Galjoen 0.39 released

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
Werner Taelemans
Posts: 91
Joined: Mon Feb 03, 2014 10:57 am
Location: Belgium
Contact:

Re: Galjoen 0.39 released

Post by Werner Taelemans » Wed Aug 15, 2018 12:32 pm

Graham Banks wrote:
Wed Aug 15, 2018 8:39 am
(under ChessGUI).
Indeed. I'll take a look at it.

User avatar
Graham Banks
Posts: 31331
Joined: Sun Feb 26, 2006 9:52 am
Location: Auckland, NZ

Re: Galjoen 0.39 released

Post by Graham Banks » Wed Aug 15, 2018 10:04 pm

Werner Taelemans wrote:
Wed Aug 15, 2018 12:32 pm
Graham Banks wrote:
Wed Aug 15, 2018 8:39 am
(under ChessGUI).
Indeed. I'll take a look at it.
Thanks.
I was hoping to include it in my Amateur Division 7, which is starting on Saturday.
My email addresses:
gbanksnz at gmail.com
gbanksnz at yahoo.co.nz

User avatar
Werner Taelemans
Posts: 91
Joined: Mon Feb 03, 2014 10:57 am
Location: Belgium
Contact:

Galjoen 0.39.2

Post by Werner Taelemans » Thu Aug 16, 2018 6:54 pm

Graham Banks wrote:
Wed Aug 15, 2018 10:04 pm
I was hoping to include it in my Amateur Division 7, which is starting on Saturday.
That would be great :)
The bug should be fixed in version 0.39.2

User avatar
Graham Banks
Posts: 31331
Joined: Sun Feb 26, 2006 9:52 am
Location: Auckland, NZ

Re: Galjoen 0.39.2

Post by Graham Banks » Thu Aug 16, 2018 7:23 pm

Werner Taelemans wrote:
Thu Aug 16, 2018 6:54 pm
Graham Banks wrote:
Wed Aug 15, 2018 10:04 pm
I was hoping to include it in my Amateur Division 7, which is starting on Saturday.
That would be great :)
The bug should be fixed in version 0.39.2
Thanks. Will try it out shortly. :)
My email addresses:
gbanksnz at gmail.com
gbanksnz at yahoo.co.nz

User avatar
Graham Banks
Posts: 31331
Joined: Sun Feb 26, 2006 9:52 am
Location: Auckland, NZ

Re: Galjoen 0.39.2

Post by Graham Banks » Thu Aug 16, 2018 9:20 pm

Werner Taelemans wrote:
Thu Aug 16, 2018 6:54 pm
Graham Banks wrote:
Wed Aug 15, 2018 10:04 pm
I was hoping to include it in my Amateur Division 7, which is starting on Saturday.
That would be great :)
The bug should be fixed in version 0.39.2
Works well now. Thanks, Werner. :)
My email addresses:
gbanksnz at gmail.com
gbanksnz at yahoo.co.nz

Gabor Szots
Posts: 93
Joined: Sat Jul 21, 2018 5:43 am
Location: Szentendre, Hungary
Full name: Gabor Szots

Re: Galjoen 0.39.2

Post by Gabor Szots » Fri Aug 17, 2018 7:33 am

Werner Taelemans wrote:
Thu Aug 16, 2018 6:54 pm
Graham Banks wrote:
Wed Aug 15, 2018 10:04 pm
I was hoping to include it in my Amateur Division 7, which is starting on Saturday.
That would be great :)
The bug should be fixed in version 0.39.2
Can we combine 0.39 and 0.39.2 results? Or have you made changes affecting strength, too?
Gabor Szots
CCRL testing group

User avatar
Werner Taelemans
Posts: 91
Joined: Mon Feb 03, 2014 10:57 am
Location: Belgium
Contact:

Re: Galjoen 0.39.2

Post by Werner Taelemans » Fri Aug 17, 2018 9:36 am

Gabor Szots wrote:
Fri Aug 17, 2018 7:33 am
Can we combine 0.39 and 0.39.2 results?
No problem, they have the same strength. Also, Arena doesn't trigger this bug.

tpoppins
Posts: 750
Joined: Tue Nov 24, 2015 8:11 pm
Location: upstate

Re: Galjoen 0.39 released

Post by tpoppins » Mon Aug 20, 2018 2:04 pm

I haven't tested v0.39.x yet but thought I'd report a drastic eval anomaly exhibited by v0.38:



The last 50+ moves are 0.00 according to Galjoen's opponent and this eval is also confirmed by SF9. Yet Galjoen persisted with an eval between +250 and +300 cp. Perhaps its passed-pawn bonus is way too high?
Tirsa Poppins
CCRL

User avatar
Werner Taelemans
Posts: 91
Joined: Mon Feb 03, 2014 10:57 am
Location: Belgium
Contact:

Re: Galjoen 0.39 released

Post by Werner Taelemans » Tue Aug 21, 2018 5:12 pm

tpoppins wrote:
Mon Aug 20, 2018 2:04 pm
Perhaps its passed-pawn bonus is way too high?
That's possible. But the eval of 0.39 looks better. When I compare the first 10 plies in the position right after the last pawn capture:
[D]8/Pb6/8/2Bp4/3Pp2R/4K3/7r/6k1 w - - 1 137

between SF-9 and Galjoen 0.39.2:

