an impressive accomplishment for a chassis packing two fans nevertheless.

Standardized Results

To correct for differences in case fans, for the standardized test we remove the standard case fans and replace them with Corsair ML120 or ML140 units. Up to three fans may be installed, preferably one at the default exhaust location and two at the front intake, in the biggest sizes the case supports. The above tests are then repeated, but because the Corsair ML fans have absurdly high maximum speeds, we run those at 50% duty for the tests as no case fan should ever need to run above those speeds. 

Testing with an identical setup on all cases, it becomes clear that most mesh cases perform quite similarly. Noise levels are naturally within a margin of error identical to similar cases, and Phanteks is trading blows with the be quiet! Pure Base 500DX, with both attempting to top our leaderboard. 

Conclusion

Add up all the things the Phanteks P500A does well, and it’s clear we have a winner on our hands. The chassis is big, and if you’re not using the second-system capabilities or at least filling up a few of the HDD slots and radiator spots, you might prefer to look at a smaller case such as the Phanteks P400A, but you can’t go wrong with the P500A. 

It packs neat, tidy European styling, excellent thermal performance, and is practical beyond all sense. It runs a little louder than we would like, but as long as you pack quiet hardware into it and spend some time optimizing your fan curves, noise shouldn’t be an issue.

If you’re looking for a big, flexible chassis that you won’t outgrow in the coming years and that focuses on thermal performance above all else, the Phanteks P500A is for you.

MORE: Best Cases

MORE: All Case Content

Dutch case, cooling, and PSU manufacturer Phanteks offers up the Eclipse P300A, an airflow-optimized variant of the P300, ditching the RGB lighting and adding a front mesh intake. This new front facade should do the case some favors in terms of temperatures, but the chassis only comes with one fan included, and that isn’t much.

But the basic setup also means the price is low; you can buy yourself the Phanteks P300A for just $60, which might just make you forgive it for its lack of included extras. At this price, the company is targeting the entry-level builder who wants an ATX chassis that cools well and isn’t overly designed or flashy. There’s a lot to like here, but will the P300A make it to our best PC cases list? First, let’s take a close look at the Eclipse P300A’s specs.

Specifications

Features

When unpacking the Phanteks P300A, it’s clear you’re dealing with a very basic chassis with hardly any features that you don’t strictly need. The tempered glass side panel is the primary aesthetically focused feature. There is no RGB lighting, but you’ll find a subtle white line that lights up at the bottom right of the glass panel that looks quite neat.

In terms of size, the P300A is about as compact as an ATX chassis gets. Gone are the days where we need 5.25-inch external bays and tons of hard drives. This chassis stands out by being one of the smallest “normal” ATX mid-towers around.

At the top of the chassis is a magnetic mesh panel to hide the screws to attach a fan. It’s a nice touch, but it slides out of place easily and doesn’t fall into its cavity as nicely as in other cases. Both the right side/back panel and the tempered-glass panel feature thumbscrews at the rear and slide off the chassis for removal. Most glass panels come with thumbscrews on the side to secure them, but despite being a budget case, Phanteks has opted for a sliding mechanism here. This is less finicky in use and results in a flush side profile, adding to the minimalist appeal.

The front I/O resides at the front of the top of the chassis, consisting of two USB 3.0 ports, separate mic and headphone jacks, along with power and reset buttons. There is no USB Type-C here, but this is just a $60 case, and front-panel USB-C headers are still somewhat rare-especially on the kind of budget boards you’re likely to pair with a case that’s this affordable.

Internal Layout

Internally, the Phanteks P300A is simple. In the main motherboard compartment you’ll find space for the motherboard, graphics card, and associated cooling components and no more.

Underneath the power supply shroud is space for the PSU as well as two 3.5-inch drives. The case also supports two more 2.5-inch drives behind the motherboard tray. But a bracket for just one drive is included in the box.

Cooling

For cooling, the chassis comes with one 1200 RPM fan that uses voltage control for regulation (3-pin). This is a simple budget fan for a budget case, but it seems to perform adequately as we found in our testing. Still, for most builds you’ll probably want to add at least one more spinner.

The top of the chassis can support a 140 mm fan, though we wouldn’t install a 140 mm radiator here as it will likely lead to clearance problems with tall VRM heatsinks. The front of the chassis supports up to two 120 mm or 140 mm fans, and while a 240 mm radiator will fit, you’ll likely run into some clearance problems with larger 280 mm units.

Phanteks opted to include only an air filter at the bottom of the chassis for the PSU intake, explaining that the front mesh is so fine that it acts as a filter. This should, in theory, reduce intake drag and therefore improve temperatures, though with all the gaps around the chassis, we would still expect some dust buildup. The front panel comes off easily without much force for cleaning and accessing the fan mounts and HDD caddy.

System Configuration

We are using the following system for our case test bed.

With the motherboard standoffs pre-installed, installation of the motherboard was simple. The central standoff anchored the board, letting us easily install the screws to fasten the Maximus XI Hero down.

