Heikki Kultala (hkultala@iki.NOSPAM.fi) on 4/20/11 wrote:
---------------------------
>>It's simple. It's 3072 cores @ 0.88Ghz versus nvidia 448 cores @ 1.2Ghz. 3072 cores
>>always win by factor 3-4 then or so.
>>
>>No discussions there.
>
>Wrong.
>
>It's 24 ATI cores per chip versus 28-32 nvidia cores per chip.
>
>And, it's 384 ATI SPMD lanes per chip versus 448-512 nvidia SPMD lanes per chip.
>
>VLIW ALU != core

I wrote it in popular language, but that doesn't stop idiots like you.

It's 3072 PE's versus 448 PE's.

What's going to be FASTER in well designed gpgpu codes?

0.83Ghz * 3072 PE's is *always* going to annihilate in well designed codes a meager 448 PE's @ 1.2Ghz Tesla.

It's not even a contest. We're not speaking about a few % differnece here.

It's in raw performance (so no theoretic bitdicking about multiply-add which both gpu's have)

0.83Ghz * 3072 = 2550 Gflops for AMD 6990 versus
1.215Ghz * 448 = 544.32 Gflops for Nvidia's Tesla

Don't take me wrong: i understand nvidia's choices and don't judge them; gamers are their core business and what they do (clocking high and big bandwidth for 1 gpu) makes sense.

For gpgpu, if you care about throughput, it's a run contest.

Regards,
Vincent