Author: Tord Romstad
Date: 11:35:33 01/14/06
Go up one level in this thread
On January 14, 2006 at 13:35:12, Tomas Casanovas Martinez wrote: >Hi, Tord, > >The settings have been very similars to those of other tests I perform >regularly... except that I have set ponder 'ON'. I think that in my computer >ponder 'ON' is not OK for the number one engine Rybka. Their nodes drop to about >50% in relation to ponder 'OFF' whereas the drop for other programs is about >25%. I can not repeat the test with ponder 'OFF' at this moment since my >computer is bussy testing Rybka 101 Beta 10 against two other engines. All right, I suppose the settings were OK, then. The results are strange, but as we all know strange results do happen, especially when the number of games is so small. >The unusual number of draws is also surprising for me. Have you set a different >'draw contempt' in relation to your previous version?. I have not changed >anything of your standard settings. No, I haven't made any such changes. A draw is evaluated as 0.00 pawns, like in all previous versions. I also haven't noticed an unusual number of draws in other games, so it was probably just a random fluctuation in your games. >One of the aspects that have impressed me more from your 'new Glaurung' is its >strengh in the endgames. It has been able to save different inferior positions >in the test, sometimes with an excellent vision. (Take my opinios with care, >please. I am not a Master, my latest Elo in Spain -20 years ago!- was about >2060). It would be very nice that anyone of our members, that own engame tests >to perform, tests Glaurung in this ares. I have also noticed that the endgame is clearly improved compared to Glaurung Mainz. This is rather strange, because I haven't added any new endgame knowledge at all. The improved endgame play must be caused by a better search and hash table code alone. Tord
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.