Author: Robert Hyatt
Date: 19:49:52 01/16/04
Go up one level in this thread
On January 16, 2004 at 15:19:10, Bryan Hofmann wrote: >On January 15, 2004 at 11:58:31, Eugene Nalimov wrote: > >>On January 15, 2004 at 11:49:01, Robert Hyatt wrote: >> >>>On January 15, 2004 at 09:21:25, GuyHaworth wrote: >>> >>>>I'd be interested to know - if you have used more than a 2-processor machine to >>>>run Eugene Nalimov's EGT-generation code - how many processors you used: SMP >>>>machines presumably. >>>> >>>>Thanks, >>>> >>>>Guy >>> >>>I don't believe his code does a parallel generation, because the algorithm >>>itself is basically serial. You might run two at a time to generate two >>>separate tables, of course, but that might swamp the I/O and memory since the >>>memory requirement is huge when you mmap() files into your virtual address >>>space. >> >>Code does parallel generation, and I run it on a quad-CPU system using all 4 >>CPUs. >> >>Thanks, >>Eugene > >Where can I get a copy of this code that supports parallel? The code I dowloaded >off of Dan Corbitt's site (very old I guess 2001) does not have this capabililty >and I could sure put my other CPUs to use creating TBs. (BTW I need the source >to compile as I'm using under Linux unless someone has this already compiled for >such). > >Thanks > >Bryan I believe he has said in the past that the new code needs a 64 bit architecture to generate the new tables (6 piece). He maps things into main memory which requires a humongous virtual address space, if I am recalling this correctly...
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.