Crysis Benchmark Results

Crysis uses a new graphics engine: the CryENGINE2, which is the successor to Far Cry's CryENGINE. CryENGINE2 is among the first engines to use the Direct3D 10 (DirectX10) framework of Windows Vista, but can also run using DirectX9, both on Vista and Windows XP.

Roy Taylor, Vice President of Content Relations at NVIDIA, has spoken on the subject of the engine's complexity, stating that Crysis has over a million lines of code, 1GB of texture data, and 85,000 shaders. To get the most out of modern multicore processor architectures, CPU intensive subsystems of CryENGINE 2 such as physics, networking and sound, have been re-written to support multi-threading.

Crysis offers an in-game benchmark tool, which is similar to World in Conflict. This short test does place some high amounts of stress on a graphics card, since there are so many landscape features rendered. For benchmarking purposes, Crysis can mean trouble as it places a high demand on both GPU and CPU resources. Benchmark Reviews uses the Crysis Benchmark Tool by Mad Boris to test frame rates in batches, which allows the results of many tests to be averaged.

The very first thing we discovered in the low-resolution tests was how seemingly poor both of our multi-GPU products performed. The Gigabyte GeForce 9800 GX2 was the lowest of the group at 1280x1024, matched in performance to a single Sapphire Radeon HD 4850 or GeForce 9800 GTX. The CrossFireX set of 4850's suffered the same rapid-response overhead bottleneck and performed almost the same as a single Radeon 4870 or GTX 260. To be fair, none of these video cards will probably ever realistically see game-play at a resolution this low, so this performance illustrates how high-end GPU power can be cut short if the monitor (resolution) doesn't match it.

Low-resolution testing allows the graphics processor to plateau maximum output performance, which thereby shifts demand onto the system components. At the lower resolutions Crysis will reflect the GPU's top-end speed in the composite score, indicating full-throttle performance with little load. This makes for a less GPU-dependant test environment, and is helpful in creating a baseline for measuring maximum output performance in the next few test results. At the 1280x1024 resolution used by 19" monitors, our results show that performance is beginning to really drop despite the small difference is pixels drawn. In terms of general performance, all of these products maintain the same performance ratio as before, except for the 9800 GX2 which seems to beneficially hold its ground.

The CrossFireX set of HD 4850's is going to soon reach it's limit, as they are in last gear and the 9800 GX2 is still shifting up the tree. Sapphire's Radeon HD 4870 flexes GDDR5 muscle at 1680x1080, running alongside the overclocked ZOTAC GeForce GTX 280 AMP! Edition ZT-X28E3LA-FCP with only a 5% difference. For widescreen users, our benchmarks below indicate that the ATI Radeon HD 4870 matches the performance of NVIDIA's GeForce 9800 GX2 video card, although the 4870 stops delivering post-processing effects at 8x AA and the 9800 GX2 can reach 32x AA if the application supports it. The CrossFireX set of HD 4850's appear to be matching performance of a single Radeon HD 4870 video card; at least as far as Crysis is concerned. Testing in high-pressure Crysis also seems to have effected the Sapphire Radeon HD 4870 X2 we tested, which finally succombs to the AMP!'ed GeForce GTX 280.

Reading the results of our 1920x1200 resolution tests using SOYO's DYLM26E6 monitor for testing, Crysis forced 2.3 million pixels to be processed by our graphical test products. At our highest widescreen resolution, the Sapphire Radeon HD 4870 performs on par with the GTX 260. Despite what 3dMark06 previously reported, the CrossFireX set of Radeon HD 4850's is not king; as a close struggle between the GeForce GTX 280 and Sapphire Radeon HD 4870 X2 comes to a head. If only by a small difference, the overclocked ZOTAC GTX 280 enjoys a 5% (2 FPS) lead over the GeForce 9800 GX2, which also seem keep its own 8% (3 FPS) lead over the CrossFireX set.

At the end of our Crysis testing, neither the GeForce 9800 GX2, GTX 280, or CrossFireX HD 4850 set could touch the multi-GPU Radeon HD 4870 X2. The new king is earning his crown, but as the pressure gets higher the challengers rise to the occasion. Before we leave Crysis though, I decided to include a look at post-processing performance with 4x AA enabled at the 1680x1050 and 1920x1200 widescreen resolutions. The chart below shows the average frame rate performance with 4x Anti-Aliasing enabled.

At 1680x1050, the Radeon HD 4870 is no match for the GeForce GTX 260 it's trailed thus far. Additionally, the CrossFireX set of 4850's falls 28% short of reaching the Gigabyte GeForce 9800 GX2. As if it was really unexpected, ZOTAC's GTX 280 AMP! Edition comes out (way) on top with an impressive 18% lead over the Sapphire Radeon HD 4870 X2 video card. But once the Honeywell 22-Inch LCD was swapped out, I began testing at 1920x1200 and the differences were made very clear.

At 1920x1200 the multi-GPU CrossFireX array of 4850's performed almost identically to a single Radeon HD 4870, while the GeForce 9800 GX2 seemed to fall flat. A heavily overclocked GTX 280 stills shows a prevailing strength, as it produces a 12% lead over the dual-GPU 4870 X2.

With only a small dose of anti-aliasing added to Crysis, there are very few products that would make for playable frame rates. Our Island timedemo mixes a some beach and water views so it's going to be on the high side of frame rates when compared to actual game play, but as you can see the Radeon products do very well when post-processing effects are added. Sapphire's hot-potato HD 4870 matched performance of our 4850 CrossFireX set, but trailed behind the GTX 280 by 52%. It appears that the G92 GPU really stands out like a sore thumb against the newer graphics processors, which yield far better frame rates in our Crysis testing.

In our next section, Benchmark Reviews tests with Unreal Tournament 3. Read on to see how a blended high-demand GPU test with low video frame buffer demand will impact our test products.

<< Previous Page - Next Page >>