Installing the power supply was less straightforward. The Corsair HX750i that we use in our standard test system did not fit with its modular cables attached, so we had to use a smaller power supply provided by Phanteks. This unit was the Phanteks AMP 650, which fit without issue. For a modular power supply, we wouldn’t choose a bigger unit than this one at 140 mm long, despite the case’s spec list allowing up to a 200 mm unit. So be sure to choose your power supply carefully.

Even though we’re not using it for the tests, two 3.5-inch drives installed easily into the HDD caddy with the toolless installation mechanism. At first, we had some trouble, but discovered quickly that the  drives need to be installed with the power and data connectors at the front of the chassis rather than at the PSU side. If you regularly swap drives, this system will make your life very easy.

Cable Management

The cutouts in the PSU shroud at the bottom of the motherboard are adequate in depth, but there aren’t enough of them to line up all connections nicely with our Maximus XI Hero motherboard, resulting in more exposed cabling than I would like.

The rear of the motherboard only has 14 mm (slightly more than a half an inch) between itself and the side panel, not enough room to work with if it were the only space for wires. However, Phanteks’ design has a large cable-routing gutter to the left of the motherboard that is 35 mm (1.4 inches) deep, providing lots of room to stow extra cables and slack. We also appreciate the inclusion of Velcro straps for keeping things tidy.

Benchmark Settings

For thermal and acoustic testing, we are using the following software and settings:

Test Results

The P300A only comes with one case fan that spins at roughly 1200 RPM, which helps extract heat from the CPU cooler, but there isn’t much airflow to help the GPU out.

Acoustics

For our acoustic tests, we run three scenarios: CPU full load, CPU and GPU full load, and an optimized idle. The CPU Full Load test runs the CPU and case fans at their maximum speed. For the CPU and GPU full load acoustic test we add the Nvidia RTX 2070 Super FE at 75% fan speed, because in practice it never runs at 100 percent and is far too loud when it does.

For the optimized idle, we run the GPU fan speed at 40 percent (the 2070 Super FE GPU does not have a Zero-RPM mode), and run the CPU and included case fans at the lowest speed they will spin.

The one included case fan can spin at a minimum speed of about 500 RPM, which is likely due to the light blade design, as it lacks momentum to keep spinning at lower speeds, and doesn’t come with PWM support. Nevertheless, in the optimized idle scenario the system is audible, and although it puts down a slightly lower dB reading than the be quiet! Pure Base 500DX, we prefer the acoustic profile of the be quiet! chassis, as its three fans spin at lower speeds and create a smoother sound.

Once the CPU and especially the GPU fan speeds ramp up, noise levels quickly become irritatingly loud. Of course, note that the GPU we have installed is especially loud specifically to bring out the biggest noise differences between different cases, and you’ll be able to achieve better results with the right hardware choices. In particular, if opting for a case with lots of mesh and a powerful graphics card, look for a model of card with a zero RPM option to cut down on noise when you aren’t gaming.

Thermal

For the thermal tests, all case and CPU fan speeds are set to 100 percent. The i9-9900K is pegged at a 4.6GHz clock at 1.1v on all cores to ensure consistent power consumption across test scenarios, and letting the GPU run at 75 percent fan speed enables it to maintain its power target while maintaining one set reasonable fan speed, so that the temperature is the only variable.

Compared to the be quiet! Pure Base 500DX that we previously tested at this location, the removal of two fans has made the CPU temperature climb by 2 degrees and the GPU by 6 degrees. Nevertheless, the ambient temperature in our testing room was 23 degrees, placing the actual CPU and GPU temperatures at 70 and 71 degrees, which are still perfectly good numbers especially if you consider that our test hardware is harder to manage than what most shoppers in this segment will be throwing in this case.

Standardized Results

To correct for differences in case fans, for the standardized test we remove the standard case fans and replace them with Corsair ML120 or ML140 units. Up to three fans may be installed, preferably one at the default exhaust location and two at the front intake, in the biggest sizes the case supports. The above tests are then repeated, but because the Corsair ML fans have absurdly high maximum speeds, we run those at 1200RPM for the load tests, because no case fan should ever need to run above those speeds.

You’ll notice that improving the fans in the chassis also dramatically improved temperatures. Although the CPU temperature only improved by 2 degrees, the GPU was very happy with the added cool air supply, running a whopping 9 degrees cooler. In the standardized test, both the Phanteks Eclipse P300A and the be quiet! Pure Base 500DX also put down near-identical numbers, though this isn’t surprising as they’re both meshed cases optimized for airflow rather than keeping noise down.

Conclusion

Phanteks’ P300A is a compelling chassis at its $60 price point, offering simple but sophisticated looks with its all-black design and tinted glass panel. Its mesh front panel also lends the chassis to a highly airflow-optimized design. Despite that it only comes with one fan from the factory, performance is acceptable thanks to the mesh front. We would still recommend adding at least one fan to blow air onto the GPU if you can spare the expense, but it’s worth noting that our test bed with the i9-9900K and the 2070 Super is far beyond what most shoppers in this segment would throw at this case, and the P300A’s single fan still managed to keep temperatures in check.