Code: Select all

Stockfish 9 64 BMI2 by T. Romstad, M. Costalba, J. Kiiski, G. Linscott
uci
id name Stockfish 9 64 BMI2
id author T. Romstad, M. Costalba, J. Kiiski, G. Linscott

option name Debug Log File type string default 
option name Contempt type spin default 20 min -100 max 100
option name Threads type spin default 1 min 1 max 512
option name Hash type spin default 16 min 1 max 131072
option name Clear Hash type button
option name Ponder type check default false
option name MultiPV type spin default 1 min 1 max 500
option name Skill Level type spin default 20 min 0 max 20
option name Move Overhead type spin default 30 min 0 max 5000
option name Minimum Thinking Time type spin default 20 min 0 max 5000
option name Slow Mover type spin default 89 min 10 max 1000
option name nodestime type spin default 0 min 0 max 10000
option name UCI_Chess960 type check default false
option name SyzygyPath type string default <empty>
option name SyzygyProbeDepth type spin default 1 min 1 max 100
option name Syzygy50MoveRule type check default true
option name SyzygyProbeLimit type spin default 6 min 0 max 6
uciok
position fen 8/Pb6/8/2Bp4/3Pp2R/4K3/7r/6k1 w - - 1 137 moves
go infinite
info depth 1 seldepth 1 multipv 1 score cp 133 nodes 53 nps 26500 tbhits 0 time 2 pv h4h2 g1h2
info depth 2 seldepth 2 multipv 1 score cp 133 nodes 124 nps 62000 tbhits 0 time 2 pv h4h2 g1h2
info depth 3 seldepth 3 multipv 1 score cp 103 nodes 192 nps 96000 tbhits 0 time 2 pv c5e7 h2h4 e7h4
info depth 4 seldepth 4 multipv 1 score cp 110 nodes 374 nps 187000 tbhits 0 time 2 pv h4h2 g1h2 c5e7 h2g3 e7d6 g3g4
info depth 5 seldepth 6 multipv 1 score cp 106 nodes 659 nps 219666 tbhits 0 time 3 pv h4h2 g1h2 c5e7 h2g2 e7d6
info depth 6 seldepth 8 multipv 1 score cp 163 nodes 3541 nps 505857 tbhits 0 time 7 pv h4f4 h2h3 e3e2 h3h2 e2e1 h2b2 f4f8 g1h2 a7a8q
info depth 7 seldepth 9 multipv 1 score cp 163 nodes 5102 nps 510200 tbhits 0 time 10 pv h4f4 h2h3 e3e2 h3h2 e2e1 h2b2 f4f8 g1h2 a7a8q
info depth 8 seldepth 10 multipv 1 score cp 163 nodes 5880 nps 534545 tbhits 0 time 11 pv h4f4 h2h3 e3e2 h3h2 e2e1 h2b2 f4f8 g1h2 a7a8q b7a8 f8a8
info depth 9 seldepth 19 multipv 1 score cp 176 nodes 24159 nps 1271526 tbhits 0 time 19 pv h4f4 h2h3 e3e2 b7a6 e2e1 a6b7 f4f8 g1g2 f8b8 b7c6 e1e2 h3f3 b8g8 f3g3
info depth 10 seldepth 18 multipv 1 score cp 140 nodes 35052 nps 1524000 tbhits 0 time 23 pv h4f4 h2h3 e3e2 b7a6 e2e1 h3e3 e1d2 e3d3 d2c2 a6b7 f4f7 b7c6 f7f8 g1g2 a7a8q

Code: Select all

uci
id name Galjoen 0.39.2
id author Werner Taelemans
option name Hash type spin default 128 min 6 max 4096
option name Ponder type check default true
option name OwnBook type check default false
option name UCI_LimitStrength type check default false
option name UCI_Elo type spin default 2000 min 1500 max 2200
option name UCI_Chess960 type check default false
option name MultiPV type spin default 1 min 1 max 256
option name Debug type check default false
uciok
position fen 8/Pb6/8/2Bp4/3Pp2R/4K3/7r/6k1 w - - 1 137 moves
go infinite
info depth 4 seldepth 6 score cp 167 time 2 nodes 536 nps 268000 pv h4h2 g1h2 c5d6 h2g2 
info depth 6 seldepth 8 score cp 158 time 7 nodes 2004 nps 286000 pv h4h2 g1h2 c5d6 h2g2 d6e5 b7c6 
info depth 7 seldepth 12 score cp 155 time 12 nodes 3981 nps 331000 pv h4g4 h2g2 g4g2 g1g2 c5a3 g2f1 a3d6 f1e1 
info depth 8 seldepth 12 score cp 149 time 17 nodes 10528 nps 619000 pv h4h2 g1h2 e3f4 h2g2 c5d6 g2f2 f4e5 
info depth 9 seldepth 12 score cp 151 time 22 nodes 17296 nps 786000 pv h4h2 g1h2 c5a3 h2g2 a3d6 b7c6 d6c7 g2f1 
info depth 10 seldepth 17 score cp 150 time 27 nodes 22935 nps 849000 pv h4g4 h2g2 g4g2 g1g2 c5d6 b7c6 d6e5 g2f1 e5g3 f1g2 g3d6 
the evals are very alike. Stockfish's eval goes down though, when it looks deeper. With the help of the search, it seems to detect that no progress can be made.

Post Reply