Author: Vincent Diepeveen
Date: 13:23:11 05/05/04
Go up one level in this thread
On May 05, 2004 at 08:49:01, Sune Fischer wrote: >On May 05, 2004 at 08:38:50, Vincent Diepeveen wrote: > >>On May 05, 2004 at 08:29:47, Sune Fischer wrote: >> >>>On May 05, 2004 at 07:26:40, Vincent Diepeveen wrote: >>> >>>>On May 05, 2004 at 06:52:06, Gerd Isenberg wrote: >>>> >>>>With bitboards perhaps it's slower. >>>> >>>>Also in endgames with non bitboards it's faster. >>> >>>It's not a bitboard issue, it's because in the endgame the attack rays are >>>generally longer so more bits need updating. >>> >>>-S. >> >>Knowing Gerd just uses bitboards to get to his goals i guess he was stuck there. > >What kind of speed difference do you usually see between midgames and endgames? > >I usually get something like 6-800 knps midgame and 11-1800 knps in endgames >(2GHz). Varies a lot depending on what type of endgame of course. 180k nps in opening up to 600k nps in far endgame (dual k7). incremental attacktables are not the problem when getting to speeds of 600k nps, but facts like using 3 hashtables (pawn,eval,transposition) with for the transposition table 4-8 probes. I hash everywhere, also in qsearch. Look the far endgame is not interesting at all. We search deep enough there. You can wake me up any time of the day to tell me how i can lossless search 1 ply deeper though without having to deal with burocracy. >I never saw that kind of speed up when I used attack tables, maybe just barely >hitting 10-1100 knps sometimes in the endgame while usually doing 6-800 knps >midgame. > >So throwing them out have given me a big boost in the endgame and no noticable >penalty in midgame. >-S. And at the same time throwing it out means we won't ever see you at a world championship. The best horseman is always on his feet...
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.