Author: Dann Corbit
Date: 17:31:11 01/24/02
Go up one level in this thread
On January 24, 2002 at 18:55:29, Tom Kerrigan wrote: >If you've got, say, a 512MB hash table that's being accessed randomly and >infrequely, I don't really see the difference between a 256k L2 cache and a 2MB >L2 cache. The chance you'll find what you're looking for in the 2MB cache is >0.4% and the chance it's in a 256k cache is 0.05%. Seems like the most it could >speed your program up is 0.35%, and that's not taking into account the stuff >chess programs do other than access the hash table. I suspect the relationship is not linear, but then again, hashing itself scatters the data on purpose, so it might be. >'Course, if you're doing very small searches, the hash table would fit in the >cache and be accessed with relative frequency due to iterated searching, but I >don't consider this a very interesting case. True, but it will be a big help in some operations like EGTB lookup. I doubt if chess is a great application for Xeon, so I certainly agree with that basic conjecture. Heuristically, Bob agrees from experience [if I understand his wallet joke]. I think it would be possible to design a chess program that benefits from that architecture more, but it probably isn't worth the bother.
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.