EGBB implementation

[Daniel Shawul]

The one i am using at office with quad with xp machine. The single core laptop is vista. I will try and see if the problem occurs with multi-cpu vista machine.
Thanks for your help

[Kempelen]

I am suspecting it is a bad variable initialization. I mean: I do a test with crash at a certain point. Then I put a few printf here and there and the dll crashes in other position.... quit strange!, this looks as a initialization bug.

[Desperado]

ok, downloaded arena 2.0.1 to view the logfile.

Attention please

1: bright a using egbb on my position

Code: Select all

305.596-->1:quit
306.662*1*-------------------------Starte Motor 1 Bright-0.3a-------------------------
306.668*1*Configured Engine 1 Type:   UCI
306.797*1*Engine 1 dir: C:\Program Files (x86)\ChessBase\Engines\UciEngines\Bright\Bright 0.3a
306.838*1*Engine 1 commandline: "C:\Program Files (x86)\ChessBase\Engines\UciEngines\Bright\Bright 0.3a\bright-0.3a.exe" 
306.882*1*Engine 1 ProcessID: 4304
307.026-->1:uci
307.128<--1&#58;EgbbProbe 3.3.1 by Daniel Shawul
312.059<--1&#58;Loading egbbs....
312.059<--1&#58;Egbbs loaded !
312.111<--1&#58;bright-0.3a
312.162<--1&#58;&#40;c&#41; 2007 AJ Siemelink
312.214<--1&#58;All rights reserved
312.266<--1&#58;inifile=bright.ini
312.317<--1&#58;book found&#58; drdeeb-ii.bbk, 861662 positions
312.318<--1&#58;egbb.3men=1
312.320<--1&#58;egbb.4men=1
312.370<--1&#58;egbb.5men=1
312.423<--1&#58;EGBB&#58; loaded, maxmen=5, ram=4
312.475<--1&#58;!--using parallel search with 2 threads!
312.476<--1&#58;starting main search thread...
312.476<--1&#58;starting worker1...
312.527<--1&#58;started 1 new worker threads
312.528<--1&#58;+---+---+---+---+---+---+---+---+
312.580<--1&#58;8 |*r*|*n*|*b*|*q*|*k*|*b*|*n*|*r*|
312.633<--1&#58;+---+---+---+---+---+---+---+---+
312.685<--1&#58;7 |*p*|*p*|*p*|*p*|*p*|*p*|*p*|*p*|
312.735<--1&#58;+---+---+---+---+---+---+---+---+
312.788<--1&#58;6 |   |   |   |   |   |   |   |   |
312.789<--1&#58;+---+---+---+---+---+---+---+---+
312.790<--1&#58;5 |   |   |   |   |   |   |   |   |
312.843<--1&#58;+---+---+---+---+---+---+---+---+
312.895<--1&#58;4 |   |   |   |   |   |   |   |   |
312.900<--1&#58;+---+---+---+---+---+---+---+---+
312.902<--1&#58;3 |   |   |   |   |   |   |   |   |
312.955<--1&#58;+---+---+---+---+---+---+---+---+
313.006<--1&#58;2 |&#40;P&#41;|&#40;P&#41;|&#40;P&#41;|&#40;P&#41;|&#40;P&#41;|&#40;P&#41;|&#40;P&#41;|&#40;P&#41;|
313.058<--1&#58;+---+---+---+---+---+---+---+---+
313.110<--1&#58;1 |&#40;R&#41;|&#40;N&#41;|&#40;B&#41;|&#40;Q&#41;|&#40;K&#41;|&#40;B&#41;|&#40;N&#41;|&#40;R&#41;|  40-40=0
313.161<--1&#58;+---+---+---+---+---+---+---+---+
313.213<--1&#58;a   b   c   d   e   f   g   h
313.265<--1&#58;rnbqkbnr/pppppppp/////PPPPPPPP/RNBQKBNR w KQkq - 0 0
313.317<--1&#58;hash=6975491b85d41ed6
313.320<--1&#58;phash=d2fbd4e9dcbfcead
313.321<--1&#58;%
313.375<--1&#58;id name bright-0.3a
313.427<--1&#58;id author AJ Siemelink
313.481<--1&#58;option name Hash            type spin  default 32 min 1 max 4096
313.533<--1&#58;option name Ponder          type check default false
313.585<--1&#58;option name OwnBook         type check default true
313.640<--1&#58;option name Threads         type spin default 2 min 1 max 8
313.693<--1&#58;option name UCI_Opponent    type string
313.761<--1&#58;option name UCI_EngineAbout type string default bright by AJ Siemelink
313.765<--1&#58;uciok
313.768-->1&#58;setoption name Hash value 128
313.773-->1&#58;isready
314.300<--1&#58;readyok
339.472*1*Start calc, move no&#58; 0
339.476-->1&#58;ucinewgame
339.480-->1&#58;isready
339.486<--1&#58;readyok
339.494-->1&#58;setoption UCI_Opponent value none 2400 human Bright-0.3a

