Advanced Micro Devices Inc (AMD)

[wbmw]

Petz, why not use a more scientific method to predict the 2.0Ghz Opteron scores, rather than half relying on unsubstantiated rumors and your own SWAG? The SPEC scores certainly give enough datapoints to get a reasonable estimate. Let's see:

http://www.spec.org/cpu2000/results/cpu2000.html

SPECint
Opteron 240: 880
Opteron 242: 991 (12.6% perf, 14.3% frequency, 88% scaling)
Opteron 244: 1081 (9.1% perf, 12.5% frequency, 73% scaling)
Opteron 246: 1081 * (1 + (11.1*0.80)%) = 1177, assuming 80% scaling

SPECfp
Opteron 240: 934
Opteron 242: 1029 (10.2% perf, 14.3% frequency, 71% scaling)
Opteron 244: 1122 (9.0% perf, 12.5% frequency, 72% scaling)
Opteron 246: 1122 * (1 + (11.1*0.70)%) = 1209, assuming 70% scaling

So my estimate for an Opteron 246 is 1177 SPECint_base and 1209 SPECfp_base.

Athlon 64 should do better with DDR400 memory, but your estimate is 1327 and 1426, which is 12.7% and 17.9% higher than mine, respectively. Unless you are somehow also accounting for undisclosed micro-architectural improvements, then I would have to say you are in wishful thinking land once again.