Catalyst 11.1a vs. GeForce 266.58 Performance Analysis
As part of an ongoing feature for AlienBabelTech, this editor is comparing the performance of 25 benchmarks with the current 11-1a (hotfix) monthly Catalyst driver release versus the one from the previous month (12-10) . As an added bonus, timing has allowed us to also compare the Geforce 266.58 which was released last week with the 263.09 WHQL driver from last month.
Best of all, this driver performance evaluation will give a natural comparison between the performance improvements for GTX 500 and 400 series versus HD 6900 and 6800 series. Catalyst 11.1 and HotFix 11.1a was released last week and we naturally ask ourselves if there are any performance improvements for the Radeon 6000 series in games. Beside testing with a single HD 6870, we are also going to compare HD 6970 performance with the latest Catalyst drivers. For the Nvidia cards, we will test GTX 580 and GTX 460 so as to give a sampling of high and mid-performance cards and from the latest two series of cards.
The last performance analysis of Catalyst 10.12 was published here and Senior Editor Leon Hyman used his Q9550S and Radeon HD 5830. We are using the hotfix Catalyst 11-1a that was released simultaneously with WHQL 11-1 and promised even more performance improvements and a “Tessellation Slider”. Now we are using our overclocked Core i7-920 at 3.8 GHz and we use the same settings that we used for our Galaxy’s GTX 560 Ti GC – Introducing Nvidia’s Titanium Hunter. This time we are not testing with CrossFire nor SLI but will return to testing with it next month. The percentage of change from upgrading the drivers should remain about the same for any capable CPU platform.
We are going to test Catalyst 10.12 against Catalyst 11.1a using our current benchmark suite of 23 games plus two synthetic benchmarks, Heaven 2.0 and Vantage. We also test WHQL 266.58, just released last week against WHQL 263.09 from last month. Our testing platform is Windows 7 64-bit using Intel Core i7-920 at 3.80 GHz, 6 GB DD3, and our video cards are a single HD 6870 and a HD 6970 to represent the Radeons and a GTX 580 and a GTX 460 to represent the GeForce cards. All of our games are tested at two of these three resolutions: 2560×1600, 1920×1080, 1680×1050, with 16xAF, and we use DX11/10/10.1 whenever possible.
Let’s get right to the test configuration, each driver’s release notes and the tests.
When I look closely at Nvidia’s, its obvious nvidia has a good aim set for the driver results. While the overall average is very positive, there are a few cases that show a slight decline. This is typical. What i want to point out is the accuracy of Nvidia’s aim. If you look only at the set backs, a clear pattern emerges. First off the 460 doesnt really have any losses. They’re completely negligible with not a single fps loss. With the gtx 580 individual game results we can see that virtually every loss in fps is met with a 460 gain. There are six cases the 580 takes a step back or has a real loss. The 460 gains out of 5 of them. This was a fine tuning for the 560 launch. I am not criticizing at all, some may not take this the right way. There are great gains to be made for any nvidia user. But you cant get a driver that improves every case or game and trade offs are the norm. In this new driver it appears those trade offs seem to be more favorable to the gf104/gf114. I can say Nvidia strategically aimed their driver from the data you present us. Nvidia has a good driver team that knows well what it is doing.
Now for AMD,
Alls i gots to say is way to go! Its nice to see the fantastic results they achieved across a majority of games. I believe there will be more and more improvements like these as AMD finds better and better ways to utilize their new architectures.
All in all, it is worth it to everyone with any modern card to upgrade, this is some nice data,
thanx apoppin.
Looking forward to the comparison next month on multi-GPU driver scaling efficiency!
I tried the new drivers and most games run considerably better, also, it seems for me that the minimum FPS and some hiccups have been eliminated, gameplay experience seems to be a bit smoother.
Thanks for the effort fellows
Appopin,
The resolution listed for the Unigine engine is backwards. (1920×1200 should be 1680×1050, and vice versa)
Very nice article though.
-Martimus
Thank-you, Martimus
Fixed.
Thanks for putting in the time to test these. Can I ask you how much time it actually takes you to do this? You’re very thorough, which is greatly appreciated
You’re welcome! I am not really sure how long it takes to run all of these benches as I am continually benching.
For this article, I will just “guess” – 4 cards x 32 benches x 2 resolutions x 2 driver sets – probably well over 30 hours of continuous benching. And of course it takes a few hours to create the charts and write the article.
Nicely done!!
I see a very minor typo, you miss labeled the old Catalyst 12-10 instead of 10-12 just under the charts.
I’m very glad to have found this site and will continue to come back as I’m just starting the process of building a new system. My other one is 11+ yrs old..lol.
Oh BTW…I noticed that you have your work cut out for you