Author: Robert Hyatt
Date: 20:57:54 02/08/00
Go up one level in this thread
On February 08, 2000 at 23:21:53, Peter W. Gillgasch wrote: >On February 08, 2000 at 23:17:08, Dann Corbit wrote: > >>On February 08, 2000 at 22:13:58, Peter W. Gillgasch wrote: >> >>>On February 08, 2000 at 20:11:50, Dann Corbit wrote: >>> >>>>2bnN1k1/6P1/4p3/1P2N1P1/3K4/8/8/8 w - - >>> >>>Including producing a full blown log file with an awful lot of >>>statistics it took 2 seconds on my 60 mhz PowerPC601 machine. >>>Iteration 7, 43914 nodes, minimal hashtable (128 K entries). >>>This time includes clearing the hash table. >>> >>>Without the hash table it took 3 seconds, iteration 7, 77162 >>>nodes. I believe that things like these can be solved easily >>>much faster with better selectivity. >>> >>>Is there anything special about that problem ? >> >>Nothing that I can see. Crafty finds the right move in less than two seconds, >>but fails to see the mate for 6 minutes. Looks like a bug or ??? > >Jesus Christ ! Looks like Bob will have to work overtime. I did >notice that one log shows something like several million nodes >but I did not notice the time. Looks like a bug to me or maybe >some of his pruning stuff fouls it up big time ?! > >-- Peter It doesn't appear to be a bug.. just that on the fail high, some mates that were pruned suddenly become "ok". And move ordering just happens to follow an incredibly deep series of checks first. IE it finds a mate in 15 or so very quickly then tries to improve. But in doing so, it searches a _bunch_ of stuff. Just runaway extensions, mainly.
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.