Advanced Micro Devices Inc (AMD)

[chipguy]

(you said that IPF is up to 20% slower than Opteron today in integer, but it's more like 10% or less).

They're basically equal today; 1408 for I2/1.5 vs 1405
for K8/2.2 on SPECint_base2k. Why don't we revisit
this comparison in July. Both architectures should
have faster parts available then but I suspect IPF
will have a modest but clear lead.

The funny thing is that SPECint2k performance is
incidental to IPF's targeted markets, and it often
gets beaten up for supposedly "lousy" integer
performance. If Opteron can't beat it then either
1) IPF integer performance isn't lousy at all, or
2) Opteron integer performance is also lousy. So
which is it boys?