Quake 2 - Open GL Performance |
||
- |
Timedemo - 640 x 480 | |
- | demo1.dm2 | crusher.dm2 |
Canopus Spectra 2500 (Riva TNT) |
41.0 | 20.5 |
- | Timedemo - 800 x 600 | |
- | demo1.dm2 | crusher.dm2 |
Canopus Spectra 2500 (Riva TNT) |
40.9 | 20.6 |
- | Timedemo - 1024 x 768 | |
- | demo1.dm2 | crusher.dm2 |
Canopus Spectra 2500 (Riva TNT) |
35.8 | 20.6 |
Notice something odd here? The Celeron 300, without any L2 cache, which was once thought to be the equivalent of a Pentium II under Quake 2 and other gaming situations is actually performing slower than a Pentium II 233. Why would that be? It seems as if the TNT chipset itself is quite dependent on the presence of a high speed L2 cache, which is why the Spectra 2500 scaled so incredibly well with the Pentium II processors in terms of clock speed. For a Pentium II, when you increase the clock speed, you are also directly increasing the speed of the L2 cache, resulting in faster overall performance and for the TNT chipset, resulting in faster gaming performance. All of you original Celeron owners out there will want to avoid the Spectra 2500 as well as any other TNT based products, sorry guys, it looks like 3Dfx is the only path to take for the best overall 3D acceleration for you.
TNT vs Voodoo2 |
||
Celeron 300 |
Timedemo - 640 x 480 | |
- | demo1.dm2 | crusher.dm2 |
Canopus Spectra 2500 (TNT) | 41.0 | 20.5 |
Canopus Pure 3D-2 (Voodoo2) | 59.6 | 28.3 |
- | Timedemo - 800 x 600 | |
- | demo1.dm2 | crusher.dm2 |
Canopus Spectra 2500 (TNT) | 40.9 | 20.6 |
Canopus Pure 3D-2 (Voodoo2) | 53.3 | 27.7 |
- | Timedemo - 1024 x 768 | |
- | demo1.dm2 | crusher.dm2 |
Canopus Spectra 2500 (TNT) | 35.8 | 20.6 |
Canopus Pure 3D-2 (Voodoo2) | N/S | N/S |
0 Comments
View All Comments