> SysMark Content Creation uses that WME program that people
> here love to hate, so I won't count that. Besides that
> benchmark, though, the 1.5GHz Pentium 4 with SDRAM beats
> the 1.2GHz Celeron with the exact same memory in Winzip
> 8.0 (no P4 optimizations that I know of)
WinZip's a good one. I like that, since it's a program I use a lot. :)
> Quake III (Intel friendly, but no painstaking
> optimizations required),
> DroneZ (includes SSE, but not sure about SSE-2),
DroneZ is a fair one, I think (though I'd like to see Palomino benches on it; anyone got?). Quake III is a sore spot, because it somehow managed to be misrepresentative of performance in just about all games (that I can tell) that are based upon its engine. For example, Heavy Metal: FAKK2 and possibly Return to Wolfenstein are based on the Quake III engine, and they are both not P4 friendly. There's nothing wrong with the P4 being strong in Quake III, but people got annoyed because it comprises about 80% of all actual in-game performance testing but in itself it only represents 5-10% of actual game use.
> 3DMark2001 (optimized, but not painstakingly so, AFAIK),
3DMark is way inconsistant. Whether it scores high or low depends on which day of the week it is. ;)
> and ScienceMark (said to be a quite objective benchmark
> for FPU performance).
It gets raped. The Pentium 4 gets horribly raped in ScienceMark so far. Primordia, which is very memory bandwidth friendly, scores respectably (it used to be, I think, king of the hill, but the addition of hardware prefetching on the Athlon XP and Pentium III sort of pushed it behind.
But I agree with your general scope about P4 performance. |
