Enthusiasts are the folks who appreciate the benefits of lower temperatures, and most of us also put high value on less noise, too. The fight between cooling performance and acoustic pollution became particularly acute in AMD’s Radeon R9 290 graphics cards. More so, even, when the company was forced to override its firmware-based fan ramp through a driver in order to deliver consistent clock rates.
Where we once saw performance variance as large as 15% from one board to the next, that “fix” still has us finding variations up to 5% caused by slight changes in room temperature.
Most of us would pay to avoid those issues, and the $400 Radeon R9 290 was supposed to be cheap enough to encourage value-minded adoption of high-end graphics hardware. But in our search for an answer to the 290’s reference cooling woes, many of us forget to ask the right question: how much would we be willing to spend on better performance and lower noise, while giving up our warranty coverage? Based on today's online prices, water block manufacturer EK thinks the answer is around $120 (add $30 for the back cover).

VisionTek does its math a little differently. On the product page of its CryoVenom R9 290, the company values your time building a water-cooled R9 290 at $100, as you also sacrifice its warranty. The marketing gets a little fuzzier as VisionTek calculates that a $120 cooler, a $30 back cover, and your $100 of time pushes the cost of a $400 Radeon R9 290 to $651.
Even if you disagree with those numbers, though, a do-it-yourself configuration with an original $400 card, the EK cooler, and the company's backplate would cost at least as much as the original $550 VisionTek CryoVenom R9 290.

Right now, some of you are probably thinking that a $550 liquid-cooled Radeon R9 290 would be one heck of a deal in a market loaded with $600 air-cooled cards, and you’d be right. Acknowledging the crazy market pricing for Hawaii-based Radeons, VisionTek admits it needs to charge more to cover cost increases on certain board components. As a result, the company recently bumped its CryoVenom up to $600 for new orders.
As of this writing, the CryoVenom R9 290 isn't available to order, though. So, there's no way for us to know if that $600 price tag is going to hold up over time. What we do know is that the cheapest R9 290s go for $550 on Newegg, so you'd still be getting $150 worth of liquid-cooling equipment and a one-year warranty at a substantial discount. But again, that's simply not something we can vouch for on a card you can't buy right now.
Our CryoVenom R9 290 arrived in the limited-edition wooden crate you saw on the previous page. VisionTek tells us that this could eventually become part of a collector's edition of the board with some additional accessories. However, you should expect the card you buy to show up in a cardboard box instead.

VisionTek uses the acrylic-covered version of EK’s FC R9-290X to show off its nickel-plated copper base. What you get is slightly pricier and more corrosion-resistant than its bare-copper sibling. This also gives VisionTek the perfect place for its logo.

EK’s aluminum back cover is also emblazoned with VisionTek branding.

We appreciate that EK does a good job building single-slot coolers. But the Radeon R9 290 itself still requires a dual-slot form factor to expose its second DVI output.

Effective overclocking is one of the reasons you'd want the liquid-cooled CryoVenom R9 290. There are, of course, overclockable air-cooled cards as well; you simply have to crank their fans up in order to exploit their higher frequency ceilings. VisionTek's CryoVenom R9 290 carries over AMD's reference 947 MHz core clock rate with DDR3-5000 memory. It's up to us to figure out how far Hawaii can be pushed manually.
Graphics guru Chris Angelini strongly recommended an upgrade from AMD’s fleshed-out Catalyst 13.12 graphics drivers used in PowerColor LCS AXR9 290X: Water Makes Hawaii Comfortable to a beta version of Catalyst 14.1. A few benchmarks revealed why, as the new driver boosted the performance of the lower-model R9 290 beyond the previous 290X results.
| Test System Configuration | |
|---|---|
| CPU | Intel Core i7-4770K (Haswell): 3.5 to 3.9 GHz, 8 MB shared L3 cache, LGA 1150 Overclocked to 4.5 GHz, 1.25 V at 100 MHz BCLK |
| Motherboard | Asus Z87 Pro: Intel Z87 Express, UEFI 1707 (12/13/2013) |
| Reference Graphics | Sapphire R9 290 (100362SR): 947 MHz GPU, 4 GB GDDR5-5000 |
| RAM | Mushkin Redline Ridgeback 997121R 16 GB Dual-Channel Kit 2 x 8 GB XMP-2133 CAS 9-11-11-28, 1.65 V |
| Hard Drive | Samsung 840 Pro MZ-7PD256, 256 GB SSD |
| CPU Cooling | Thermalright MUX-120 w/Zalman ZM-STG1 Paste |
| Case | Nanoxia Deep Silence 1 |
| Sound | Integrated HD Audio |
| Network | Integrated Gigabit Networking |
| Power | Seasonic X760 SS-760KM: ATX12V v2.3, EPS12V, 80 PLUS Gold |
| System Software | |
| OS | Microsoft Windows 8 Professional RTM x64 |
| Graphics | AMD Catalyst 14.1 beta 1.6 |
Main system components are carried over from the LCS AXR9 290X review, except for the part that matters most: a retail-purchased Sapphire Radeon R9 290 (with reference cooling) replaces the air-cooled R9 290X in this comparison of VisionTek’s CryoVenom R9 290.

