v 2.25 (all settings on for D3D / Nali Castle flyby)
Q3Test1.07 (largest texture size /
high quality sound)
“timedemo=1; demo Q3TestdemoX” commands:
3D Mark 99 Max (all done without
the n-polygon test)
32-bit tests: 24bit Z-Buffer / Trilinear Filtering
tests: 16bit Z-Buffer / Trilinear Filtering
needless to say, the Elsa card performs worthily of any TNT2 chip
and is on par with the rest of them in the market. As you can see, the CPU dependency becomes less apparent at
higher resolutions and the card is getting fill-rate limited then. Comparing this with my 32MB V770 (non-Ultra) clocked at
162/192, it is understandably slower because of its lower mem / core
clock settings (158/165) especially at higher resolutions. Now, if you’ve seen the V770 benchmarks in my previous
Spectra 5400PE review and wonder why they are slower at low
resolutions (compared to the Elsa’s here), it can be readily
explained by the slower P3 [email protected] setting used for benchmarking
my V770 then.
A TNT2’s just a TNT2.
Now give me the 3D glasses!!