Author: Mark Young
Date: 03:00:16 11/12/98
Go up one level in this thread
On November 12, 1998 at 05:45:58, Howard Exner wrote: >On November 12, 1998 at 04:49:44, odell hall wrote: > >> >>On November 11, 1998 at 19:48:45, Mark Young wrote: >> >>>On November 11, 1998 at 11:33:23, odell hall wrote: >>> >>>> >>>>On November 11, 1998 at 06:48:14, Mark Young wrote: >>>> >>>>>On November 11, 1998 at 06:25:36, Mark Young wrote: >>>>> >>>>>>sorry I see the hash size now, 16mb. I will recheck with that size. >>>>> >>>>> >>>>>Now it plays your rebel 10 moves when set to 13mb hash and 10 mb hash. At lower >>>>>hash size it plays like my Rebel 10 set at 120mb hash. Try this, play the same >>>>>opening but set rebel 10 at 8 mb hash, and see if you still can win. >>>> >>>> >>>> ??? I don't understand the logic. >>> >>>You don't need to understand my logic. Just play the move of the game at 8mb >>>hash, you will see what I mean. Rebel at certain hash sizes plays like a >>>different program with big eval changes when set on 8 mb hash for Rebel 10 >>>playing with 10 mb hash. >> >> >>Mark I tried your suggestion and tested rebel10 at hash=8mb, and Hash=10mb my >>results were that their was absolutley no difference at all!!! >> >> >>19. rebel chooses BxG7(0-33sec)at33sec it chooses 19.Nxb5 8ply 0.45 then at 6min >>it still chooses nxb5 at 10ply 0.45 total positions searched 22,468,608 >> >>I got the exact same results above with the hash table set to 10mb! So much for >>your theory > >I have a theory also about what might be happening. Help me test this out. >Set combination to off as it would be in tournament mode. Now run the position >three times in either 10 MB hash or 8 MB hash, using first AG=Off, then AG= >Active and finally AG=Strong. Does Ag=off produce Bxg7, followed for a awhile by >Bd3 then returns to Bxg7? Does AG=Active get Bxg7 and holds on to it? >Finally, do you get the Nxb5 sequence when AG=Strong? > >That's what I'm getting on my machine. The trouble is the default of AG=Smart >seems to just randomly choose one of the 3 options (Off,Active or Strong) >with no real consistency. Can you Mark and Odell confirm this. > >When AG=Smart I can force a move, take it back, replay the exact same position >and then get an entirely different move sequence. Are we all doing something >wrong or did Ed and the beta testers miss something? I know it is doing something different, I will work the problem and see what I find.
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.