Crafty 8.22 Bratko-Kopec test results
Dikran B. Karagueuzian
compiled with the options -DCOMPACT_ATTACKS and -DUSE_SPLIT_SHIFTS
Note that the results (13 solve, 11 unsolved) are worse than those
reported by Steven J Edwards for a 486/33 machine, although the mean
node frequency is higher. A strange phenomenon, indeed.
Script started on Wed Dec 6 20:38:46 1995
zerah:crafty
unable to open book file [book.bin].
unable to open book file [books.bin].
Crafty v8.22
White(1): epdscor BK_180.epd
total: 24
total acn: 38430618 mean total acn: 1601275
total acs: 4153 mean total acs: 173
total mean node frequency: 9253 Hz
solve: 13
solve acn: 22544960 mean solve acn: 1734227
solve acs: 2165 mean solve acs: 166
solve mean node frequency: 10413 Hz
unsol: 11
unsol acn: 15885658 mean unsol acn: 1444150
unsol acs: 1988 mean unsol acs: 180
unsol mean node frequency: 7990 Hz
White(1): quit
execution complete.
zerah:exit
zerah:
script done on Wed Dec 6 20:39:05 1995
Robert Hyatt
Dikran B. Karagueuzian <san...@Csli.Stanford.EDU> wrote:
--> Here are the test results for Crafty 8.22 running on a Sparc 5,
-->compiled with the options -DCOMPACT_ATTACKS and -DUSE_SPLIT_SHIFTS
-->Note that the results (13 solve, 11 unsolved) are worse than those
-->reported by Steven J Edwards for a 486/33 machine, although the mean
-->node frequency is higher. A strange phenomenon, indeed.
-->
-->
-->Script started on Wed Dec 6 20:38:46 1995
-->zerah:crafty
-->unable to open book file [book.bin].
-->unable to open book file [books.bin].
-->
-->Crafty v8.22
-->
-->White(1): epdscor BK_180.epd
-->total: 24
-->total acn: 38430618 mean total acn: 1601275
-->total acs: 4153 mean total acs: 173
-->total mean node frequency: 9253 Hz
-->solve: 13
-->solve acn: 22544960 mean solve acn: 1734227
-->solve acs: 2165 mean solve acs: 166
-->solve mean node frequency: 10413 Hz
-->unsol: 11
-->unsol acn: 15885658 mean unsol acn: 1444150
-->unsol acs: 1988 mean unsol acs: 180
-->unsol mean node frequency: 7990 Hz
-->White(1): quit
-->execution complete.
-->zerah:exit
-->zerah:
-->script done on Wed Dec 6 20:39:05 1995
I have not run these positions in a very long time. The tactical problems I suspect
it does pretty well at. The pawn lever problems are something else, as they are
dependent on eval tuning, which still needs a lot of work. At tactical problem sets,
it seems to be doing well. For positional chess, a lot of work yet to be done.
I'm not sure, but I think I recall that Cray Blitz solved all 12 tactical problems in
under 1 minute, and solved 10 of the 12 pawn lever positions in under one minute. I
think this was published somewhere in a revised Bratko-Kopec paper of some sort. If
you want, I can dig this up. Bottom line is, Crafty's significantly better than Cray
Blitz overall, but has not had the years of pawn structure tuning to make this type of
test produce good results. Just needs time.
Bob
--
Robert Hyatt Computer and Information Sciences
hy...@cis.uab.edu University of Alabama at Birmingham
(205) 934-2213 115A Campbell Hall, UAB Station
(205) 934-5473 FAX Birmingham, AL 35294-1170
Herbert Pohlai
Here are the test results for Crafty 8.22 running on a NekoTech Alpha(289 MHz),
compiled with gcc from the Makefile (without -pipe)
Crafty v8.22
White(1): time
using cpu time
White(1): time
1 moves/3 minutes primary time control
1 moves/3 minutes secondary time control
White(1): hash
hash table memory = 4210688 64-bit words.
White(1): hashk
king hash table memory = 65536 64-bit words.
White(1): hashp
pawn hash table memory = 98304 64-bit words.
White(1): epdpfga
...
White(1): epdscor BK_180.epd
total: 24
total acn: 146480085 mean total acn: 6103336
total acs: 4231 mean total acs: 176
total mean node frequency: 34620 Hz
solve: 16
solve acn: 101427365 mean solve acn: 6339210
solve acs: 2691 mean solve acs: 168
solve mean node frequency: 37691 Hz
unsol: 8
unsol acn: 45052720 mean unsol acn: 5631590
unsol acs: 1540 mean unsol acs: 192
unsol mean node frequency: 29255 Hz
Herbert