Author: Frank Quisinsky
Date: 13:18:35 02/27/04
Go up one level in this thread
On February 27, 2004 at 14:23:07, Uri Blass wrote: Hi Uri, >It is possible that a program may be 100 elo better than previous version with >ponder off or at x minutes/40 moves and not stronger with ponder on or at x >minutes per game(the way sarah tested it). >The question is if that is the case with Ruffian. > >Uri I added for minutes the final results after 160 games with 40/20 in my forum and the links in CCC. Absolutly clear that the Ruffian results with ponder=on are not good. The results by Leo, SSDF too. Other persons add very good results with ponder=off from Ruffian in my forum (same results as my results). And now: Ponder alone is not over 100 ELO but in combination with ponder its a problem. Hash delete, hash problem with ponder=on ... not easy to see for me! But at the moment Per-Ola check the games. Harald Faber send me games too with autoplayer and ponder=on. The results are bad too, very what I saw with ponder=on! This could be explain and I hope Per-Ola found it and we have ... we have a strong Ruffian and all are happy :-) Best Frank
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.