Mr. DOug, your statement, it is clearly not spec_fp, nor spec_int, but their _rate equivalents, in 4S systems --
That statement is a DIRECT LIE.
The article said nothing of the sort. Here's what the article actually said:
Barcelona will deliver 42 percent better floating-point performance than Intel's Xeon X5355 Clovertown, said Mario Rivas, executive vice president of the computing products group at AMD, making the comparison with the SPEC_fp benchmark that measures mathematical calculation speed. Using SPEC_int, which measures more common integer-processing tasks, Barcelona will outperform Clovertown by greater than 10 percent, AMD said, without being specific.
It is using SPEC_fp and it is comparing AMD's fastest at release to Intel's fastest when Barcelona ships. I said over a month ago that the 42% claim was based on comparisons of equal number of sockets, and this article confirms it.
There is absolutely no indication of using SPEC_int_rate or SPEC_fp_rate in that article. Absolutely impossible for AMD to compare them in a 4S system, anyway, as you well knew when you lied about what was said.
And guess what. Because of the FSB choke problem, SPEC_RATE benchmarks will favor Barcelona by even larger amounts.
Petz |
