Scumbria,
Don't get too optimistic. Remember Tom Pabst's prerelease PII scores? Or Thresh's prerelease K7 scores? Those previews basically said that the processors in question were crap. But they turned out just fine in the end.
There is substantial reason (the long pipe, perhaps the trace cache issue rate, Intel's own word) to believe that the P4 will have lower general per-clock performance than PIII/Athlon. But regardless of loyalties in this most interesting of wars, I think it's best to give Intel the benefit of the doubt as to the severity of this situation.
Also, not counting fpu performance (which, imho, seems to act differently to integer performance in these limited benchmarkfests I've seen), Willamette may merely not like Sandra very much. For all we know, Sandra might be the worst case scenario of the popular benchmarks. We may find that ZDNet's and PC World's benchmarks, for example, very highly favour the P4. And it's these benchmarks (well, and Q3/UT/MDK2) that will determine performance guesstimates for the mass media.
-JC |
