Author: Drexel,Michael
Date: 04:32:36 09/29/03
Go up one level in this thread
On September 29, 2003 at 07:02:04, Tord Romstad wrote: >On September 29, 2003 at 04:03:10, Gian-Carlo Pascutto wrote: > >>It seems that the ChessBase UCI support contains a rather 'convenient' bug >>that severely handicaps UCI engines when playing tournaments under the >>ChessBase interface: >> >>An extract from Deep Sjeng's logs: >> >>First game: >> >>< setoption name Hash value 128 >>Allocated 4194304 hash entries, totalling 134217728 bytes, in shared memory. >> >>All subsequent games: >> >>< setoption name Hash value 1 >>Allocated 32768 hash entries, totalling 1048576 bytes, in shared memory. >> >>So, they basically disable hashing for UCI engines after the first game. > >Oh, my God. This is horrible. How widely known is this bug? > >My engine slows down by about 50% when I limit the hash table size to 1MB. > >Tord Yes, if this is really true a lot of testing would be meaningless. There are many questions: Are only tournaments affected or also engine matches? UCI.dll version used? Does it happen also with older versions of UCI.dll? ... Michael
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.