Author: Bryan Hofmann
Date: 11:16:03 02/16/04
Go up one level in this thread
On February 16, 2004 at 14:02:07, Peter Skinner wrote: >On February 16, 2004 at 13:41:32, Bryan Hofmann wrote: > >>Sorry I gave the Linux switches, it should be "/Qprof_genx /Qprof_dir ." for the >>first compile and "/Qprof_use /Qprof_dir ." (Also note the space between dir the >>the . this means the prof directory is the current directory). The prof file i >>gave you is simply a text file that will run crafty through several different >>position. You can name it prof.txt just make sure the command you give to run >>for step 2 is "crafty < prof.txt" Also after reading the documentation that >>intel provides they recommend that you remove the /Qip & /Qipo for the first >>compile and put them back in for the second compile. >> >>Bryan > >I compiled it with the PGO like you suggested and here are the results: > >Initializing multiple threads. >System is SMP, not NUMA. >EPD Kit revision date: 1996.04.21 >unable to open book file [./book.bin]. >book is disabled >unable to open book file [./books.bin]. > >Crafty v19.10 (1 cpus) > >White(1): bench >Running benchmark. . . >...... >Total nodes: 89942714 >Raw nodes per second: 755821 >Total elapsed time: 119 >SMP time-to-ply measurement: 5.378151 >White(1): > >It is actually slightly slower than the normal Intel 8 build, but still much >better than the MSVC 6.0 builds. > >Peter It is not an exact science and some times you have to play with the switches in the different stages to get the best results. You may try using the /QO2 instead of the /QO3 in the 1st and second stages. I played with it for hours one day until I got the best speed. Good Luck
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.