[Nel] memcpy for SSE ?

[Gaylord Foureau]

Hi,

I see a strange thing in the file code\nel\src\misc\fast_mem.cpp.
The memcpy (and precache) for SSE doesn't seem to be optimise for SSE but
for 3DNow!

It uses some MMX instruction but movntq and prefetchnta are 3dNow! extended
instructions !

I'm sure of that, you can verity this here: www.cpuid.com
and in this document:
"Using Block Prefetch for Optimized Memory Performance
Advanced Micro Devices
Mike Wall Member of Technical Staff Developer Performance Team"

This document is avaible here:
http://www.amd.com/us-en/Processors/DevelopWithAMD/0,,30_2252_7451,00.html


Si I'm crazy or not ? ;-)


Elrick