>> Is it just me, or does this seem as underwhelming as the QuadFarter to anyone??
It is completely underwhelming. And not surprising. I had been surprised to read a few times your expectation that K8L would be an outstanding product. From the fact of late intro of K8 65 nm chips, it seemed to me they were having trouble. I think yesterday's FX demo and today's are confirmation.
AMD's next generation QUAD core that's supposed to be all that more impressive than Intel's "two dual cores on a package" ONLY GETS A 40% IMPROVEMENT IN SPECFP_RATE?!?
LOL, I wouldn't call that impressive.
Yeah that caught my eye. The Opteron has such a high memory bandwidth to FLOP ratio that going quad core should drive up SPECfp_rate2000 by more than 40%. Add to that the L3 and a reported doubling of SIMD DP FLOPs/clock to 4 like NGMA and it is completely underwhelming.
How much clock speed did AMD lose going from dual core to quad core to keep device TDP in check? Apparently quite a bit if AMD isn't seriously sandbagging.
They'll need at least two more steppings of the CPU with this level of functionality, and you don't even have time for full validation between now and then.
2 more? I'd guess 2 more full mask set steppings plus some minor metal layer ones.
Beamer, personally I count TPC-C scores more than SPECfc_rate, since the latter seems to vary too much to be reliable.
Even so, 70% scaling for TPC-C is not all that impressive to me. I was expecting more given the scalability one would expect from HyperTransport and the local memory controllers.