Re: Data and common sense about PIV optimizations, Gcc and the Intelcompiler;

[Thomas Dodd <ted@xxxxxxxxxxx>]

Didn't we just do this thread a few month ago?
Mid October, subject: optimised glibc

Me, Jean, and Florin atleast.

Dan, Check the list archives for that one.

Jean Francois Martinez wrote:

> At least if you are using a PIII.  Let's remember that RedHat's ordinary
>  
Or athlon. While I'd like an optimized glibc, I suspect it would take a 
lot of work to find the correct tuning, just adding -O3 or msrch=athlon 
won't do it. Glibc is to complex.

> I don't know for PIV.  For Athlon I can only do an educated guess: AMD
> knows well most of the time its processors will be running code who has
> been optimized for Intel ones AMD cannot make processors whose
> performance crumbles if sequence is not exactly optimized for them.
> So AMD processors have either to be agnostic (ie sequnce A and sequence
> B are equally fast on them) or haves speed tables close to the speed
> table of their main Intel rival (ie if A is faster than B on Intel, AMD
> will ensure it is also faster on Athlon).  That is why I doubt compiling
>  
No, only that A is as fast on AMD as on Intel. B could be faster on AMD, 
but it will seldom be used. Read the Optimization guidelines for Athlon. 
Not sure if it has changed in Intel CPUs, but for the Athlon,FEMMS and 
EMMS are the fastes way to clear the FP stack, but that was very slow on 
Intel. The Intel recomended way was still fast, but there was a faster 
way in the Athlon.

I have a copy of it, Doc #22007, Rev. I, September 2000, "AMD Athlon 
Processor x86 Code Optimization Guide"
New revs would cover the newer CPUs, but I doubt much has changed.
I can forward it to anyone (3 MB), or you can get the newest from AMD, 
either download, or request a free CD of all the current AMD docs
<http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/22007.pdf> 
is rev K, Feb. 2002.
<http://www.amd.com/us-en/Processors/DevelopWithAMD/0,,30_2252_739_1055,00.html> 
has instructions to get the current CD.

   -Thomas




_______________________________________________
Redhat-devel-list mailing list
Redhat-devel-list@redhat.com
https://listman.redhat.com/mailman/listinfo/redhat-devel-list