Author: Claude Dubois
Date: 02:21:32 04/20/04
Hello, I have disussions with Fabien Letouzey on le Forum du Fou Numérique about the attribution of 1 mb hashtable in the Fritz GUI I have let play a gaunlet with Yace (who is making marvelous logfile) of 36 games. The games was played on a PC Pentium III/800 Mhz, 256 mb Ram, Windows XP Pro. On 36 games, the fault of 1 mb hashtable was discovered on the game 2, 3, 5, 7, 14, 16, 17, 18, 19, 27, 30 et 33. There are 1/3 of the games who has had this problem. Yace is playing the impaired games with the white. Game 02 : Yace beats Fritz 8 (!) Game 03 : Yace drawed in 16 moves vs Chess Tiger 15.0 Game 05 : Yace loses vs Junior 8 Game 07 : Yace beats Deep Sjeng 1.6 Game 14 : Yace loses vs SOS 4 for Arena Game 16 : Yace draws vs Ktulu 4.2 Game 17 : Yace beats Patriot 1.1.0 Game 18 : Fruit 1.0 vs Yace Game 19 : Yace loses vs Shredder 8 Game 27 : Yace beats DelfiT 4.4 W2U Game 30 : Yace lose contre WildCat 3 W2U Game 33 : Yace lose contre Pharaon 2.62 Against, I have not till watched this problem on my tournament. Perhas, it's due that the engines are working one after one and there are no gaunlets. But this problem is pertubing me and I shoult want that ChessBase could ajust this problem. The games was played with the tempo 40'/40 moves. Before to make this test I sent a message to ChessBase with the Support formular and I am waiting for an answear since 4-5 days. Here you can read the logfile when Yace is receiving the information about 1 mb. Begin input: ; Use 30 Megabytes for hash tables. Change this to a size suitable for input: ; your hardware. You can use numbers like 10.5M, 10.5m, 10500k or 105000000. input: hash 64M hash size wanted 64000000 Stored 22 learned positions into hash table entries 5333331 size 63999972 size_wanted 64000000 tts3 1777776 input: ; set the chache size for endgame tables. Default is 2M. I suggest to use input: ; not less than 1M and not more than 8M. input: ; egtb_cache 4M input: ; Where to find the endgame tables, Nalimov format, compressed or uncompressed input: ; If you don't have endgame tables, delete the following line input: tbldir c:\tbs 5-men endgame table bases found input: ; save all my games input: sgames yacegam.pgn input: ; positional learning now is on by default input: book_learn 6 book_learn = 6 DBBOOK101 10000 128 20 ff 10578 input: ; to disable book learning, I suggest input: ; book_learn 4 input: ; Yes - 4 and not 0, so that Yace will have access to NAGs in dblearn.bin white ( 1): input: uci id name Yace Paderborn id author Dieter Buerssner option name Ponder type check default false option name Hash type spin default 1 min 1 max 1024 option name NalimovCache type spin min 1 max 64 default 2 option name NalimovPath type string default <empty option name Position Learning type check default true option name Log File type check default false option name Show hash performance type check default false option name Show add. EGTB info type combo default at end of search var never var at end of search var allways option name Endgame-Database Usage type combo default Normal var Never var Only at root var Very Low var Low var Normal var High option name Use only 3 and 4-men TBs type check default false option name OwnBook type check default true option name BookFile type string default dbbook.bin option name LearnBookFile type string default dblearn.bin option name BookLearning type check default true option name Book Random type check default true option name BitbasePath type string default <empty option name Selectivity type combo default Normal var None var Normal var Risky option name Extensions type combo default Normal var Normal var Aggressive option name Add. time for so many moves type spin min 0 max 15 default 5 option name null_high type spin min 0 max 600 default 300 option name null_low type spin min 0 max 600 default 300 option name null_endgame type spin min 0 max 600 default 200 option name null_split type spin min 0 max 2000 default 550 option name Recapture extension type spin min 0 max 100 default 59 option name Check extension type spin min 0 max 100 default 100 option name Perpetual check extension type spin min 0 max 100 default 59 option name Mate threat extension type spin min 0 max 100 default 100 option name Passed pawn extension type spin min 0 max 100 default 100 option name Autoplayer delay (ms) type spin min 0 max 2000 default 0 option name Clear Hash type button uciok white ( 1): input: setoption name Hash value 1 Stored 22 learned positions into hash table entries 87381 size 1048572 size_wanted 1048576 tts3 29126 white ( 1): input: isready readyok white ( 1): input: setoption name Log File value true end I give the end of the logfile when all is ok and we can see that the activation of the access to the nalimov's table are ok here and not with the false attribution. begin uciok white ( 1): input: setoption name Hash value 64 Stored 6 learned positions into hash table entries 5592405 size 67108860 size_wanted 67108864 tts3 1864134 white ( 1): input: setoption name NalimovPath value C:\Tbs 5-men endgame table bases found white ( 1): input: setoption name NalimovCache value 1 white ( 1): input: isready readyok white ( 1): input: setoption name Log File value true end I don't know how to sensitive ChessBase with a known problem and I think that CB doesn't need of that to let win the commmecials engines. What do you think about this problem ? Thanks Best wishes Claude
This page took 0.01 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.