Author: Robert Hyatt
Date: 17:57:16 02/13/04
Go up one level in this thread
On February 13, 2004 at 17:42:19, J. Wesley Cleveland wrote: >On February 12, 2004 at 13:24:35, Aaron Gordon wrote: > >>On February 12, 2004 at 12:35:03, Andreas Guettinger wrote: >> >>>A fellow student of me got an Athlon64 3200 for his phD work at university and >>>installed the x86-64 version of Suse9. I could not resist to quickly compile >>>crafty to test the speed. :) >>> >>>unable to open book file [./book.bin]. >>>book is disabled >>>unable to open book file [./books.bin]. >>>EGTB access enabled >>>using tbpath=./TB >>>0 piece tablebase files found >>> >>>Crafty v19.10 >>> >>>White(1): bench >>>Running benchmark. . . >>>...... >>>Total nodes: 89942714 >>>Raw nodes per second: 1835565 >>>Total elapsed time: 49 >>>SMP time-to-ply measurement: 13.061224 >>>White(1): >>> >>> >>>regards >>>Andy >> >>With a good Windows binary running 64bit stuff you can probably figure you will >>be able to get a 40% increase over GCC. This is what I got going from GCC to one >>of my Intel C 5.0.1 binaries. This should put you up around 2.5 million >>nodes/sec @ 2GHz. :) > >I wish. I just managed to compile a 64 bit version of crafty on WinXP64 on my >Athlon64 3000 using the DDK compiler, and got >Raw nodes per second: 1578860. >for comparison, the 32 bit version I compiled got >Raw nodes per second: 1332804 > >I haven't made any special effort to optimize it, and my A64 3000 runs at the >same clock speed as the 3200 with half the L2 cache. When I get Linux running, >I'll compile that and compare. Note that for linux, I have a linux-amd option in the Makefile. And that there is an inlineamd.h that will add a couple of percent although the linux-amd makefile already takes advantage of that...
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.