Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Deep Sjeng Opteron Performance Results

Author: Gerd Isenberg

Date: 16:45:54 08/06/03

Go up one level in this thread


On August 06, 2003 at 18:25:11, Gian-Carlo Pascutto wrote:

>I had the opportunity to run some tests on an Opteron machine,
>and benchmarked Deep Sjeng on it, compared to the Athlon XP.
>
>The Opteron is 70% faster.
>
>Yes, that's _SEVENTY PERCENTS FASTER_, clock for clock.
>
>An 1.6Ghz Opteron runs Deep Sjeng faster than a (nonexisting) 2.7Ghz
>Athlon XP would.
>
>Operating system was Linux 2.4.19, SuSE 8.0 Opteron version.
>I compiled with GCC 3.2.2, pure (portable) C source.
>
>The reference was a MSVC6 SP5 with Processor Pack compile, with
>optimized parameters, and with optimized assembly language for
>most move generators and some other functions. (The fastest IA32
>binary I have for the Athlon), running on an Athlon XP of the same
>clockspeed.
>
>--
>GCP

Hi Gian-Carlo,

nice to hear ;-)

I'm not familar with GCC. I guess you compared a x86-64 executable running on a
single opteron processor, all sixteen 64-bit gp-registers used, with a x86-32
executable.

A few questions:

Do you use a lot of 64-bit stuff in Deep Sjeng?

Did you use profile guided optimization?

Did you benchmark the x86-32 Deep Sjeng binary on opteron?

Thanks in advance,
Gerd



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.