Some of the case’s other shortcomings include a lack of front USB Type-C, cheap materials, and a lack of RGB. But we can easily forgive those shortcomings if you consider all the things the case does well at its price.

If the one-fan thing isn’t an issue for you, the Eclipse P300A is an excellent choice as a beginner’s chassis or for a modest build. It’s superbly easy to work in,adding almost no complexities to the build process – seriously, it’s one of the easiest cases to manage. It’s also one of the smallest ATX cases around. So if you’re sticking to the ATX platform but don’t want a huge tower on your desk, want good performance and aren’t too fussed about noise levels, you should definitely put the Phanteks Eclipse P300A on your shortlist.

The Phanteks AMP Series is based on the newest revision of the Seasonic Focus platform, which is quite popular in the PSU market. The strongest model of the AMP line, with 750W max power, offers high performance, but its noise output is quite high, reaching 33 dB(A). On the other hand, the similar capacity Corsair RM and RMx units, along with the Asus ROG Strix 750, offer high performance along with silent operation,

We have already reviewed the AMP models with 550W and 650W max power and found them to be highly competitive in the tough PSU market. For users that need more Watts, though, the AMP model with 750W is a more suitable choice. Like the other two members of the family, this one also uses the newest version of the Seasonic Focus platform, which is among the most popular in today's market.

The Phanteks AMP 750W has compact dimensions, and it is fully modular, featuring two EPS and four PCIe connectors. It can deliver almost full power at +12V, while the combined power of the minor rails is restricted at 100W since no modern system will require more Watts from these rails. 

Specifications

Power Specifications

Cables & Connectors

There are two EPS and four PCIe connectors, along with a large number of peripheral ones. All cables are long enough for a 750W power supply, but the distance between the 4-pin Molex connectors is low at 120mm. It should be 150mm, at least. Finally, there are in-cable caps only on the ATX cable, and all connectors use the standard 18AWG gauges, with only the sense-wires being thinner (22AWG).

Component Analysis

We strongly encourage you to have a look at our PSUs 101 article, which provides valuable information about PSUs and their operation, allowing you to better understand the components we're about to discuss.

As we already mentioned, all AMP models are based on Seasonic's newest Focus Plus platform. The main PCB is compact. Still, there is enough clearance between parts to allow for proper airflow. The build quality is excellent, and all parts are by good manufacturers so that they will have a pretty long life even under harsh operating conditions.

The transient filter has all the necessary components, but as usual, our EMC pre-compliance testing will reveal if it does a good job. 

There are two bridge rectifiers, bolted on a dedicated heat sink. 

Right in front of the APFC converter is the NTC thermistor and bypass relay combo, that restrict inrush currents. 

The main FETs are arranged into a full-bridge topology, which is supported by an LLC resonant converter.

The +12V FETs are installed on the solder side of the main PCB. Two DC-DC converters are powered from this rail, to generate the 5V and 3.3V rails. 

Japanese manufacturers provide the filtering caps, and most of them belong to good lines with an increased lifetime. Besides electrolytic caps, a large number of polymer is also present on the main PCB and the modular board.

The standby PWM controller, responsible for the 5VSB rail, is partially hidden by the primary heat sink. 

At the front side of the modular board, we find four polymer and three electrolytic caps. As you can see in the photos above, there is room for more polymer caps. 

The soldering quality is good, and the same goes for the parts that Seasonic used to make this platform. 

As usual, for a Seasonic-made product, the cooling fan is provided by Hong-Hua, and its model number is HA1225H12F-Z. It uses a fluid dynamic bearing and measures 120mm. The use of a larger fan was not possible, due to the compact chassis. 

MORE: Best Power Supplies

MORE: How We Test Power Supplies

MORE: All Power Supply Content

To learn more about our PSU tests and methodology, please check out How We Test Power Supply Units. 

Primary Rails And 5VSB Load Regulation

The following charts show the main rails' voltage values recorded between a range of 40W up to the PSU's maximum specified load, along with the deviation (in percent). Tight regulation is an important consideration every time we review a power supply, because it facilitates constant voltage levels despite varying loads. Tight load regulation also, among other factors, improves the system’s stability, especially under overclocked conditions and, at the same time, it applies less stress to the DC-DC converters that many system components utilize.

The load regulation is extra-tight on all rails but 5VSB, where it is satisfactory. 

Hold-Up Time

Put simply; hold-up time is the amount of time that the system can continue to run without shutting down or rebooting during a power interruption.

The hold-up time is close to 20ms while the power-ok signal exceeds the threshold that the ATX spec sets, and it is accurate. 

Inrush Current

Inrush current, or switch-on surge, refers to the maximum, instantaneous input current drawn by an electrical device when it is first turned on. A large enough inrush current can cause circuit breakers and fuses to trip. It can also damage switches, relays, and bridge rectifiers. As a result, the lower the inrush current of a PSU right as it is turned on, the better.

The inrush currents are a bit higher than the competition's. 

10-110% Load Tests

These tests reveal the PSU's load regulation and efficiency levels under high ambient temperatures. They also show how the fan speed profile behaves under increased operating temperatures.