|MSI R6870 Radeon HD 6870 Video Card|
|Reviews - Featured Reviews: Video Cards|
|Written by Bruce Normann|
|Friday, 05 November 2010|
Page 7 of 18
Crysis Performance Tests
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 (DirectX 10) framework, but can also run using DirectX 9, on Vista, Windows XP and the new Windows 7. As we'll see, there are significant frame rate reductions when running Crysis in DX10. It's not an operating system issue, DX9 works fine in WIN7, but DX10 knocks the frame rates in half.
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 CryENGINE2 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.
Low-resolution testing allows the graphics processor to plateau its maximum output performance, and shifts demand onto the other 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, but it is sometimes helpful in creating a baseline for measuring maximum output performance. At the 1280x1024 resolution used by 17" and 19" monitors, the CPU and memory have too much influence on the results to be used in a video card test. At the widescreen resolutions of 1680x1050 and 1900x1200, the performance differences between video cards under test are mostly down to the cards themselves, but there is still some influence by the rest of the system components.
With medium screen resolution and no MSAA dialed in, the MSI R6870 is about half way between the HD 5830 and the HD 5870. Unlike many so-called TWIMTBP titles, Crysis has always run quite well on the ATI architecture, and the 6xxx series doesn't change that at all. The GTX 460 is not quite competitive here at current pricing, but don't look at the performance in this title as anything like a failure. It's just not a slam dunk victory for NVIDIA this time, unless you are looking at results for a massively overclocked version, which is always a potential threat with the GTX460.
Crysis is one of those few games that stress the CPU almost as much as the GPU. As we increase the load on the graphics card, with higher resolution and AA processing, the situation may change. Remember all the test results in this article are with maximum allowable image quality settings, plus all the performance numbers in Crysis took a major hit when Benchmark Reviews switched over to the DirectX 10 API for all our testing.
At 1900 x 1200 resolution, the relative rankings stay the same; the raw numbers just go down. With the increased load on the GPU, the R6870 still gets over the 30FPS hump, but only by about 10%. It takes more than any mid-range GPU can muster to play Crysis at high resolution with all the bells and whistles turned on, but that should be no surprise to anyone.
Now let's turn up the heat a bit on the ROP units, and add some Multi-Sample Anti-Aliasing. With 4x MSAA cranked in, the top cards lose about 5 FPS at 1680x1050 screen resolution but they manage to stay above the 30 FPS line. Compared to the ATI offerings, the MSI N460GTX Cyclone with out-of-the-box settings still can't convincingly edge out the HD 5830. The HD 6870 once again slots itself very nicely between the old HD 5830 and the HD 5870. Unlike the much-maligned 5830, the 6870 seems to have cleanly hit the exact middle of the performance target. I think in this case, the marketing people and the engineers had the good fortune to share a common vision about what performance was needed to kick off the new product line.
This is one of our toughest tests, at 1900 x 1200, maximum quality levels, and 4x AA. Only one GPU gets above 30 FPS in this test, and until recently it was the fastest single-GPU card on the planet, the Radeon HD 5870. In the middle ranges, the HD 6870 holds on to its spot as the performance leader, and also pushes out the GTX 460 as the value leader. Even a massive overclock on the GTX 460 won't come up even with the HD 6870 on this test. I ran the MSI GTX460 HAWK at 950 MHz core clock last month, and it only got 26 FPS at these settings.
Our next benchmark of the series is not for the faint of heart. Lions and tiger - OK, fine. Guys with guns - I can deal with that. But those nasty little spiders......NOOOOOO! How did I get stuck in the middle of a deadly fight between Aliens vs. Predator anyway? Check out the results from one of our toughest new DirectX 11 benchmarks in the next section.