Although all of the R9 270X’s and R9 280X’s that we have covered have been extremely exciting, especially the Toxic cards, to date we hadn’t had a chance to take a look at the newest cards from AMD. Today we finally have that chance. Sapphire sent over their R9 290 Tri-X for us to put through our tests. The Toxic cards have dominated in our other reviews so I am extremely interested in how well the 290 Tri-X will perform with it sharing a very similar cooling design. With high expectations, let’s jump into it and let the numbers do the talking.
Product Name: Sapphire R9 290 Tri-X
Review Sample Provided by: Sapphire
Written by: Wes
Pictures by: Wes
Specifications |
|
Display Support |
Supports up to 3 display monitor(s) without DisplayPort 4 x Maximum Display Monitor(s) support |
Output |
1 x HDMI (with 3D) 1 x DisplayPort 1.2 2 x Dual-Link DVI-D |
GPU |
1000 MHz Core Clock 28 nm Chip 2560 x Stream Processors |
Video Memory |
4096 MB Size 512 -bit GDDR5 5200 MHz Effective |
Dimension |
305(L)X113(W)X38(H) mm Size. 2.2 x slot |
Software |
Driver CD |
Accessory |
8 PIN to 4 PIN Power Cable 6 PIN to 4 PIN Power Cable HDMI 1.4a high speed 1.8 meter cable(Full Retail SKU only) |
Packaging
Having already seen the 270X and 280X Toxic cards, I wasn’t surprised at all when the 290 Tri-X came in with its eye catching orange packaging. There are a few differences though. First the image on the cover is different than the other cards. You also have more feature icons along the bottom, due to this being a higher end card and of course up in the top right corner you have the model name. Around on the back you have a few more icons representing some of the cards features, like the Tri-X cooling and the OC. This time there is a little more information included with each icon.
Inside you could say that Sapphire went a little overboard protecting the 290 Tri-X. Not only is the card packed up in inches of foam and covered with foam on top, but they also wrapped it up in a padded static protective bag as well. Inside up under the foam you will find another box for all of the accessories included with the Tri-X. Unlike most other cards, they include a free HDMI cable as well as a Molex to six pin adapter and a double Molex to 8 pin adapter. For documentation you get a quick installation guide, a card with information on how to register your card, and a driver installation CD.
Card Layout and Photos
The overall look and design of the R9 290 Tri-X from Sapphire is very similar to the R9 280X Toxic. You have a black and orange design that really catches your eye. Unlike previous models, the middle fan matches the other two fans rather than being the runt of the family. That mean Sapphire did step up the cooling slightly when going from the 280X to the 290. Each of the three fans is a 9 blade design and their glossy black finish goes perfectly with the black trim on the top and bottom of the card.
The Tri-X cooler on the R9 290 Tri-X is similar to the previous models in its cooling design. You have a triple fan design (this time with three fans of the same size) that blows air down over heatsinks. This is different from how a reference cooler would cool, reference designs blow across the card, not down at the PCB. This does have the side effect of warmed air venting from the top and bottom of the 290 Tri-X, like any other aftermarket cooled video card. If you look from the underside you can see the heatpipes that pull the heat off of the GPU itself and distribute the air across the heatsinks for the best cooling. Also up on the top edge you have a Sapphire logo like on the other cards, but this time around it isn’t backlit. It’s interesting that the model up from the 280X is missing features that both it and the 270X had.
Sapphire didn’t change things up for power from the reference design this time around. We have the same 8-pin + 6-pin combo that the 280X had as well as the reference 290. This should still be more than enough to push things. On the top edge of the PCB, over on the left side you also have a small switch to be able to switch from the standard BIOS to the UEFI BIOS should your PC support it. Interestingly enough the 290 Tri-X, like the 290X, does not have a Crossfire bridge connection at all. This is because with their new cards AMD went with a new design that takes advantage of the additional bandwidth that PCIExpress 3.0 provides and lowers latency in triple and quad crossfire configurations where data used to have to transfer across multiple busses.
Sapphire went with the standard dual DVI, full sized HDMI, and full sized DisplayPort combo on the end of the 290 Tri-X. This is a great combination that seems to cover almost every possible option except VGA. Normally you can still run a DVI to VGA adapter but in this case you don’t have that option because you don’t have analog pass-through on either of the DVI ports. Sapphire did pack in as many cooling vents as possible around the four connections although most of the air that the Tri-X cooling is pushing will still have to vent in other directions.
