Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Are bitboards really faster on 64-bit hardware?

Author: Gerd Isenberg

Date: 14:04:11 11/06/03

Go up one level in this thread


On November 06, 2003 at 15:55:10, Russell Reagan wrote:

>In this thread:
>
>http://www.chess-archive.com/ccc.php?art_id=310212
>
>Gian-Carlo Pascutto reported that his program, Deep Sjeng (compiled for 64-bit
>hardware), ran 70% faster on an Opteron, clock for clock. As far as I know, Deep
>Sjeng is not bitboard based.
>
>In this thread:
>
>http://www.talkchess.com/forums/1/message.html?325912
>
>Eugene Nalimov reports that Crafy (compiled for 64-bit hardware) gets 1,761,569
>nps on a 1.8GHz Opteron. On my 2GHz Athlon, Crafty gets 1,230,931 nps.
>
>(2.0 / 1.8) x 1,761,569 = about 1,957,298 nps on a 2GHz Opteron
>
>So, clock for clock, Crafty is 1,957,298 / 1,230,931  = 1.59 times faster (60%)
>on an Opteron, while Deep Sjeng was 70% faster. I suspect Eugene was using his
>magic unreleased compiler also, which probably produces faster executables than
>gcc. But that is speculation.
>
>So either Deep Sjeng is bitboard based, or the expected advantage that bitboard
>engines were going to get that non-bitboard engines were not going to get does
>not exist. Or I'm overlooking something else.
>
>Thoughts?

hmm, a few aspects: are the Athlons you compare really equal?
Let's see how other programs will perform. Sjeng may profit more from opteron
features independent from 64-bit stuff, e.g. bigger cache, more registers and
therefore less store/loads of locals and function parameters, faster memory and
smarter branch prediction. And 64-bit bsf is still expensive vector path
instruction with 9 cycles latency where all other pipes are blocked.

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.