Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Simple quad-opteron test

Author: Sune Fischer

Date: 12:48:07 12/04/03

Go up one level in this thread


On December 04, 2003 at 13:47:17, Russell Reagan wrote:

>On December 04, 2003 at 07:45:58, Sune Fischer wrote:
>
>>That huge speedup Sjeng gets, assuming Sjeng isn't largely bitboard based,
>>indicates that other factors may be equally important.
>
>Who knows what he is doing. I keep asking him and he says vague things that make
>me think two different things. I can't imagine that he gets a 70% speed increase
>while Crafty only gets a 60% boost if Sjeng is not bitboard based. He says,
>"Sjeng can take advantage of 64-bit hardware, and, "Crafty is the classic
>bitboard program, so what do you think?" That makes me think it's bitboard
>based. But then he goes and says something like, "What I think about bitboards
>isn't a secret," and what I remember is him arguing about why other approaches
>are better than bitboards quite a bit in the past. Dann Corbit also said
>something that made me think he saw Sjeng code that was bitboard based at some
>point. But who knows.

I guess he is just too polite to come right out and tell us it's none of our
damn business what he does, I can take a hint however :)

But unless he has turned 180 since the old days I doubt he is using bitboards.
Of course people have been known to grow smarter over time, so... ;-)

>I personally don't get it. He posts benchmark numbers, but doesn't just state
>information that is VERY relevant to the interpretation of those numbers. As it
>is, we can only guess, and that makes his numbers kind of meaningless.

I think it tells us something, that engines in general can look forward to a
good speedboost.
Exactly what's causing it for Sjeng we don't know, but then again we also don't
know exactly what is causing it for Crafty.

>Anyway, I am looking into buying one at some point, so I can just write my own
>bitboard and array based programs and see how they fair.

Yeah, nothing like good old cumbersome tedious testing.

-S.



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.