Author: Andreas Guettinger
Date: 16:13:22 12/19/05
Go up one level in this thread
On December 19, 2005 at 18:32:07, Frank Wies wrote: >I think that the handicap put on Rybka is unrealistic. When I run engine matches >on my quad opteron 2.2GHz Rybka wins about 2-1 verses Deep Shredder 9x64. On >this rig, Fruit 2.2.1 was slightly ahead of Deep Shredder 9x64. >These were 1'0" games, where I expected DS9x64 to have a substantial advantage. >If you put enough of a hardware handicap on any engine, it will fail. >Try switching the hardware and I expect the Rybka with only one of the Power >Pc's will beat the Pentium. > >Cheers, >Frank I disagree. First, to initialize and coordinate the parallel search might give some overhead, therefore it can be that on such fast time controls (1 min per game) the Quad version is actually weaker than the single CPU version. I don't speak about NPS. I remember crafty purposely using only 1 CPU in the recent WCCC blitz tournament of it's 8 available. If this is the same for Deep Shredder9 I don't know, depends on search implementation. Of course one could make a match between Shredder9 and Deep Shredder9 with 1' 0" to test this. Therefore I think you cannot compare 1'0" matches with 4'2" matches. (I consider 1' 0 matches a joke anyway, because I think you don't measure chess playing strength with this.) Secondly, as stated previously, Rybka is not available for PowerPC. regards Andy
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.