For game tests high resolutions/high detail settings were used in the hopes that even for highly CPU bound games, that would help show some differences in performance of the various cards. I tried to stress them as much as possible but in some cases (as we've seen before) the results are more CPU bound than video card. (The card can't draw more frames than it's being sent.)
One game that I'm eagerly awaiting for the Mac isn't available yet - Doom3. On the Mac everything is OpenGL (no DirectX) so it will be interesting to see if the 6800 delivers better performance than the X800 on the Mac with that game. (On the PC, the first Doom3 benchmarks for the 6800 led to a surge in sales of Nvidia 6800 cards, even though the X800 had better performance in some other games.)
I used Halo 1.5.1 (latest to date) with maximum detail/image quality settings (shadows/specular/particles, etc.). Only Lens Flare was set to Low (although 1.5 reportedly fixed the lens flare issue) and Audio Quality set to Low (variety left at medium setting). I tested at 1600x1200 and 1920x1080 with no FSAA and 4x FSAA. BTW - although Particles were enabled in the game video setup menu, the timedemo reports always listed them as "Off". The image quality issues I mentioned in last fall's 6800 Ultra review with Halo 22.214.171.124 seemed to be fixed.
Enabling 4x FSAA in Halo 1.5.1 delivered some surprises
as far as the 6800's performance.
At 1600x1200 with 4x FSAA enabled, surprisingly the OEM 9800 Pro delivered better performance than the 6800 Ultra on the Mac, so I can only assume the code isn't optimal for the 6800. At 19x12 however, the 9800 runs out of breath. The X800 XT's shader performance was clearly the best. It would easily be my choice of the current Mac graphics cards for this game.
Unreal Tournament 2004 (v3339) Tests:
Although it's limited to 1600x1200 resolution max in the version I have (although you could edit the ini files to change that), I used
Santaduck's UT2004 benchmark utility (www.santaduck.com/ut2k4/santa_toolpak_bench_ut2004.sit) to automate running tests with the latest (to date) Mac UT2004. For FSAA/AF tests, I had to manually edit the config files.
The 6800 was a hair faster on this test, but to stress the cards more I enabled 4x FSAA + 8x Anisotropic Filtering. The 6800Ultra's performance took a huge hit with both of these enabled and again the 9800 shows its age:
In the original 6800 Ultra review I tested the 6800 card with both MultiSampleHint2 (default - "nicest smoothing") and MultiSampleHint1 ("fastest smoothing") modes, but due to time constraints I didn't
test Hint1 this time. (It didn't make a big difference last time.)
Botmatch tests are very heavily CPU bound so I didn't graph those
although there was some performance differences in the cards at 1600x1200, especially with FSAA, etc enabled where the X800 XT was a few FPS faster than the 6800Ultra, and the 9800 Pro trailed both by about 25%.
Splinter Cell 1.0 Benchmarks:
I used the Mac benchmark scripts from Mac Splinter Cell FPS db. (Attempts to use beyond3d's benchmark file failed to run properly, due to the control prefs setup apparently.)
Splinter Cell does not support FSAA, but high quality settings were used:
SHADOWLEVEL High, SHADOWRESOLUTION High, EFFECTSQUALITY VeryHigh, SHADOWMODE Projector
Here's a graph of the Average FPS rates of the 3 cards cards at 1600x1200 mode.
Some readers asked for lower resolution tests like 1024x768 (that they normally play at), so I graphed 640x480 to 1600x1200
The 6800 has an edge at the low end (driver efficiency perhaps?)
but the gap narrows as resolutions rise. The older 9800 nosedives as
resolutions rise and can't match the bandwidth of the newer cards.
Call of Duty MP Timedemo:
COD is one of my favorite games on the Mac. The X800 XT delivered the best performance with it I've seen so far in this system, although
the 6800 Ultra's performance is very close.
Return to Castle Wolfenstein Tests:
I used the same RTCW Multiplayer demo test used in previous Video card and CPU upgrade reviews. All tests used High Quality game settings, audio enabled, but music volume set to 0. I tested at 1600x1200 and 2048x1536 modes to put as much stress on the cards as possible. (No FSAA options for this game and no control panel for the Nvidia card to
force it on as there is for the ATI cards.)
The 6800 Ultra was slightly faster than the X800 XT at the high end in this test.
Quake3 Arena Tests:
Yes, it's an oldie, but one of the few games on the Mac that scales very well with graphics cards AND shows a big boost from dual processors. (There's also enough mods that it's still played.) Due to the rush I tested only at 1600x1200 and 2048x1536 (no 1900x1200 option). The last Quake3 1.32 'G4' version was used with Quake3's "high quality" settings PLUS high geometic detail and max texture quality, with all game options on. Audio was not disabled, but music volume set to 0. (r_smp set to 1 to use dual CPUs and my typical "s_chunksize 4096" vs default 2048 was used as well as "com_hunkmegs" set to 256 and "s_mixadhead" set to 0.1.)
I was pleased with the X800's overall game performance (and image quality) especially since all Mac 3D games are OpenGL and on the PC, the 6800 generally has better OpenGL performance than the X800. (Although I've not kept up with later ATI PC driver releases, which may have closed the gap somewhat.) As I mentioned on the intro page, personally I'd choose the X800 over the 6800 on the Mac, all things considered. The lower cost (compared to the Mac 6800 Ultra) and ATI Control Panel features are also factors in that decision.