339.498-->1&#58;position fen 8/8/5P2/8/3B2k1/1P6/1K6/8 w - - 0 1

339.504-->1&#58;go wtime 300000 btime 300000 winc 0 binc 0
339.509<--1&#58;info depth 1
339.513<--1&#58;info currmove d4c5 currmovenumber 1 time 0 nodes 0 nps 0 hashfull 0
339.524<--1&#58;info depth 1 seldepth 1 score cp 1280 time 0 nodes 1 nps 1 pv d4c5
339.567<--1&#58;info depth 1 seldepth 1 score cp 1372 time 0 nodes 10 nps 10 pv f6f7
339.574<--1&#58;info depth 2
339.575<--1&#58;info currmove f6f7 currmovenumber 1 time 0 nodes 18 nps 18 hashfull 0
339.576<--1&#58;info depth 2 seldepth 1 score cp -4525 time 0 nodes 19 nps 19 pv f6f7
339.586<--1&#58;info depth 2 seldepth 1 score cp 0 time 0 nodes 28 nps 28 pv d4c5

CRASHING WITHOUT ANY REPORT...

on some other runs on fermins position with white to move,
the engine send Kh5g6 as bestmove (WRONG COLOR) and of
course it is reported as error.

...

[Desperado]

same for toga...

Code: Select all

2.275.767-->1:setoption name Toga Extended History Pruning value false
2.275.770-->1:setoption name Toga King Safety Margin value 1700
2.275.771-->1:setoption name Toga King Safety value false
2.275.774-->1:setoption name Toga Lazy Eval Margin value 200
2.275.775-->1:setoption name Toga Lazy Eval value true
2.275.778-->1:setoption name Passed Pawns value 100
2.275.781-->1:setoption name Pawn Structure value 100
2.275.782-->1:setoption name King Safety value 100
2.275.785-->1:setoption name Piece Square Activity value 100
2.275.786-->1:setoption name Piece Activity value 100
2.275.789-->1:setoption name Material value 100
2.275.790-->1:setoption name Quiescence Check Plies value 1
2.275.793-->1:setoption name Delta Margin value 50
2.275.796-->1:setoption name Delta Pruning value true
2.275.799-->1:setoption name Extended Futility Margin value 300
2.275.801-->1:setoption name Futility Margin value 100
2.275.804-->1:setoption name Futility Pruning value true
2.275.808-->1:setoption name History Threshold value 70
2.275.811-->1:setoption name History Pruning value true
2.275.816-->1:setoption name Verification Reduction value 5
2.275.819-->1:setoption name Verification Search value Always
2.275.822-->1:setoption name NullMove Reduction value 3
2.275.826-->1:setoption name NullMove Pruning value Always
2.275.829-->1:setoption name MultiPV value 1
2.275.833-->1:setoption name Bitbases Cache Size value 16
2.275.834-->1:setoption name Bitbases Path value c:/egbb/
2.275.838-->1:setoption name BookFile value performance.bin
2.275.842-->1:setoption name Ponder value false
2.275.845-->1:setoption name Search Depth value 0
2.275.851-->1:setoption name Search Time value 0
2.275.884<--1&#58;EgbbProbe not Loaded!
2.275.885<--1&#58;EgbbProbe 3.3.1 by Daniel Shawul
2.280.819<--1&#58;Loading egbbs....
2.280.823<--1&#58;Egbbs loaded !
2.311.231*1*Start calc, move no&#58; 0
2.311.232-->1&#58;ucinewgame
2.311.234-->1&#58;isready
2.311.309<--1&#58;readyok
2.311.312-->1&#58;position fen 8/8/5P2/8/3B2k1/1P6/1K6/8 w - - 0 1
2.311.312-->1&#58;go wtime 300000 btime 300000 winc 0 binc 0
2.311.315<--1&#58;info depth 1
2.311.316<--1&#58;info multipv 1 depth 1 seldepth 1 score cp 994 time 0 nodes 2 pv b2a1
2.311.341<--1&#58;info multipv 1 depth 1 seldepth 1 score cp 1022 time 15 nodes 3 pv b2b1
2.311.356<--1&#58;info multipv 1 depth 1 seldepth 1 score cp 1033 time 15 nodes 4 pv b2c1
