Texel 1.07

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
Rebel
Posts: 4541
Joined: Thu Aug 18, 2011 10:04 am

Re: Texel 1.07

Post by Rebel » Sat Sep 30, 2017 5:58 pm

petero2 wrote: You should be aware though that this causes texel to report 872 UCI parameters,
:)

User avatar
Graham Banks
Posts: 32876
Joined: Sun Feb 26, 2006 9:52 am
Location: Auckland, NZ

Re: Texel 1.07

Post by Graham Banks » Sat Sep 30, 2017 6:25 pm

petero2 wrote:Version 1.07 of my chess engine Texel is now available for download.
Excellent news.
Thanks for your efforts. :)
My email addresses:
gbanksnz at gmail.com
gbanksnz at yahoo.co.nz

MikeB
Posts: 3267
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: Texel 1.07 macOS build

Post by MikeB » Sat Sep 30, 2017 6:36 pm

link to a PGO Build macOS executable

https://www.dropbox.com/s/lknu09kvtj7f9 ... S.zip?dl=1

should work on most macs with an i7 core

Code: Select all

uci
id name Texel 1.07
id author Peter Osterlund
option name UCI_EngineAbout type string default Texel 1.07 by Peter Osterlund, see http://hem.bredband.net/petero2b/javachess/index.html#texel
option name Hash type spin default 16 min 1 max 524288
option name OwnBook type check default false
option name BookFile type string default 
option name Ponder type check default true
option name UCI_AnalyseMode type check default false
option name UCI_Opponent type string default 
option name Strength type spin default 1000 min 0 max 1000
option name Threads type spin default 1 min 1 max 512
option name MultiPV type spin default 1 min 1 max 256
option name UseNullMove type check default true
option name GaviotaTbPath type string default 
option name GaviotaTbCache type spin default 1 min 1 max 2047
option name SyzygyPath type string default 
option name MinProbeDepth type spin default 1 min 0 max 100
option name AnalysisAgeHash type check default true
option name Clear Hash type button
option name BufferTime type spin default 1000 min 1 max 10000
uciok

pseudo bench ( go depth 20) 

info nodes 76040820 nps 1374438 hashfull 1000 time 55325
bestmove e2e4 ponder e7e6

pseudo bench using 18 threads on a 2010 12 core Mac Pro with 2GB hash ( go depth 20)
info nodes 49466301 nps 17104530 hashfull 1000 time 2892
bestmove e2e4 ponder e7e6 

two questions to Peter on this position
[d]8/8/8/1p5r/p1p1k1pN/P2pBpP1/1P1K1P2/8 b - - 0 1
what does mate 945 mean? see output below and why does nps drop off so dramatically after the mate announcement at depth 28 ?

Code: Select all

position fen 8/8/8/1p5r/p1p1k1pN/P2pBpP1/1P1K1P2/8 b - - 0 1
setoption name hash threads value 18
setoption name hash value 2048
setoption name syzygypath value /applications/tablebases
info string Found 510 syzygy tablebases
go infinite
info string eval mtrlraw:-11
info string eval qncorr :-11
info string eval trade  :-11
info string eval majred :-99
info string eval imbala :-132
info string eval pst    :-165
info string eval pawn   :-304
info string eval castle :-304
info string eval rook   :-346
info string eval bishop :-332
info string eval knight :-336
info string eval threat :-344
info string eval protect:-333
info string eval king   :-333
info string eval endgame:-333
info string eval halfmove:-333
info string eval staleP :-335
info string eval total  :-3.35
...
info depth 28 score cp 5096 lowerbound time 197717 nodes 4922045247 nps 24894395 tbhits 5368092 pv b5b4 a3b4 h5h4 d2c3 h4h1 b4b5 h1b1 e3d2 e4d5 d2g5 a4a3 b2a3 b1b5 g5f6 b5b3 c3d2 b3a3 d2e3 a3a2 e3f4 d3d2 f4g5 d2d1q g5h6 a2e2 h6g7 d1c2 f6d8 e2e6 g7h8 c2b2 h8h7 b2b7
info nodes 4949351599 nps 24906533 hashfull 1000 tbhits 5371291 time 198717
info depth 28 score cp 6330 lowerbound time 199554 nodes 4972992423 nps 24920534 tbhits 5374957 pv b5b4 a3b4 h5h4 d2c3 h4h1 b4b5 h1b1 e3d2 e4d5 d2g5 a4a3 b2a3 b1b5 g5f6 b5b3 c3d2 b3a3 d2e3 a3a2 e3f4 d3d2 f4g5 d2d1q g5h6 a2e2 h6g7 d1c2 f6d8 e2e6 g7h8 c2b2 h8h7 b2b7

info depth 28 score mate 945 lowerbound time 199566 nodes 4973816144 nps 24923163 tbhits 5375144 pv b5b4 a3b4 h5h4 d2c3 h4h1 b4b5 h1b1 e3d2 e4d5 d2g5 a4a3 b2a3 b1b5 g5f6 b5b3 c3d2 b3a3 d2e3 a3a2 e3f4 d3d2 f4g5 d2d1q g5h6 a2e2 h6g7 d1c2 f6d8 e2e6 g7h8 c2b2 h8h7 b2b7

