Author: Mike Byrne
Date: 09:35:37 02/07/04
Go up one level in this thread
On February 07, 2004 at 12:22:59, Peter Skinner wrote: >On February 07, 2004 at 08:41:36, Mike Byrne wrote: > >>On February 07, 2004 at 08:39:42, Mike Byrne wrote: >> >>>On February 07, 2004 at 08:20:37, Drexel,Michael wrote: >>> >>>>Comparison Win32 executables with identical Crafty.rc file for AMD 2000+ >>>>processor: >>>> >>>>Skinner 3.2.2004: >>>> >>>>Crafty v19.10 >>>> >>>>White(1): bench >>>>Running benchmark. . . >>>>...... >>>>Total nodes: 71227364 >>>>Raw nodes per second: 962531 >>>>Total elapsed time: 74 >>>>SMP time-to-ply measurement: 8.648649 >>>>White(1): >>>> >>>>Byrne 2.2.2004: >>>> >>>>Crafty v19.10 >>>> >>>>White(1): bench >>>>Running benchmark. . . >>>>...... >>>>Total nodes: 71227364 >>>>Raw nodes per second: 890342 >>>>Total elapsed time: 80 >>>>SMP time-to-ply measurement: 8.000000 >>>>White(1): >>>> >>>>Corbit 4.2.2004: >>>> >>>>Crafty v19.10 >>>> >>>>White(1): bench >>>>Running benchmark. . . >>>>...... >>>>Total nodes: 79723137 >>>>Raw nodes per second: 937919 >>>>Total elapsed time: 85 >>>>SMP time-to-ply measurement: 7.529412 >>>>White(1): >>>> >>>>It seems Skinners executable is clearly the best for AMD computers. >>>>Thanks to all for making them available. >>>> >>>>Michael >>> >>> >>>Interestingly, the Byrne exectuable is faster on Skinner's computer than his >>>own. I have rec'd an email from Peter stating this. >>> >>>================================================================ >>>Windows XP Professional - 2.0Ghz Pentium 4 768 megs ram. >>> >>>Byrne Crafty PGO >>>White(1): bench >>>Running benchmark. . . >>>...... >>>Total nodes: 89942714 >>>Raw nodes per second: 719541 >>>Total elapsed time: 125 >>>SMP time-to-ply measurement: 5.120000 >>>White(1): >>> >>>Byrne Crafty SE >>> >>>White(1): bench >>>Running benchmark. . . >>>...... >>>Total nodes: 89942714 >>>Raw nodes per second: 795953 >>>Total elapsed time: 113 >>>SMP time-to-ply measurement: 5.663717 >>>White(1): >>> >>>Peter's Crafty Compile >>> >>>White(1): bench >>>Running benchmark. . . >>>...... >>>Total nodes: 89942714 >>>Raw nodes per second: 719541 >>>Total elapsed time: 125 >>>SMP time-to-ply measurement: 5.120000 >>>White(1): >>>=============================================================== >>> >>>So it's back to the old adage -- "your mileage may vary". I think this also >>>demostrates again that the AMD processors are usually better for chess than >>>Intel P4 that run from 1.7 Ghz to at least 2.4 Ghz - I'm not sure about newer >>>processors. >> >>BTw, for my p4 1.7 Ghz, I also get results similiar to Peter - about 10% faster >>than his executable. > >Hi Mike, > >Yes when running the benchmark tests with a reboot between each one, yours >indeed is faster than my compile. The odd thing is during play online I notice I >_never_ seem to see it achieve 100% cpu, normally around 97%, where my exe >nearly always gets 100%. I wonder what the difference is. > >Also in little test games here and online mine performs better than yours as >others have pointed out as well. I am wondering if you have changed something in >the evaluation that is hurting it. > >Peter. No, when used as Crafty "Standard" , it is exactly the same. I have no idea why it does not get 100% on your machine and ditto for you other comment as well. As an aside. an engine rarely gets 100% on my machine as I'm usually doing something my nachine si doing something in the background. 99% is more common for me. I keep falling back on 'ymmv".
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.