Replies to post #62087 on Advanced Micro Devices Inc (AMD)

[chipguy]

I calculated roughly 3.2 GHz, assuming SUNs submission is Specfp2000 peak for 2.8 GHz Socket939 Opteron, from hardware alone.

I think you have got a big disconnect between peak and base.
Sun can play whack-a-mole with compiler option tweaking for
peak until the cows come home but there is little they can do
with base.

[chipdesigner]

It wasn't for a 939 Opteron, but for a 254 (940).

(You can tell because the benchmark is for the Galaxy X4200 2-socket model, not the Aquarius X2100 939-based server.)