Author: Aaron Gordon
Date: 01:46:02 10/28/01
Hey guys. Been tinkering around a bit more with the Intel C 5.0 compiler, messed with the -Qopenmp option. Turns out it may have done the trick. Slater Wold says he got about a 200kn/s increase over my older CraftyK7 18.12. I profiled it with MMX now also, seems slightly faster with a single cpu as well. Here's slate's Dual AthlonXP 1.53GHz (1800+) benchmarks with the 'test' version of this (this verion is %2 faster than the one he got 200kn/s more on). Total nodes: 118647746 Raw nodes per second: 1501870 Total elapsed time: 79 SMP time-to-ply measurement: 8.101266 Anyone else out there with dual Athlon/Thunderbird/Durons, etc please try this version out. Would like to see how much faster it is with dual cpu's. Also interested to see if it helps on intel cpu's. Thanks. You can grab it at ftp://65.10.79.9/pub/crafty/c1812-k7.zip (if you're using a single cpu and downloaded this in the past it wouldn't hurt to get it again, slightly faster). The reason I decided to try that compiler flag is I think I remember hearing Hyatt say something about using OpenMP, not sure. So far so good <crossing fingers> :) Oh, if anyone wants the optimized Sjeng11 for K7's, here ya go. ftp://65.10.79.9/pub/chess/sjeng11-k7.zip Seems to be about 15% faster than Dan Corbits binary.
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.