|
|
Advertisement:
|
|
ATI Radeon 9700 Pro Video Card |
|
Join the community - in the OCAU Forums!
|
UT2003, Drivers
Unreal Tournament 2003 Demo Benchmark
The UT2003 Demo has only very recently been released and has a built-in benchmark. The engine used in this game will probably be used as the basis for a few more games in the future. Note that there are no benchmarks from the SiS648 testbed as the R9700 would crash during this benchmark as described in the "AGP8X?" section on page 2.
In this one the results are a little stranger. The cards fare almost identically in the "Botmatch" sections of the benchmark - I assume this is because there's not actually a lot going on visually in this part of the benchmark. It mostly consists of static views with small animated figures fighting. With only small sections of the screen requiring updates, this benchmark is likely CPU limited rather than video-card limited, so we see similar performance with either card. In the Flyby section, which involves swooping through canyons etc, the entire screen is constantly changing. In this video-intensive benchmark we see once again, as the resolution rises, so does the R9700's lead over the GF4 Ti4600 extend.
Drivers
ATI's drivers have traditionally been the letdown with their products. Bloated, feature-heavy control panels, dissapointing performance and flaky compatability have plagued them literally for years. Recently, they have changed their approach to drivers with their "ATI Catalyst" system. This splits the driver package into smaller chunks - the most important being a basic driver portion updated fairly frequently, similar to the DetonatorXP setup from nVIDIA (but of course, not updated as regularly as nVIDIA's almost daily "leaked" betas). There's also a Control Panel which enables control of a lot more options. One feature of the control panel is a slider for DirectX and another for OpenGL. These let you choose presets along a scale from "Performance" to "Quality". You can individually turn on things like Anti-Aliasing, Vertical Sync etc, or simply use this slider.
Out of fairness, benchmarking was done with the sliders at the "Default" setting. This is how I normally approach benchmarking because a review environment doesn't allow the time to investigate the effect of each setting - and we must also consider comparable settings on the Ti4600 we're comparing it to. Obviously, we ensure VSync and FSAA are turned off for normal benchmarking, and the "Default" setting has these disabled or set to "Application Preference", and we have them turned off in the various benchmarks.
So, just be aware that there may be more speed available for those who want to tweak the drivers, possibly at the expense of visual quality. How much more? On the AthlonXP 1600+ testbed, "Performance" achieved a default 3DMark2001SE score of 11489, while the default "Balanced" option, as we used for benchmarking, scored 11385. This difference of 100 points out of 11,000 can be safely ignored, but may be larger on more powerful systems or more texture-intensive benchmarks - the sliders seem to only change "Texture Preference" between Performance and Balanced. The "Quality" (which includes 4X FSAA and 16X Anisotropic Filtering) dropped the score to 7984.
Did ATI's bugbear of dodgy drivers appear? Partly. As mentioned earlier, I found the UT2003 Demo's benchmark would hang when entering the first "Botmatch" phase, the third phase overall. This happened only in AGP8X mode on both KT400 (AthlonXP) and SiS648 (P4) systems. On KT333 in AGP4X it worked fine, so perhaps it's not so much a driver issue as an AGP8X issue, although I've not heard other reports of KT400 AGP8X problems with these cards - reports of problems with SiS boards and early-revision R9700's are more common and ATI say they are working on a fix.
One annoying problem did occur though, which brought flooding back nightmares of trying to get ATI's Rage Pro XL working under WinNT4 years ago (yes - the same nightmares that recurred during the Radeon 8500AIW review I did for PC PowerPlay mag). It happens on a freshly-built machine with Win2K SP3, DirectX 8.1b and appropriate motherboard drivers loaded. Installing the Catalyst drivers is fine, selecting 1024x768x32 is fine and the machine can happily be used and will reboot etc. However, selecting (seemingly) any refresh rate other than 60Hz would work, but the machine when rebooted would present a corrupted login screen which made logging in impossible. Rebooting in VGA mode, selecting 60Hz and rebooting fixed it. This behaviour was reproduced on KT333, KT400 and SiS648 test machines over several reinstalls of Win2K.
So, still some work to be done on drivers, it seems. These definitely seem less problematic than when I was testing the Radeon 8500, so ATI are at least moving in the right direction.
|
|
Advertisement:
All original content copyright James Rolfe.
All rights reserved. No reproduction allowed without written permission.
Interested in advertising on OCAU? Contact us for info.
|
|
