Catalyst 10.6 and GeForce 257.21 performance analysis
As part of an ongoing feature for AlienBabelTech, this editor will be comparing the performance of 21 benchmarks with the current monthly Catalyst driver release versus the one from the previous month. Catalyst 10-6 was released last week and we naturally ask ourselves if there are any performance improvements for the Radeon 5870 series in games. Beside testing with a single HD 5870, we are also going to compare HD 5870 CrossFire performance with the latest Catalyst drivers.
As part of something new, since we are also benching GTX 480 for an unrelated review, we will give separate performance charts showing the performance changes in the latest WHQL GeForce drivers, 257.21 versus 197.75. We will not directly compare the Radeon to the GeForce performance as difference sets of criteria were used for the benching of each card. Since we are testing performance of Catalyst 10-6 for other articles and reviews, the HD 5870 was mostly benched with 4xAA and the GTX 480 was largely benched at 8xAA. However, you can see the relative performance change of each driver by looking at the charts.
We are going to test Catalyst 10.5 against Catalyst 10.6 using our current benchmark suite of 20 games plus one synthetic benchmark, Heaven 1.0. We also added Alien versus Predator benchmark and we are now using the full retail version of DIRT2’s DX11 path instead of our earlier DX9c path from the demo version in previous reviews. Our testing platform is Windows 7 64-bit using Core i7 920 at 3.80 GHz, 6 GB DD3, and our video cards are HD 5870 and GTX 480. All of our games are tested at 2560×1600, 1920×1200 and 1680×1050 resolutions and with maxed-out in-game settings plus 4xAA or 8xAA/16xAF and we use DX11/10/10.1 whenever possible.
Let’s get right to the test configuration and the tests.
Thanks for the benchmarks apoppin! It’s nice to see side by side comparisons between drivers versions.
Thanks for yet another driver review. I think I’ll wait for 10.7
I do miss the “%-Changed” to the right of the scores and an overall percentage gain or loss however, like you did in the 197.13 Driver-review. But this is still great, keep it up!
Nice review!
I’m curious though, why was such a high level of aa used on Crysis for the GTX 480? At 26fps, 8xq is not a realistic and playable amount of aa. No aa or 4x would be more appropriate I think.
Thank-you. The percentage of change might get added back in future driver reviews.
The reason that such a high level of AA was used on Crysis for the GTX 480 is that these results are being used in another review where most of the testing is at 8xAA or higher. Realistically, when I play Crysis at 2560×1600, I use 2xAA and lower some of the settings.
For a overall driver review, I prefer to set the bar very high and use similar (maxed out) settings. The purpose is to find the change from one driver revision to the latest.
he, why no article in ABT has BFBC2 bench, i mean c’mon, that’s all guys’ favourite today. and also empire: tw/ napoleon: tw, too.
one more question please, is there any difference forcing AA from ati Catalyst driver vs in-game AA? which one that you use? what about the general image quality? do they differ? THANX
I haven’t even got BFBC2 yet. As I am SP only, it is not high on my priority list. It also takes me awhile to add a game to my benching suite. I need to play it first and relate the benchmark to actual performance in gameplay. Metro 2033 is my next scheduled game for adding to my suite. Perhaps Leon will do some BFBC2 benchmarks for us.
Applying AA depends on the specific game; with most games there is no difference where it is applied – from the vendor’s CP or in-game. With some games – and specifically with Batman AA – Anti-aliasing is optimized for NVIDIA drivers in the game and can be set in the game’s control panel. However, with ATI drivers, you must force AA with CCC and it takes a bigger performance hit because it must now use a brute-force approach.
Set AA in game if possible first; if there is no option for it, then set it in your vendor’s CP.
Hello, am extremely interested in becoming a tech reviewer, problem is, i dont know how!!!! If you would please lead me in the right direction, maybe i can review producs for your website. Thank you~ luis campos
It looks like Call of Duty 4’s performance is still hosed.
Hi,
why there is no nvidia result in Arma2?
I have been using the ARMA2 demo and the results have been unreliable for this set of drivers. This game is such a difficult game on PC HW resources that I am dropping it altogether from my benching suite.
Would it be possible to do minimum FPS numbers in future comparisons?
There was one site ([H] possibly) which did a pre-release vs 9.12 vs 10.3 or so, and the minimum fps showed a large improvement over time.
It would be interesting to see minimum FPS changes as well as just average changes with new driver releases.
I actually do have the minimums and maximums and they mostly show no improvement over the averages; it is proportional in most cases.
The purpose of doing these comparisons is to get a quick idea of the driver changes in about 20 games before the drivers get old. If these charts were to be expanded, it would take much longer for them to be published.
However, there is a *series* of articles that I am working on right now – these will show the minimums and maximums as well as the averages and they will span from Catalyst 10-5 through 10-7 and we can compare to GeForce 197.75, 257.21 and also the next set.
You will be able to see for yourself beginning next weekend with a review of PowerColor HD 5870 PCS+ in Part One as we attempt to find the limits of HD 5870 architecture; and we continue right on through analysis of the Dragon Platform and explore the number of CPU cores and scaling on game performance. GTX 480, HD 5870 and HD 5870 CF will be featured and the minimums and maximums will be mostly shown in the graphs.