One area where Sapphire did seem to skimp on compared to their Toxic cards was around on the back of the card. Both of the Toxic cards I recently tested had very sharp backplates that both show off the card name and also protect the card itself. It’s interesting that Sapphire opted to not do the same on the R9 290 Tri-X when it shares an otherwise similar design. You do still have a black PCB though. You can also see how the Tri-X cooler does still hang over the PCB on the end as well.
Our Test Rig and Procedures
Our Test Rig |
|
CPU |
|
Memory |
|
Motherboard |
|
Cooling |
|
Power Supply |
|
Storage |
|
Case |
Our Testing Procedures |
|
Bioshock Infinite |
Using the Adrenaline Action Benchmark Tool we run Bioshock Infinite on the “Xtreme” quality setting. This has a resolution of 1920x1080, FXAA turned on, Ultra Texture detail, 16x Aniso Texture Filtering, Ultra Dynamic Shadows, Normal Postprocessing, Light Shafts on, Ambient Occlusion set to ultra, and the Level of Detail set to Ultra as well. |
Tomb Raider |
Using the Adrenaline Action Benchmark Tool we run Tomb Raider on the “Xtreme” quality setting. This has a resolution of 1920x1080, Exclusive Fullscreen turned on, Anti-Aliasing set to 2xSSAA, Texture Quality set to Ultra, Texture Aniso set to 16x Aniso, Hair Quality set to TressFX, Shadow set to Normal, Shadow Resolution on High, Ultra SSAO, Ultra Depth of Field, High Reflection quality, Ultra LOD scale, Post Processing On, High Precision RT turned on, and Tessellation is also turned on. |
Hitman: Absolution |
Using the Adrenaline Action Benchmark Tool we run Hitman: Absolution on the “Xtreme” quality setting other than the MSAA setting is turned down from 8x to 2x. That setting puts the resolution at 1920x1080, MSAA is set to 2x, Texture Quality is set to High, Texture Aniso is set to 16x, Shadows are on Ultra, SSA is set to high, Global Illumination is turned on, Reflections are set to High, FXAA is on, Level of Detail is set to Ultra, Depth of Field is high, Tessellation is turned on, and Bloom is set to normal. |
Sleeping Dogs |
Using the Adrenaline Action Benchmark Tool we run Sleeping Dogs on the “Xtreme” quality setting. That means our resolution is set to 1920x1080, Anti-Aliasing is set to Extreme, Texture Quality is set to High-Res, Shadow Quality is High, Shadow Filter is set to high, SSAO is set to High, Motion Blur Level is set to High, and World Density is set to Extreme. |
F1 2012 |
We use the built in benchmark for F1 2012. We set our resolution to 1920x1080 and then use the “Ultra” setting. |
Total War: Shogun 2 |
Direct X11 Benchmark High setting |
Crysis 2 |
Using Adrenaline Crysis 2 benchmark. 1080p, 4x Anti-Aliasing, DX11, Laplace Edge Detection Edge AA, on the Times Square map, with hi res textures turned on. |
Battlefield 3 |
Using Fraps with the game set to Ultra settings with 4x MSAA Antialiasing Deferred, 16X Anisotropic Filter, at 1920x1080. |
Sniper V2 Elite |
1920 x 1080 resolution, graphics detail set to ultra |
Dirt Showdown |
1920 x 1080 resolution, 4x MSAA multisampling, Vsync off, Shadows: ultra; Post Process: High; Night Lighting: High; Vehicle Reflections: Ultra; Ambient Occlusion: Ultra; Water: high; Objects: Ultra; Trees: Ultra; Crowd: Ultra; Ground Cover: High. |
Metro Last Light |
Using the included benchmark tool. The settings are set to 1920x1080, DirectX 11, quality is set to very high, Texture filtering is untouched at 4x, and motion blue is set to normal. SSAA is unselected, PhysX is unselected, Tessellation is off. We run through scene D6 three times to get an average score. |
Synthetic Benchmarks |
For video cards our synthetic benchmarks are limited to 3DMark Vantage 2011, and 3DMark 2013 (AKA 3DMark). In 3DMark Vantage 2011 we run both performance and extreme benchmarks. The same goes for the most current version of 3DMark, we run through Fire Strike on standard and extreme settings. |
Unreal Heaven Benchmark 4.0 |
Using the “Extreme” preset |
Unreal Heaven Benchmark 4.0 heat testing |
We run through Unreal Heaven at 1080p for 30 minutes to test in game heat performance and noise output of the card while under load. |
Power Usage |
Using Unreal Heaven Benchmark 4.0, we get our “load” power usage number from the peak power usage during our test. We get our numbers from a Kill-A-Watt connected to the test benches power cord. |
Noise Testing |
Our Noise testing is done using a decibel meter 3 inches away from the video card on the bottom/fan side of the card. We test an idle noise level and then to get an idea of how loud the card will get if it warms all the way up we also turn the fan speed up to 100% and test again. The 100% test isn’t a representation of typical in game noise levels, but it will show you how loud a card can be if you run it at its highest setting or if it gets very hot. |
Cooling, Noise, and Power
Before we get into the synthetic and in game benchmark results I would like to take a look at the other measurable aspects of the R9 290 Tri-X. To start things off, I put the R9 290 through our power consumption test that compares both the idle and in game performance wattages against other cards to see how much power you will truly need to run the card. As you can see below, even though the R9 290 Tri-X is up fairly high on the list it is actually surprisingly close to the R9 280X Toxic. The load power consumption falls in between the R9 280X Toxic and the GTX 780 Ti with a peak usage of 482 watts. You may see higher in some other games, but this is the peak wattage seen while running the Heaven 4.0 benchmark on our 6 core test bench that includes an SSD and hard drive as well as water cooling.
