Author: Peter Berger
Date: 14:15:25 07/31/02
Go up one level in this thread
This is from the actual logfile of the game:
white (15): input: time 140922
white (15): input: otim 118954
white (15): input: d1a4
usetime = 45.06, mintime = 22.53 maxtime = 225.32 tl 1407.22 ml 35
10 -34 2813 12653719 t 15...Bd5 16.Rfc1 Qg6 17.Ne1 e6 18.Nxb7 Kxb7 19.Qb5+ Ka8
20.Rxc6 Bxc6 21.Qxc6+ Kb8 {80}
10 -33 3486 16014615 t+ 15...Bxf1 16.Rxf1 a6 17.Rc1 Nb8 18.Qa5 Nc6 19.Ng5
10 -33 3753 17195834 t 15...Bxf1 16.Rxf1 a6 17.Qb3H Na5H 18.Qc3H Nb3H 19.axb3H
Qxf3H 20.gxf3H Rxd4H {HT} {-761}
10 -33 4443 20549951 . 15...Bxf1 16.Rxf1 a6 17.Qb3 Na5 18.Qc3 Nb3 19.axb3 Qxf3
20.gxf3 Rxd4 {HT} {-761}
10 -33 4443 20549951 f. 15...Bxf1 16.Rxf1 a6 17.Qb3 Na5 18.Qc3 Nb3 19.axb3 Qxf3
20.gxf3 Rxd4 {HT} {-761}
20549951 44.430 -0.33 10f. 15...Bxf1 16.Rxf1 a6 17.Qb3 Na5 18.Qc3 Nb3
19.axb3 Qxf3 20.gxf3 Rxd4 {HT} {-761}
20549951 Nodes, 45.17% Leavenodes, 462524 Nodes/sec
16152784 eval, 37.08% score, 77625688 genmoves, 23.26% captures le 363/231
ext: pawn 7918, rcp 130828, chk 317323, repchk 48336, null 9258, prune 1323291
htable: 16898269 store, 0 rejected, 20072317 probe, 19.8% f/p, 23.5% f/s
entries 5333331 age 15 renew 0
interior recognizer probes 0, found 0 egtb probes 0, found 0 max_depth 28
This is what I got when I just gave the position to Yace to analyze:
white ( 1): input: setboard 2kr1b1r/ppp1pppp/2n5/2N1P2q/Q1bP4/4BN1P/PP3PP1/R4RK1
b - - 0 1
Stored 68 learned positions into hash table
black ( 1): input: analyze
usetime = 19.93, mintime = 9.97 maxtime = 49.83 tl 598.00 ml 40
626568 1.477 -0.32 7t 1...Be2 2.Rfc1 Bxf3 3.gxf3 Kb8H 4.Qb5H Qg6+H
5.Kf1H b6H {0}
769054 1.741 -0.32 7. 1...Be2 2.Rfc1 Bxf3 3.gxf3 Kb8 4.Qb5 Qg6+ 5.Kf1
b6 {0}
1339537 2.891 -0.42 8t 1...Be2 2.Rfc1 Bxf3 3.gxf3 a6 4.Qb3 Qg6+ 5.Kf1
Na5 6.Qb4 {0}
2013197 4.423 -0.41 8t+ 1...Bd5 2.Qb5 Nb8 3.Ng5 a6 4.Rfc1 axb5 5.Nxb7
{931}
3468757 7.966 -0.29 8t 1...Bd5 2.Rac1 e6 3.Nxb7 Kxb7 4.Qb5+ Kc8H
5.Rxc6H Bxf3H 6.gxf3H Qxh3H {0}
3822087 8.646 -0.29 8. 1...Bd5 2.Rac1 e6 3.Nxb7 Kxb7 4.Qb5+ Kc8 5.Rxc6
Bxf3 6.gxf3 Qxh3 {0}
8105115 18.155 -0.27 9t 1...Bd5 2.Rfc1 e6 3.Nxb7 Kxb7H 4.Qb5+H Kc8H
5.Rxc6H Bxf3H 6.gxf3H Qxh3H {0}
9172390 20.191 -0.27 9. 1...Bd5 2.Rfc1 e6 3.Nxb7 Kxb7 4.Qb5+ Kc8 5.Rxc6
Bxf3 6.gxf3 Qxh3 {0}
15825992 34.932 -0.34 10t 1...Bd5 2.Rfc1 Qg6 3.Ne1 e6 4.Nxb7 Kxb7 5.Qb5+
Ka8 6.Rxc6 Bxc6 7.Qxc6+ Kb8 {HT} {80}
19176934 41.924 -0.34 10. 1...Bd5 2.Rfc1 Qg6 3.Ne1 e6 4.Nxb7 Kxb7 5.Qb5+
Ka8 6.Rxc6 Bxc6 7.Qxc6+ Kb8 {HT} {80}
38133219 1:24.7 -0.42 11t 1...Bd5 2.Rfc1 Qg6 3.Ne1 e6 4.b4 Bxc5 5.Rxc5
Qe4 6.Rac1 Kb8 7.Nf3 {HT} {0}
50118884 1:50.8 -0.42 11. 1...Bd5 2.Rfc1 Qg6 3.Ne1 e6 4.b4 Bxc5 5.Rxc5
Qe4 6.Rac1 Kb8 7.Nf3 {HT} {0}
input: quit
51788467 1:56.0 -0.42 11f. 1...Bd5 2.Rfc1 Qg6 3.Ne1 e6 4.b4 Bxc5 5.Rxc5
Qe4 6.Rac1 Kb8 7.Nf3 h6H {0}
51788467 Nodes, 44.32% Leavenodes, 446360 Nodes/sec
40161851 eval, 37.96% score, 196360232 genmoves, 22.63% captures le 383/243
ext: pawn 25361, rcp 314976, chk 803503, repchk 120142, null 17567, prune
3245724
htable: 42135330 store, 1325 rejected, 50499110 probe, 22.0% f/p, 26.4% f/s
entries 5333331 age 1 renew 0
interior recognizer probes 0, found 0 egtb probes 0, found 0 max_depth 31
It shows that Yace's position learning really works.
