We began our testing for this year's updated charts with the newest and fastest graphics cards available from ATI and Nvidia. Right now, you can compare 32 retail products and graphics chip classes to one another; detailed performance results are available from 30 different gaming benchmark environments.
From here on out, the graphics charts will be updated on a monthly basis as we introduce additional cards (reference and retail) at every price point. All our benchmarks here are new, whether you're looking at the Mainstream or High-End categories.
To provide the best possible performance picture, all of the graphics products are measured first at their standard clock rates. This year we aren't restricting these charts to reference models: you'll find commercially-available products, special editions, and overclocked versions from all the well-known vendors for comparison, too.
This is how Tom’s Hardware tracks the graphics card market closely and carefully, where vendors are always changing clock rates, cooling solutions, and various other design details so that they can improve upon the performance available from basic reference cards.
Our new Benchmark Suite includes another large collection of different games and 3D engines. Our goal is to represent a broad mix of real-time strategy, simulation, role-playing, and FPS games in these results. For some time now, state-of-the-art graphic cards have offered enough performance for fluid game play on most monitors, even at resolutions up to 1920x1200 (a bit higher than standard HD resolution). That’s why we’ve raised the bar in this year’s charts for our test resolutions: High-End now starts at 1680x1050 resolution with anti-aliasing (AA) enabled. That’s because lower resolutions run smoothly across the board, and are more likely to be limited by CPU (rather than GPU) performance.
At a resolution of 1920x1200, the biggest differences become visible, so we feature three different test variations for this case. First, there's a test with anti-aliasing turned off (so that we can record the most playable frame rates). We also test with 4xAA to improve image quality somewhat (and to better tax the more powerful graphics cards). Finally, we test with 8xAA to most accurately distinguish between otherwise-subtle differences in clock speeds, graphics chip classes, and frame buffers on high-end cards.
|Benchmark Suite 2009||DirectX||Standard||Quality||8xAA||3D Engine|
|Fallout 3||DX9||0xAA+0xAF||4xAA+8xAF||8xAA+15xAF||Gamebryo (Oblivion)|
|Far Cry 2||DX10||0xAA+0xAF||4xAA+8xAF||8xAA+16xAF||Dunia Engine|
|F.E.A.R 2||DX10||0xAA+0xAF||4xAA+8xAF||LithTech Jupiter Extended (EX)|
|Left 4 Dead||DX9||0xAA+0xAF||4xAA+8xAF||8xAA+16xAF||Source Engine (Halflife 2)|
|The Last Remnant||DX10||0xAA+4xAF||0xAA+4xAF||Unreal 3|
|Tom Clancy's Endwar||DX10||0xAA+0xAF||4xAA+8xAF||Unreal 3.1|
|Tom Clancy's H.A.W.X||DX10||0xAA+0xAF||4xAA+8xAF||Ubisoft Bukarest (Blazing Angels)|