Moving on to noise testing, I ran the R9 290 through decibel tests at idle, 50% fan speed, and 100% fan speed. Our 50% fan speed results are still new so I don’t have a full graph to include with them, but I do have results against 28 other video cards. The idle results are fairly moot really; most of the cards come in within a few decibels. 100% fan speed testing on the other hand has a wide range of results. The R9 290 Tri-X comes up fairly close to the top of the charts for noise, in the bad way. This is to be expected to a point really when you keep in mind it does have three fans on the Tri-X cooler. Even still, it still performed better than a few other dual and even single fan designs as well so Sapphire isn’t doing too badly with their triple fan design. For those who are curious at 50% fan speed the R9 290 Tri-X runs at a much quieter 68 decibels, this is more like what you should expect most of the time.
The last thing we have to look at with Sapphires triple fan Tri-X cooler is its actual cooling performance. To do that I ran the R9 290 Tri-X in the Unreal Heaven benchmark 4.0 until it reached a peak temperature. The R9 290 Tri-X warmed up to 72 degrees. This isn’t really all that much farther from the R9 280X Tri-X’s results. What is most impressive is when you compare the overclocked R9 290 Tri-X next to the various reference cards that we have tested in the past.
Synthetic Benchmarks
My first taste of the performance of the R9 290 Tri-X was running it through our synthetic benchmark suite. That includes both extreme and performance benchmarks in 3DMark 11 and the new 3DMark Fire Strike benchmark. I also include the Unreal Heaven Benchmark 4.0 in this lot because it is a synthetic benchmark based on the same engine that some games run on. With that said, I was extremely impressed with the performance of the R9 290 Tri-X. In both Fire Strike benchmarks the R9 290 Tri-X came in second only to the GTX 780 Ti from Nvidia. That means it out performed the GTX 780 and even the overclocked GTX 780 as well. In 3DMark 11 the results were similar, but with more of the multi card tests sitting above both the 780 Ti and the R9 290 Tri-X. In Unreal Heaven Benchmark 4.0 the overclocked GTX 780 did slightly pull ahead. All in all I was very impressed with my first round of testing.
In Game Benchmarks
Synthetic benchmarks are great and all, but in game testing will give you a better idea of what to expect from the R9 290 Tri-X. Because of that I ran it through our 11 game benchmark suite that includes a variety of today’s most popular games. We know the R9 290 Tri-X has some power, but how well did it handle all of the games? Well we consider 60 FPS to be the ideal FPS goal, out of the 11 games tested the R9 290 Tri-X ran well over 60 FPS on 10 of them. The single game that it didn’t break 60 FPS on had 58 FPS, it might not be 60 but you can’t get much closer than that. When compared to the GTX 780, GTX 780 Ti, and the R9 280X’s on the charts the R9 290 Tri-X performed similarly to what I saw previously in the synthetic benchmarks. In nearly every test the only single card that out performed it was the GTX 780 Ti with a few cases where the R9 290 Tri-X topped the charts as well as a few games that favor Nvidia more that had the GTX 780 ahead of it. One thing is for sure, the R9 290 might not be AMDs top card, but it still held its own against the competition.