info nodes 4979959584 nps 24829530 hashfull 1000 tbhits 5515696 time 200566
info nodes 4987626945 nps 24744386 hashfull 1000 tbhits 5657382 time 201566
info nodes 4991749419 nps 24642582 hashfull 1000 tbhits 5811998 time 202566
info nodes 4994112080 nps 24533134 hashfull 1000 tbhits 5961052 time 203566
info nodes 4997097562 nps 24427801 hashfull 1000 tbhits 6109580 time 204566
...
info nodes 5393858862 nps 19087562 hashfull 1000 tbhits 17926349 time 282585
info nodes 5398914506 nps 19038082 hashfull 1000 tbhits 18097144 time 283585
info nodes 5405008013 nps 18992596 hashfull 1000 tbhits 18269750 time 284585
info nodes 5410111728 nps 18943963 hashfull 1000 tbhits 18442051 time 285585
info nodes 5415438766 nps 18896448 hashfull 1000 tbhits 18597247 time 286585
info nodes 5421044738 nps 18850234 hashfull 1000 tbhits 18756210 time 287585
info nodes 5425811617 nps 18801433 hashfull 1000 tbhits 18923105 time 288585
...
info nodes 5507004944 nps 17845932 hashfull 1000 tbhits 22228507 time 308586
info nodes 5510296489 nps 17798920 hashfull 1000 tbhits 22380564 time 309586
info nodes 5513941081 nps 17753061 hashfull 1000 tbhits 22534364 time 310591
info nodes 5517812023 nps 17708452 hashfull 1000 tbhits 22678059 time 311592

petero2
Posts: 580
Joined: Mon Apr 19, 2010 5:07 pm
Location: Sweden
Contact:

Re: Texel 1.07 macOS build

Post by petero2 » Sat Sep 30, 2017 9:25 pm

MikeB wrote:link to a PGO Build macOS executable

https://www.dropbox.com/s/lknu09kvtj7f9 ... S.zip?dl=1

should work on most macs with an i7 core
Thanks! I can confirm I get the same node count for a "go depth 20" search.
MikeB wrote:two questions to Peter on this position
[d]8/8/8/1p5r/p1p1k1pN/P2pBpP1/1P1K1P2/8 b - - 0 1
what does mate 945 mean? see output below and why does nps drop off so dramatically after the mate announcement at depth 28 ?

Code: Select all

info depth 28 score mate 945 lowerbound time 199566 nodes 4973816144 nps 24923163 tbhits 5375144 pv b5b4 a3b4 h5h4 d2c3 h4h1 b4b5 h1b1 e3d2 e4d5 d2g5 a4a3 b2a3 b1b5 g5f6 b5b3 c3d2 b3a3 d2e3 a3a2 e3f4 d3d2 f4g5 d2d1q g5h6 a2e2 h6g7 d1c2 f6d8 e2e6 g7h8 c2b2 h8h7 b2b7
"score mate 945 lowerbound" means that the position is a mate in 945 moves or fewer. Probing syzygy WDL TBs made texel able to prove the the position is a forced win for black. An upper bound on the number of moves to mate is obtained by considering how many irreversible moves are remaining on the board, and the maximum number of reversible moves between each irreversible move (which is never more than 50 due to the 50-move draw rule.)

As soon as texel has been able to prove a forced win it starts trying to find the shortest possible win. To do this it will probe syzygy DTZ and/or gaviota DTM tables. Those probes are slower than syzygy WDL probes because they require more work to decompress/decode and those TB files might also reside on a slower hard disk.

More details about texel's TB implementation are available here and here.

petero2
Posts: 580
Joined: Mon Apr 19, 2010 5:07 pm
Location: Sweden
Contact:

Re: Texel 1.07

Post by petero2 » Sat Sep 30, 2017 10:22 pm

jpqy wrote:Wow..that are a lot settings!

Is there a trick when typing uci that we can see in pages or can hold it as list is too long to see everything..

like uci page..and every time press space to see next part to scroll these settings..

Thanks Peter.
You could type:

Code: Select all

texel >uci.txt
uci
quit
Then open uci.txt in wordpad or some other text editor.

User avatar
cdani
Posts: 2104
Joined: Sat Jan 18, 2014 9:24 am
Location: Andorra
Contact:

Re: Texel 1.07

Post by cdani » Sun Oct 01, 2017 3:18 am

Congratulations!! Very nice changes :-)

Dann Corbit
Posts: 9894
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: Texel 1.07

Post by Dann Corbit » Sun Oct 01, 2017 3:26 am

Texel is a marvel.
It scales very well to large processor count.
I also enjoy examination of the code base.
For some engines, it is a chore and a bore and not a joy.

Keep it up. For a single author, you have created a true marvel.

I bow to thee sir. quite a feat.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

Dokterchen
Posts: 102
Joined: Wed Aug 15, 2007 10:18 am
Location: Munich

Re: Texel 1.07

Post by Dokterchen » Sun Oct 01, 2017 9:28 am

Good news. :D Thank you Peter!

Damir
Posts: 2098
Joined: Mon Feb 11, 2008 2:53 pm

Re: Texel 1.07

Post by Damir » Sun Oct 01, 2017 9:52 am

Now all we need is Brendan to come here and tune the parameters. :o
I am surprised he has not shown up yet. :) :)

jpqy
Posts: 518
Joined: Thu Apr 24, 2008 7:31 am
Location: Belgium

Re: Texel 1.07

Post by jpqy » Sun Oct 01, 2017 11:06 am

petero2 wrote:
jpqy wrote:Wow..that are a lot settings!

Is there a trick when typing uci that we can see in pages or can hold it as list is too long to see everything..

like uci page..and every time press space to see next part to scroll these settings..

Thanks Peter.
You could type:

Code: Select all

texel >uci.txt
uci
quit
Then open uci.txt in wordpad or some other text editor.
Thanks Peter :!:
After some tries i got it in this way via command prompt

C:\Program Files\Chess\Engines\Texel\Texel 1.07\Tuning>texel64.exe >uci.txt
uci
quit


and creat a .txt file from 57kb with all settings!
JP.

Post Reply