Author: Mogens Larsen
Date: 09:57:04 07/13/00
Go up one level in this thread
On July 13, 2000 at 12:32:58, Dieter Buerssner wrote: >To me, this is not to be a null move problem. I tried this with Yace default >settings. On my AMD K6-2 300 with 30 MB hash, it needed 350 s. >Mogens showed a log with about 100 sec. He shows almost exactly 3 times >the Nodes/s, that I get, and has about double hash - so this is consistent. >Then I tried a Crafty like null move algorithm, with higher reduction at >high depth: 240 sec. Then I tried even more aggressive null move with a >depth reduction of 4 at all depths: 180 sec. Then I tried no null move at >all - after 15 minutes without solution I lost patience. With null move, >Yace always found the solution at ply 12, regardless of the aggressifity >of the null move algorithm. I did suspect that it wasn't a null move problem, because Yace could find the solution relatively easy with default null move setting at ply 12. Crafty can find it too, but it takes a lot of time. >Also, I suspect that this position may not be too objective, when comparing >different chess programs. At the ply before Yace finds the solution, the >moves Qxa6 and Qxf3 are not far apart. Doubling positional scores halfed >the solution time in one test I have done. I guess that tweaking some eval >parameters should give the solution even faster. Of course, the 14 s reported >for Fritz are impressive ... ZChess found it in 10 seconds on my machine with 64Mb hash. But sadly abandoned the move after one minute, when the evaluation dropped for some reason. Best wishes... Mogens
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.