Overclocking
When it came time to overclock the R9 290 Tri-X, I was surprised to dig in and find a completely new way over overclocking from AMD. The 290 cards use a system that is similar to what Nvidia have been doing recently with their offsets. The difference here being you don’t set an overclock clock speed offset, you set a percentage overclock. I’ll be honest I wasn’t a big fan of this, simply because this is even harder to see what results you are shooting for than with Nvidia’s method. You have a graph (as pictured below) that lets you set your clock speed overclock percentage as well as the power overclock that you will be giving the card. Because of the new system it took me a little while to adjust. I did however still run through slowly overclocking the GPU on the R9 290 Tri-X from its already overclocked clock speed. It didn’t take too long before I hit its cap without adding power, then I added some power and continued on. In the end I was able to bump the card up to 12% while still being able to pass our test.
When it came to memory overclocking I did the same thing. Things went smoothly until I attempted a 30% overclock. At that point the screen went black. After rebooting the issue was gone and I attempted to drop to back down to the 25% that I had tested before with good luck. This time it failed again with a black screen. After that attempt I ran into an issue where after rebooting, everything would run fine until AMDs catalyst booted up and it would automatically overclock the R9 290 Tri-X back up to the memory clock speed that was causing me issues. Because of this I discontinued my overclocking with concerns that if I tried to push the card any harder in our combined testing that I might cause damage to the card. My recommendation is to go crazy with the overclocking on the GPU, as you can see it gives you a nice performance increase. But don’t push the memory clock too far as it might be a little more fragile. More importantly the memory overclocking didn’t give me an improvement in performance that would justify risking damaging the R9 290 Tri-X.
GPU Clock Speed Overclocking |
||||||||
Overclock % |
Power % |
Resulting Clock Speed |
Pass/Fail |
FPS Result |
Notes |
|||
0% |
0% |
1000MHz |
Pass |
71.64 |
||||
10% |
0% |
1100MHz |
Pass |
76.57 |
||||
15% |
0% |
1150MHz |
Fail |
N/A |
||||
12.5% |
0% |
1125MHz |
Fail |
N/A |
||||
12% |
0% |
1120MHz |
Pass |
78.73 |
||||
12% |
10% |
1120MHz |
Pass |
78.89 |
||||
15% |
10% |
1150MHz |
Fail |
N/A |
Artifacts then Driver Crash |
|||
13% |
10% |
1139MHz |
Fail |
N/A |
Driver Crash |
|||
Memory Clock Offset Overclocking |
||||||||
Overclock % |
Resulting Memory Clock Speed |
Pass/Fail |
FPS Result |
Notes |
||||
0% |
1300MHz |
Pass |
71.64 |
|||||
5% |
1365MHz |
Pass |
71.80 |
|||||
10% |
1430MHz |
Pass |
71.81 |
|||||
15% |
1495MHz |
Pass |
72.12 |
|||||
20% |
1560MHz |
Pass |
72.15 |
|||||
22% |
1586MHz |
Pass |
72.14 |
|||||
25% |
1625MHz |
Pass |
72.29 |
|||||
30% |
1590MHz |
Fail |
N/A |
|||||
Overall and Final Verdict
Considering the last few cards I have taken a look at have been R9 270X’s I was especially excited for the R9 290 Tri-X to come in. On top of finally getting a chance to see the performance of one of AMDs newest cards, I was very happy with the last Toxic card I had in the office. The R9 290 Tri-X from Sapphire really caught my eye once again with its orange styling and Tri-X triple fan cooling design. When taking a tour around the card I was a little disappointed to see a few of the key features from the R9 280X Toxic not included, specifically the backpate and the backlit Sapphire logo on top. It seems a little weird to me that they would drop features when moving up in performance, but the Toxic cards are considered a step up from the Tri-X. Even so Sapphire did a great job of adding in value in other places like with the free HDMI cable that they bundled in with the card.
My other concerns were with the slightly noisy fans when turned all the way up to 100% and the memory overclocking issues I ran into. Honestly at idle and 50% the noise performance was great. As long as you aren’t manually cranking the fans up you shouldn’t require earplugs. That is good because otherwise the Tri-X cooling did a good job of keeping the card cool, especially when dominating in nearly all of our performance tests. The only card that outperformed the R9 290 Tri-X was the GTX 780 Ti, and let’s be honest it is in a different price range currently.
At $449.99 the R9 290 Tri-X isn’t a cheap card by any means, but it did perform up to its price. You are paying a nice premium over the reference design, specifically $50 over the reference R9 290. At that price difference it would have been nice to have the backplate included in my opinion. But the fact is, right now these cards are selling for WELL over their MSRP because the demand is there. That shows you how bad people want them, so if you have the chance to pick one up for your next build I would recommend it.