Because AMD’s driver attempts to keep fan speed under 60% and underclocks the GPU when its temperature climbs to 94° Celsius, air-cooled cards are highly impacted by ambient temperature. Unfortunately, it's still super cold here, and I'm unable to keep my lab above 19 °C, which means that the air-cooled card in this comparison will consistently perform better than it would in a more typical 21 to 24 °C gaming room.
| 3D Game Benchmarks | |
|---|---|
| Arma 3 | Version 1.08.113494, 30-Sec. Fraps "Infantry Showcase" Test Set 1: Standard Preset, No AA, Standard AF Test Set 2: Ultra Preset, 8x FSAA, Ultra AF |
| Battlefield 4 | Version 1.0.0.1, DirectX 11, 100-Sec. Fraps "Tashgar" Test Set 1: High Quality Preset, No AA, 4X AF, SSAO Test Set 2: Ultra Quality Preset, 4X MSAA, 16X AF, HBAO |
| Far Cry 3 | V. 1.05, DirectX 11, 50-sec. Fraps "Amanaki Outpost" Test Set 1: High Quality, No AA, Standard ATC., SSAO Test Set 2: Ultra Quality, 4x MSAA, Enhanced ATC, HDAO |
| F1 2012 | Steam version, in-game benchmark Test Set 1: High Quality Preset, No AA Test Set 2: Ultra Quality Preset, 8x AA |
| Metro: Last Light | Steam version, Built-In Benchmark, "Frontline" Scene Test Set 1: DX11, Med Quality, 4x AF, Low Blur, No SSAA, No Tesselation, No PhysX Test Set 2: DX11, High Quality, 16x AF, Normal Blur, SSAA, Tesselation Normal, No PhysX |
| Tomb Raider | Steam version, Built-In Benchmark Test Set 1: High Quality Preset (8x AF, FXAA), Motion Blur, Screen Effects Test Set 2: Ultimate Quality, (16x AF, FXAA), Tesselation, TressFX |
| Synthetic Benchmarks | |
| 3DMark Professional | Version 1.1, SystemInfo 4.17.0.0, Fire Strike Benchmark (Extreme Off/On) |
Benchmark charts look a little bare with only two data points on them. When I reviewed my last water-cooled card, I filled-out those graphs with four unique configurations, including the Quiet and Uber firmware switch settings. This made particular sense in that story because PowerColor's LCS AXR9 290X's Uber mode actually applies an overclock.
VisionTek's CryoVenom R9 290 doesn't offer that provision, and AMD recently narrowed the difference between air-cooled cards by overriding the Quiet mode's fan speed ceiling using its Catalyst driver. Extensive comparisons between the two BIOS settings suggest that there's really no performance difference between them, though Quiet mode does drop to lower power consumption at idle.
So, I decided to start today's review with overclocking, which lets me use the factory and overclocked settings to generate some additional benchmark results.

The air-cooled, retail-purchased Radeon R9 290 from Sapphire is the card to beat for VisionTek's liquid-cooled contender. Using MSI Afterburner, I set the maximum power limit (+50%) and MSI’s custom fan curve (100% at 90 °C) to keep clock rates steady. But I was only able to hit 1100 MHz at those settings. Still, it’s a steady 1100 MHz, which is roughly 50% faster the frequency floor you'll see when the R9 290 throttles all the way down.
I’m incredibly critical of marketing material, so I was dubious of this quote from VisionTek’s product description:
Obtaining maximum performance from the CryoVenom R9 290 couldn’t be easier. You just enter the specifications from the included build sheet in the Catalyst Control Center and in less than a minute, the CryoVenom R9 290 can safely deliver up to 1175 MHz GPU speed...a 24% increase from the stock setting of 947 MHz. The memory clock speed is also increased to 1450 MHz...16% faster than the 1250 MHz stock setting.
So, what did I get?

