Author: James T. Walker
Date: 06:51:52 12/07/00
Go up one level in this thread
On December 06, 2000 at 23:21:28, William Penn wrote: >On December 06, 2000 at 09:54:24, James T. Walker wrote: > >>I've been playing Chessmaster (1 Meg hash) vs Chessmaster (16/32M hash) >>overnight at game/25 minutes to compare the advantage of using 16/32 meg hash >>tables vice the default of 1 meg. The CM8K/16M won by 28-24 and the CM8K/32M >>won by 30-22 for a combined 58-46. The same settings at Game/5 minutes produced >>a win for the default CM8K of 104.5-95.5. >>There is a bug in the Chessmaster which causes it to lose on time ocasionally >>when the game is a draw by repetition. You have to check the "games" after all >>matches and when you see a "black/white lost on time" message then you need to >>check the game because it usually is a 3 time draw in which neither side claimed >>the draw but one side stopped playing and it's clock ran out. >>I'm also playing a match between Chessmaster vs Rebel Century 3.0 (G/1hour). >>After 20 games Chessmaster is winning by 13-7. I will stop this match at 24 >>games and return to Chessmaster vs Chess Tiger 13. Anyone wanting the PGN of >>the Rebel games just email me. >>Jim > >I've done many tests with different hash table sizes in CM6000, and a few with >CM8000. Larger hash table sizes have more effect at longer time controls. The >default 1MB hash table size is probably as good as any at fast time controls. >Also I've found that the hash table size generally doesn't have a big effect, >maybe on the order of a 10%-20% speedup at best. For those who need really >significant speedups, then higher MHz processors are the way to get it. >WP Hello William, In the middle game you are right but try the difference in the endgames with only a few pawns on the board and you will see sometimes 3 ply difference. Jim
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.