2.311.363<--1&#58;info multipv 1 depth 1 seldepth 1 score cp 1058 time 15 nodes 6 pv b2c2
2.311.372<--1&#58;info multipv 1 depth 1 seldepth 2 score cp 1060 time 15 nodes 19 pv b3b4 g4f5
2.311.379<--1&#58;info multipv 1 depth 1 seldepth 6 score cp 1434 time 15 nodes 80 pv f6f7 g4g5
2.311.387<--1&#58;info depth 1 seldepth 6 time 15 nodes 160 nps 0
2.311.390<--1&#58;info depth 2
2.311.391<--1&#58;info multipv 1 depth 2 seldepth 6 score cp 0 time 15 nodes 82 pv f6f7
2.311.401<--1&#58;info depth 2 seldepth 6 time 15 nodes 196 nps 0
2.311.404<--1&#58;info depth 3
2.311.405<--1&#58;info multipv 1 depth 3 seldepth 6 score cp 0 time 15 nodes 100 pv f6f7
2.311.413<--1&#58;info multipv 1 depth 3 seldepth 6 score cp 5249 time 15 nodes 112 pv b2c2 g4f5
2.311.430<--1&#58;info multipv 1 depth 3 seldepth 6 score cp 5268 time 15 nodes 130 pv b2c3 g4f5
2.311.439<--1&#58;info depth 3 seldepth 6 time 15 nodes 312 nps 0
2.311.441<--1&#58;info depth 4
2.311.444<--1&#58;info multipv 1 depth 4 seldepth 6 score cp 5268 time 15 nodes 166 pv b2c3 g4f5
2.311.446<--1&#58;info depth 4 seldepth 6 time 15 nodes 392 nps 0
2.311.449<--1&#58;info depth 5
2.311.450<--1&#58;info multipv 1 depth 5 seldepth 6 score cp 0 time 16 nodes 264 pv b2c3 g4g3 c3c4
2.311.463<--1&#58;info multipv 1 depth 5 seldepth 6 score cp 5279 time 32 nodes 708 pv d4e5 g4f5 b2c2
2.311.479<--1&#58;info multipv 1 depth 6 seldepth 6 score cp 5180 time 32 nodes 862 pv d4e5 g4f5 b2c3 f5e4

//no report,only crashes

on a second run it crashed exactly again after the output of "nodes 862"...

[Daniel Shawul]

Bright works in arena here,i guess it is a system dependent issue. Are you sure you are using the new bitbases (224mb in size) witn v3.3, because if you use the old ones (about 360mb) with the v3.3 it will definately crash. Please check the size of egbbs. If that is not it, the only way to find out for me is to find a vista machine with multiple cores and test directly.

[Desperado]

1:
=====

Total fileSize of all *.cmp: 223 MB (234.479.030 Bytes)
egbbdll: 23.01.2009 size 86016Bytes (84KB)

Ok, 1. my totalsize was about 300MB ( )

2:
=====

crashing continues...


wait a second...

FIXED:

using the own compiled dlls of today with the correct tablebases

bright 0.3a ok!
bright 0.4a ok!
cyclone 3.4 ok!
Toga 14beta5c ok!

me OK

conclusion:
=========

- seems as i was reinstalling the wrong bitbases today
- when i opened the thread maybe various things did go wrong
(wrong dlls,bitbases mixed... dont know)

well now there is definitely no bug reported on the position.

Thx Daniel very much!

@Fermin:

using the dlls i mailed you, on the folder "spinlocked32/64" with the
correct bitbases should work now.

[Daniel Shawul]

Nothing more.
thanks Michael and Fermin,
Daniel

[Kempelen]

@Fermin:

using the dlls i mailed you, on the folder "spinlocked32/64" with the
correct bitbases should work now.

Could you tell me exactly the url from where you downloaded the correct bitbases...? I dont know now if the one I have is the correct....

[Desperado]

hm, i dont really remember, i will help searching...

[Desperado]

http://wbec-ridderkerk.nl/html/details1/Scorpio.html

these seem to be the right ones.

Good luck, and enjoy. Hope it will work for you too