Hitting a 1160 MHz core clock and 1440 MHz GDDR5 memory frequency is still in the neighborhood of what I was told to expect. Then again, I'm a little fussier than the marketing folks when it comes to things like exact specs and game stability. I did, after all, increase the power limit by 50% to prevent the card from throttling back the instant I needed its maximum performance.
Even in a cold room, we can plainly see the impact of thermal throttling in 3DMark's Fire Strike test using an air-cooled Radeon R9 290. Overclocking helps fix that board's performance, since I'm increasing the fan speed and power limit. But the liquid-cooled card's higher frequencies assure its victory.

Though it’s possible to force power throttling on the CryoVenom at stock settings, this only happens under unusually heavy loads like FurMark. This phenomenon would have been easier to trigger with the clock rates turned up, except that I also increased the power limit by 50%, creating headroom.

The CryoVenom R9 290 retains its lead through 3DMark’s Fire Strike “Extreme” settings.
Remembering that my lab's low ambient temperatures give the air-cooled card a best-case environment, VisionTek's CryoVenom R9 290 continues to lead at the same clock rates.

With both cards overclocked, the water-cooled CryoVenom even manages to trump the retail Sapphire board's optimized fan settings.

I might have expected water cooling to give VisionTek's offering an advantage using Tomb Raider's Ultimate quality preset, but the lead instead remains constant.

F1 2012's High quality setting presents too light of a load for the fastest graphics cards.

The more strenuous workload of a multi-monitor configuration starts weighing on both cards once we dial in this title's Ultra detail preset. They boast the same GPU, and the most effective cooling solution (water) wins.
Both Radeon R9 290 graphics cards appear bottlenecked by platform components at 1920x1080. Adding a couple of monitors helps the water-cooled card stand out, though. Even with both cards set to stock frequencies, and benchmarking in an exceedingly chilly lab, the air-cooled board appears to pull back on its peak performance.

Frame rates remain smooth through our Arma 3 test using standard quality settings.

Average frame rates look terribly close to the minimums we measured under the influence of Arma 3's Ultra quality preset, necessitating a closer look at frame-rates over time.

The lowest I'm willing to go is 20 FPS. Yet, even the stock-clocked boards maintain more than 25 FPS. The CryoVenom’s victory is a little hollow here.

Overclocked cards breeze through Battlefield 4’s High quality preset, with average frame rates exceeding 50 at 5760x1080.

Sags in the frame rate over time graph illustrate why averages aren't sufficient for evaluating performance. Even then, though, both overclocked cards maintain more than 40 FPS throughout our benchmark.

We like to call 30 FPS our lowest acceptable average, though this typically depends on the game and minimum frame rates. None of these configurations violate that guideline when we apply Battlefield 4's Ultra quality preset, fortunately.

When we break that chart down into frame rate over time, it indeed appears that 30 FPS is a good average target for this title. The air-cooled card drops to 25 FPS in its stock form, while the overclocked CryoVenom board demonstrates notably higher performance that doesn't fall under 30 FPS.

A single 1080p monitor doesn’t appear to strain any of these cards using Far Cry 3’s High detail preset. Higher resolutions benefit from overclocking, but the load still appears too light to trigger throttling on the air-cooled card.

Anyone who believes they need at least 50 FPS at 5760x1080 will probably want to overclock. VisionTek's CryoVenom card overclocks better.

We hit the Ultra preset and 5760x1080 before the stock air-cooled card begins to stumble. That’s also where the overclocked water-cooled card excels.

Charting out frame rate over time shows that the stock card drops below 20 FPS during our benchmark. This game has always appeared a little choppy, even at higher frame rates. So we'll take whatever performance we can get in order to make Far Cry 3 a more enjoyable experience. We'd stick to the liquid-cooled CryoVenom at this specific setting.

The stock CryoVenom R9 290 clock rates are identical to Sapphire's retail card with AMD's reference heat sink and fan, yet we often encounter a wider range of results on that board due to the effects of throttling. That doesn't appear to be a big problem in Metro: Last Light using the Medium detail preset, however.

VisionTek's small win still counts though, and its overclocking headroom remains slightly superior.

The High quality settings apply more serious stress in this game, forcing Sapphire's air-cooled board to back down a little more.

Overclocking remains an option, and you might even say it's required at 4800x900. The air-cooled card barely stays at or over 20 FPS during our test.

