Well then I'll just keep my PA-2013 but disable the 1mb L3 then, also I'm considering buy a K6-3+ 450 to replace my K6-3 333, and this K6-3 the owner says runs at 600mhz fine, so might do that.
My board has 2MB of L3 instead of 1MB (this board came in a range of 512KB, 1MB, and 2MB of L3), so it'd probably be even less of an impact for yours.
Pentium MMX 233 | 64MB | FIC PA-2013 | Matrox Mystique 220 | SB Pro 2 | Music Quest MPU Clone | Windows 95B
MT-32 | SC-55mkII, 88Pro, 8820 | SB16 CT2230
3DFX Voodoo 1&2 | S3 ViRGE GX2 | PowerVR PCX1&2 | Rendition Vérité V1000 | ATI 3D Rage Pro
wrote:3DBench: 286.2 fps vs 286.2 fps PCBench: 122.1 fps vs 110.1 fps Doom: 710 realtics (~105.20 fps) vs 722 realtics (~103.45 fps) Q […]
3DBench: 286.2 fps vs 286.2 fps
PCBench: 122.1 fps vs 110.1 fps
Doom: 710 realtics (~105.20 fps) vs 722 realtics (~103.45 fps)
Quake: 61.5 fps vs 57.5 fps
these tests seem quite old and probably not very sensitive with memory bandwidth, which is what onboard cache helps. especially doom is obviously very limited by dos vga performance, i can't get much higher than 200fps(probably around 240) even with a core2duo at 3ghz and geforce2ultra, the fastest video card in doom that i ever tested.
i suggest that the impact would be higher if you run some modern 3d games and tests, like quake3, unreal and 3dmark2000/2001, and also run the cpu at higher clock ratio like 5.5x when memory bandwidth become more of a bottleneck.
