Summary: It's that time again! Right on the heels of ATI's CATALYST 3.2 driver release, we're here to bring you our results with ATI's latest driver. See what performance increases have been made and what issues have been fixed with the 3.2 driver. Is it worth the upgrade or should you stick with your current driver? Find out in this article!
Life in the driver development department of ATI must be pretty hectic, as the CATALYST driver team has provided a plethora of driver updates since their initial inception nearly a year ago. As a result of their efforts ATI was first to market with DirectX 9 drivers, in fact the CATALYST 3.0 driver release was delivered the same day as Microsoft’s DirectX 9.0 API. We found these drivers brought tremendous performance gains to RADEON 9500 users, offering double the performance in some cases. The CATALYST team then followed the initial DX9 driver release with their 3.1 driver, which also came with some performance enhancements.
So when ATI announced the RADEON 9800, 9600, and 9200 family of graphics processors earlier this month, we half expected that the CATALYST team was hard at work on those products, then we got the surprising news of CATALYST 3.2, just a week after ATI’s Game Developer’s Conference launch event in San Jose. Could they have developed a new top to bottom driver solution that quickly?
It turns out that these drivers aren’t quite the drivers we were expecting (the 3.2 driver was submitted for WHQL certification back in February), but this driver does offer official support for the RADEON 9800 PRO. In a sign of the CATALYST team’s dedication however, hints have been dropped that the upcoming CATALYST 3.3 driver will offer substantial performance gains for existing RADEON 9700 users. We’ll just have to wait and see if those reports hold true, today we’re here to evaluate CATALYST 3.2.
One quick glance over the release notes and you’ll see the emphasis ATI has placed on resolving compatibility issues with this driver release. Thankfully we can report that the 3.2 driver is the most solid 3.x driver release we’ve come across. For instance, the stability issues we’ve had with the RADEON 8500 and RADEON 9000 in Jedi Knight II (while operating in high quality mode) are gone. We were able to run our complete suite of tests (and then some) with this combination without a single hitch. Something we haven’t been able to say since CATALYST 2.5 last year.
Unreal Tournament 2003
While we see some slight differences with the RADEON 9700, 9500, and 8500 at lower resolutions in flyby, for the most part performance is the same among these graphics cards regardless of the API. Therefore, if you feel uncomfortable upgrading to DirectX 9, feel free to stick with DX8 and reconsider the upgrade once DX9 content becomes available.
As ATI promised, CATALYST 3.2 offers performance increases for RADEON 8500 and RADEON 9000 users in 3DMark 03. At 800x600x32, we’re looking at a score that is 19% higher for the RADEON 9000, and 21% for the RADEON 8500! Meanwhile, performance is roughly the same for the RADEON 9500 and 9700 families. Lets take a closer look at the results.
3DMark03 – Wings of Fury
3DMark03 – Battle of Proxycon
3DMark03 – Troll’s Lair
3DMark03 – Mother Nature
Not only do we get a performance boost for the DX8 tests 2 and 3 for RADEON 8500/9000 users, we also see that test 1’s (Wings of Fury) score is higher. Performance is relatively unchanged for the DX9 cards.
Serious Sam 2 - OpenGL
Serious Sam’s performance is roughly unchanged, there are some slight cases where performance increases, but the difference is well within the margin of error.
Quake III - High Quality
Like Serious Sam, we see some slight variances in performance, the 9500 PRO actually drops a few frames at lower resolutions, but the declines weren’t as severe as what we saw with CATALYST 3.1.
Comanche 4 demo
The RADEON 9700 PRO slips in Comanche 4, we see a 5% performance decline in lower resolutions. This is a bit of a surprise considering the CPU-limited nature of this benchmark, the other cards offer roughly the same performance regardless of driver.
Unreal Tournament 2003 - flyby
Unreal Tournament 2003 - botmatch
Our flyby test results are similar regardless of CATALYST driver used, but when bots are introduced we do see performance declines at 1280x1024. Looking back at our CATALYST 3.0 performance however, it looks like our CATALYST 3.1 results at that resolution were an anomaly, as the 3.2 performance is similar to what we saw with 3.0.
Jedi Knight II – High Quality
Unreal Tournament 2003 Flyby
Unreal Tournament 2003 Botmatch
Quake III – High Quality
Quake III – High Quality
Quake III – High Quality
DirectX 8 card owners
While we did see the performance improvements ATI has claimed in 3DMark 03, we’ve got to wonder if this is worth the risk of updating your display driver. If your games and other applications are running fine with the driver you’re using now, is it really worth the risk of upgrading to CATALYST 3.2 for the performance increase in a synthetic benchmark?
DirectX 9 card owners
For RADEON 9500/9700 users, it looks like the CATALYST 3.2 driver has been released solely for compatibility reasons. ATI has addressed the stuttering issues with Freelancer and Rallisport Challenge, and with 2D applications running in the background in general, but performance enhancements won’t be found here. If you’re looking for more performance, you’ll have to wait for CATALYST 3.3, whose release schedule has been accelerated to bring it to our PCs sooner. Perhaps even as an open beta!
|© Copyright 2003 FS Media, Inc.|