Despite its higher average frame rate, VisionTek's CryoVenom R9 290 also experiences frame rate drops.
AMD's latest Catalyst drivers override AMD's firmware settings, resulting in the Quiet and Uber modes performing pretty much the same (at least according to the results of my maddeningly-long time in the lab with these cards). The only difference I found was that the air-cooled card reached a lower idle state in Quiet mode, and I'm using that more conservative idle state in these power numbers.

The system with our air-cooled Radeon R9 290 only idles down to 102 W with the Uber firmware selected. Overclocked, the air-cooled card could have dropped to 105 W if we were using the Quiet BIOS and our more aggressive clock rate settings. Conversely, VisionTek's CryoVenom R9 290 doesn't exhibit the same behavioral difference.

The overclocked air-cooled card's higher idle temperature can be partly attributed to the fact that its power consumption doesn't drop as low with Uber mode engaged. I checked, and at the same clock rates, Quiet mode would have allowed it to drop to 22 °C over-ambient.
Overclocked, the retail card's lower temperature comes from a higher fan speed that adversely affects noise.

Less thermal throttling allows VisionTek's CryoVenom R9 290 to gain 2% performance over the air-cooled board running at the exact same frequencies. It also enjoys 6%-higher overclocked performance.

Using the stock Radeon R9 290 as a baseline, we find that the CryoVenom can improve efficiency by up to 19% under the effects of overclocking. It's even 11% more efficient than the stock card, using the same BIOS, at stock clock rates. Lower temperatures do wonders for efficiency.
Where do you find value in VisionTek’s air-cooled CryoVenom R9 290? Well, personally, I have three reasons to ditch air cooling when it comes to AMD's Hawaii-based Radeon cards. First, the reference design runs too hot, even at stock frequencies, and even after AMD increased its maximum fan speed setting, resulting in thermal throttling. If you go with a centrifugal cooler, you're going to have to accept lots of noise. Opt instead for a cooler armed with axial fans and you're dumping all of the card's heat back into your chassis, potentially affecting other components adversely. The CryoVenom R9 290 avoids those issues.
Ideally, you want to mount the radiator for your water-cooling setup with its fans facing outward, pushing waste heat right out of your case. Experienced builders can do this with low-speed fans that make high-end gaming machines extremely quiet. But even if we set aside the hit to portability and risk of a leak associated with liquid cooling, the CryoVenom could have a big hill to climb when it comes to calculating performance per dollar.

Because of market volatility and the fact that VisionTek's CryoVenom R9 290 is currently unavailable, we have to base the above chart on suggested retail pricing. We can’t know the actual street price of both boards until they're available simultaneously. We can only observe that a lack of supply is pushing most Hawaii-based cards between $150 and $200 higher than AMD's launch pricing.
The above chart also adds our $180 cooling system to what you'll pay for the liquid-cooled card. If you already have some old liquid-cooling gear laying around, rendering it "free", in a sense, you can push the CryoVenom’s overclocked and stock value ratings to 107% and 94%, respectively. And if you’re really tired of choosing between throttled performance, an overheated case, and excessive noise, that sounds like a bargain able to transcend basic frame rate calculations.

Value always looks worse when you're comparing individual components to each other. The rest of your platform isn't free though; we have to add in everything else supporting these parts. Let's say you're an enthusiast building a $2000 PC specifically for gaming, and a $100 upgrade boosts performance by 10%, you actually achieve a 5% value gain, regardless of how low the baseline part was priced.
If VisionTek can deliver on its promise to re-launch the CryoVenom R9 290 at $600, it will sell for close to the same price as a lot of air-cooled cards. That would make its value comparatively equal to comparative performance, with VisionTek's offering serving 2% more value at stock clock rates and 16% more value than stock when we overclock it, and 6% more value than our retail air-cooled sample when it's overclocked as well. But that $600 price tag is purely theoretical until VisionTek makes the CryoVenom R9 290 available for purchase.
So enough of the theory. In practice, a liquid-cooled product costs more than air cooling. Working out the value equation to favor water cooling is very difficult unless hammering away at heat and noise are your priorities. If you've used a Radeon R9 290 or 290X with AMD's reference heat sink and fan, those two attributes are no doubt important. Once you make the decision to water-cool, VisionTek would like to be the company selling you a turnkey solution. You get an extra $150 in cooling hardware for $150 at most (cool), without the work that goes into installing your own block (cool), and the hardware is still covered by a warranty (also cool). I think that's a good deal, even if the CryoVenom R9 290's one-year warranty is 50% shorter than what VisionTek gives you with its other cards.
You did catch that last bit, right? I mention it because the card we reviewed in PowerColor LCS AXR9 290X: Water Makes Hawaii Comfortable gives you two years of coverage.