AM (myname4rwt@jee-male.com) on 8/16/10 wrote:
---------------------------
[...]
>>That's alas the case that most (all?) papers providing
>>benchmark results never provide a way to reproduce them, and
>>this also applies to the papers linked by nVidia.
>
>You didn't even care to read them, did you? For otherwise you'd know that unlike
>Intel's paper, there's source for at least some of them.

I quickly went through two of the articles and found them to
be completely biased, or at least not fair with the CPU; I
posted in that very same thread... Can't you read or are
you forgetting on purpose?

Since then I also looked at one of the C codes that was
used to benchmark on the CPU and all I can say is that it
can't be qualified as optimized code. I didn't bother
optimizing that code as I don't have access to the GPU code
they wrote, so even if I could speed up that code by 50x or
any other number that wouldn't prove anything in a way or
another.