It's nothing new to see AMD
arguing over benchmarks, but news that came across our desk today put a new twist on a very old battle. A few days ago, NVIDIA contacted us and inquired if we'd be interested in testing HAWX 2's stand-alone benchmark for evaluation, noting that tessellation effects were employed for better terrain realism. We said sure—a number of DX11
-enabled games and/or test programs make use of the effect currently. Synthetic benchmarks aren't much better at this point; Unigine might as well be called TesselationMark. NVIDIA's Reviewer's Guide also put great emphasis on the game's use of tessellation and Fermi's superior performance in that area. We read the guide, benchmarked the game, wrote down some numbers, and that was that—or so we thought. Imagine our surprise when we got an email from AMD, claiming that HAWX 2 was a very poor test and ought not to be used. The letter is quoted below:
"It has come to our attention that you may have received an early build of a benchmark based on the upcoming Ubisoft title H.A.W.X. 2. I'm sure you are fully aware that the timing of this benchmark is not coincidental and is an attempt by our competitor to negatively influence your reviews of the AMD Radeon™ HD 6800 series products. We suggest you do not use this benchmark at present as it has known issues with its implementation of DirectX® 11 tessellation and does not serve as a useful indicator of performance for the AMD Radeon™ HD 6800 series. A quick comparison of the performance data in H.A.W.X. 2, with tessellation on, and that of other games/benchmarks will demonstrate how unrepresentative H.A.W.X. 2 performance is of real world performance.
AMD has demonstrated to Ubisoft tessellation performance improvements that benefit all GPUs, but the developer has chosen not to implement them in the preview benchmark. For that reason, we are working on a driver-based solution in time for the final release of the game that improves performance without sacrificing image quality. In the meantime we recommend you hold off using the benchmark as it will not provide a useful measure of performance relative to other DirectX® 11 games using tessellation."
This argument over what is and isn't an accurate tessellation performance test underscores the fact that there are precious few titles to choose from currently. Fermi's tessellation performance is much stronger than AMD's at this point, but superior tessellation does not automatically translate into superior frame rates. Meanwhile, we're curious to see how this situation develops. Some of you might remember the ATI/NVIDIA Doom 3 vs. Half Life 2 knock-down battles from years ago. Back then, NVIDIA was nailed to the wall by the GeForce FX's performance in a relevant tech demo. Thanks in-part to Gabe Newell's evangelism of his new game engine, journalists took the terrible FX numbers as an indication that no one wanting to play HL2 should buy an NVIDIA card to do so.
Its status as a useful benchmark may be debatable but high tessellation does enhance in-game realism.
By the time the game actually shipped, NVIDIA had launched an entirely new GPU that didn't suffer from the same performance limitations as the earlier FX cards, but the damage had been done. Despite how the numbers had changed, NVIDIA was seen as a less-than-ideal choice for gamers wanting to play Half Life 2. In this case, we won't have to wait a year to find out how much tessellation really matters across a wide range of games—with a range of titles launching in the next few months we'll soon have a comprehensive view on how much tessellation performance affects gameplay.