This is what I get when I move away the learned files ( no more elegant way
availlable to disable learned positions?):
white ( 1): input: setboard 2kr1b1r/ppp1pppp/2n5/2N1P2q/Q1bP4/4BN1P/PP3PP1/R4RK1
b - - 0 1
Stored 0 learned positions into hash table
black ( 1): input: analyze
usetime = 19.93, mintime = 9.97 maxtime = 49.83 tl 598.00 ml 40
555875 1.299 -0.32 7t 1...Be2 2.Rfc1 Bxf3 3.gxf3 Kb8H 4.Qb5H Qg6+H
5.Kf1H b6H {0}
700091 1.596 -0.31 7t+ 1...Bxf1 2.Rxf1 a6 3.Nxa6 Qg6 4.Bg5
772743 1.772 -0.27 7t 1...Bxf1 2.Rxf1 a6 3.Qb3 b6 4.Qa4 Na5 5.Nxa6
{80}
925652 2.091 -0.27 7. 1...Bxf1 2.Rxf1 a6 3.Qb3 b6 4.Qa4 Na5 5.Nxa6
{80}
1212344 2.648 -0.13 8t 1...Bxf1 2.Rxf1 a6 3.Qb3 b6 4.Qa4 Na5 5.Rc1
bxc5 6.Qxa5 {160}
1695268 3.628 -0.13 8. 1...Bxf1 2.Rxf1 a6 3.Qb3 b6 4.Qa4 Na5 5.Rc1
bxc5 6.Qxa5 {160}
2468407 4.999 -0.19 9t 1...Bxf1 2.Rxf1 a6 3.Qb3 Na5 4.Qb4 b6 5.Nxa6 e6
6.Qb5 {80}
3768929 7.488 -0.19 9. 1...Bxf1 2.Rxf1 a6 3.Qb3 Na5 4.Qb4 b6 5.Nxa6 e6
6.Qb5 {80}
9518041 19.472 -0.32 10t 1...Bxf1 2.Kxf1 a6 3.Qb3 b6 4.Nxa6 Nb8 5.Nb4 e6
6.Bg5 Rd7 7.Rc1 {80}
16494106 34.082 -0.32 10. 1...Bxf1 2.Kxf1 a6 3.Qb3 b6 4.Nxa6 Nb8 5.Nb4 e6
6.Bg5 Rd7 7.Rc1 {80}
31863417 1:05.3 -0.62 11t 1...Bxf1 2.Rxf1 a6 3.Rc1 Nb8 4.Qa5 Nc6 5.Qc3
Qg6 6.Nh4 Qh5 7.Nxb7 {HT} {80}
input: quit
33388430 1:09.9 -0.62 11u. 1...Bxf1 2.Rxf1 a6 3.Rc1 Nb8 4.Qa5 Nc6 5.Qc3
Qg6 6.Nh4 Qh5 7.Nxb7 {HT} {80}
33388430 Nodes, 40.50% Leavenodes, 477537 Nodes/sec
25296013 eval, 34.93% score, 125305349 genmoves, 21.04% captures le 353/206
ext: pawn 11362, rcp 213399, chk 560426, repchk 77351, null 8794, prune 2490092
htable: 26497886 store, 993 rejected, 32746974 probe, 25.7% f/p, 31.8% f/s
entries 5333331 age 1 renew 0
interior recognizer probes 0, found 0 egtb probes 0, found 0 max_depth 30
So the move is perfectly reproducable at depth 10 ( only -0.32 instead of -0.33
in the game ?!)
Yace.ini used during the tests:
REM initialization file for Yace
; If you want to put all logfiles in a directory, i.e. first create the
; subdirectory logs and then uncomment the line.
log_path logs\
auto_log
; Use 30 Megabytes for hash tables. Change this to a size suitable for
; your hardware. You can use numbers like 10.5M, 10.5m, 10500k or 105000000.
hash 64M
; set the chache size for endgame tables. Default is 2M. I suggest to use
; not less than 1M and not more than 8M.
egtb_cache 8M
; Where to find the endgame tables, Nalimov format, compressed or uncompressed
; If you don't have endgame tables, delete the following line
tbldir c:\tbs
; save all my games
sgames yacegam.pgn
; positional learning now is on by default
book_learn 6
; to disable book learning, I suggest
; book_learn 4
; Yes - 4 and not 0, so that Yace will have access to NAGs in dblearn.bin
bbpath c:\bb
noise 500k
resign om
rmoves 6
rscore -9
Regards,
Peter
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.