modified epd2wb where to download?

[Uri Blass]

In the Gerbil site I find only the old epd2wb that does not work with UCI engines.

I would like to know where to download modified epd2wb

I like to test programs on test suites by that tool and I think that by the right batch file it is possible to run many programs in many test suites in automatic way.

Unfortunately the tool in Bruce's site is only for winboard engines and not for UCI engines and it is also an old tool.

I have a modified version that I got by email from Thomas Mayer some years ago(newer than the old version in bruce site) but I am not sure if it is the newest version.

The version that I have also does not work with rybka or maybe I do not know how to make it work with rybka.

I would like to do automatic test of UCI engines in test suites by a batch file and today I know only to test winboard engines by batch file.

The following batch file simply skip rybka and test only movei when both movei and rybka are in d:\engines

start /w epd2wb Rybka v1.2f.w32 ECM.epd 40 -p d:\engines -a -f ECM3_rybka.log
start /w epd2wb Movei00_8_355 ECM.epd 40 -p d:\engines -a -f ECM3_8_355.log

Uri

[Volker Pittlik]

...
I would like to know where to download modified epd2wb
...

http://wbforum.volker-pittlik.name/viewtopic.php?t=2613

[Uri Blass]

Thanks

It works for me now with UCI engines.

I had a problem with rybka but after deleting a space from it's name there are no more problems.

Uri

[ShaunBrewer]

Hi Uri,

Glad you spotted the space issue - always annoying when that happens - one tends to look for bigger problems that are not there .

How is Rybka behaving during epd analysis - I have been using Polyglot for epd analysis - this works fine for me with Toga, Glaurung, Spike etc but does not work with Rybka - it does not seem to obey the time limit?

If you are having more luck with epd2wb - I will have to investigate

Thanks

Shaun

[Uri Blass]

Hi Uri,

Glad you spotted the space issue - always annoying when that happens - one tends to look for bigger problems that are not there .

How is Rybka behaving during epd analysis - I have been using Polyglot for epd analysis - this works fine for me with Toga, Glaurung, Spike etc but does not work with Rybka - it does not seem to obey the time limit?

If you are having more luck with epd2wb - I will have to investigate

Thanks

Shaun

Yes

I have luck with epd2wb

Rybka solves 281 out of 300 of WAC in 1 second per position on my A3000.

failures:
2 71 80 92 100 141 157 163 196 222 229 230 235 265 272
275 287 293 297

For comparison
knightdreamer3.1 solve in the same time 294 out of 300

Failure 100 163 213 230 243 269
Uri

[Volker Pittlik]

... I have been using Polyglot for epd analysis...

Can you explain how you have done that? I don't find anything about that in the readme.

Volker

[ShaunBrewer]

... I have been using Polyglot for epd analysis...

Can you explain how you have done that? I don't find anything about that in the readme.

Volker

Hi Volker

Its an undocumented feature

In the directory where I have Polyglot installed

I have the following command in a .bat file:

Code: Select all

polyglot epd-test -epd GF1hour(2).epd -min-time 64 -max-time 3840 -depth-delta 5 > GF1hour(2)(64mLaptop-GF).txt


GF1hour(2).epd is my test suite
min-time is in seconds - 64 seconds
max-time is in seconds - 3840 secomds (64 minutes)
depth-delta assumed found when solution does not change over 5 iterations

.ini

Code: Select all



[PolyGlot]

EngineDir = .
EngineCommand = GF.exe

Log = false
LogFile = GFruit-4bx.log

[Engine]

Hash = 256

NalimovPath = C:\TB
NalimovCache = 128



Sample output from MATS suite test - Gambit Fruit 5sec per move:

Code: Select all

PolyGlot 1.4 by Fabien Letouzey
MATS001 1    1    1 -  6   0.08     44312  +0.91 d6 O-O dxe7 Rxc3 exf8=R+ Kxf8 bxc3 Bxe4 Rxe4 Qb2 Rd1 Qc2 Rd2 Qc1+ Re1 Qxc3
MATS002 1    2    2 -  7   0.38    195596  +0.68 e4 dxe4 Nxe4 Nxe4 Bxe4 Ra7 Bxh7+ Kh8 Bd3 Nc6 O-O Bg4 Rae1 Bxe2 Rxe2 Nxd4
MATS003 1    3    3 -  3   0.00      3570  +3.41 Qg4 Qb5 Qc4 Qd7 Qc7 Qb5 a4 Qxa4 b3 Rf8 Qxc8 Qxb3
MATS004 1    4    4 -  8   0.50    356706  +4.63 Rxc3 Bxb7 Rxg3 Rac1 Rh3 Rc2 Nf2+ Qxf2 Bxf2 Rcxf2 Rhd3 Rc2 Rd1
MATS005 0    4    5 -  6   0.06     50698  +0.00 Bd8 a3 Bf6 h3 Bd4+ Kh2 Be5+ Kg1 Kg6 Nc3 Bd4+ Kh1 Bf6 Kh2 Be5+ Kg1
MATS006 0    4    6 -  8   0.45    283999  +1.03 Rxc8+ Qxc8 h5 Nf8 Bd6 Qe8 Qb4 Nd7 Qa4 b5 Qxa7 Qxe4 Qa8+ Nf8
MATS007 1    5    7 -  1   0.00         4  +0.30 b4 Bb3 Kc8 Bc4 Kd7 Bb3 Kc7 Bd1 Kb6 Bb3 Kc5 g4 Kb6 Bc4 Kc5
MATS008 0    5    8 -  2   0.00       219  +0.28 b4 Nd7 a4 Bb7 axb5 Rfc8 bxa6 Rxa6 Rxa6 Bxa6 Ra1 Rxc3 Rxa6 Rxc2 Ra8+ Nf8 b5
MATS009 1    6    9 - 11   0.34    371266  +0.39 Rc5 bxc6 e4 dxe4 fxe4 Ke7 Ke2 Kd6 Ra5 Ra8 Kd3 Ke6 Ke3 f6
MATS010 0    6   10 -  4   0.00      3744  +0.01 Rac1 Ne6 Qd3 Ng5 Rc3 Rf6 Rc5 b6 Rc2 Ne4 Nxe4 Rxe4
MATS011 0    6   11 -  5   0.04     19197  +0.83 Rfe1 Qa5 Qg3 Kh8 Rc1 Rad8 Qf3 Bb4 Re2 Qb5
MATS012 1    7   12 -  7   0.20    122051  +1.37 Be6 Qc2 Be3+ Kh1 O-O Ra4 Qh4 Rxd4 Bxd4 Nc5 Bxc5 Qxc5 Rxf4
MATS013 1    8   13 -  6   0.09     58722  +0.42 Nxf5 Bxc3 Nxd6 Qb8 Nxe4 Bf6 Rab1 Qe8 Qf3 Rd8 e3 Qg6
MATS014 1    9   14 -  8   0.16    130967  +0.20 c4 bxc4 Re5 c3 Be6 g3 Ra5 Rf2 Ra3 e5 dxe5
MATS015 1   10   15 -  1   0.00       197  +0.78 Qh4 Qc2 Bb4 Bd2 Bxd2 Qxd2 Qxc4 Nd6 Qf4 Qxf4 Nxf4 Nxb7
MATS016 1   11   16 -  5   0.02      8860  -0.29 hxg4 fxg4 f5 g5 Bg7 Ne3 f4 Ng4 Ng8 Kf2 Bb7 Kf3
MATS017 0   11   17 - 10   0.42    374288  -0.81 Bg1 Bg4 Bc5 Bc7 Bg1 Kf3 Bc5 Ke2 Bg1 Nb4 Na7 Kd2 Bd4 Nd3 Nc8
MATS018 0   11   18 - 10   2.14   1617399  +0.55 g6 Rb1 Re3 c5 Nxd5 Rxb7 Rh3 c6 Bg4 Nb5 Rxh2 Nxc7 Nxc7 Rxc7
MATS019 0   11   19 - 12   1.82   1653178  +0.41 Rc8 Rb1 Rc7 Rb5 Kg7 Bf3 h5 Kf2 Rc2+ Ke3 Rxh2 Rxb7 Ra2 Be2 Bf5
MATS020 1   12   20 -  9   0.36    322229  -0.82 h5 R6d4 Rh6 Ra4 Ra8 Ra5 Ke7 Rg5 g6 Rc5 Rc8
MATS021 0   12   21 -  5   0.03     16509  +0.17 Qf3 Qxf3 Nxf3 Be7 Be3 Nxb3 axb3 Rf8 Rfd1 Rf5 Ra2 a5 c4 Nxe5 Nxe5 Rxe5 cxb5 Rxb5
MATS022 0   12   22 -  8   1.38    846440  -0.35 Qg4 Kh1 Nc5 cxd5 Nxb3 Nxb3 cxd5 Qe3 Rhg8 Rc1+ Kb8 Rf2
MATS023 1   13   23 -  7   0.07     53608  +0.08 O-O Qxd1 Raxd1 Re8 Bc5 Be6 f3 b6 Bd4 exf3 Rxf3 Rad8
MATS024 0   13   24 -  6   0.03     15637  +0.58 O-O-O Be3 Kb8 Qd2 Qf6 Rg1 Qxf3 Rxg7 Rhg8 Ng1 Qh1


From above 13 found from 24

90 sec per move

Code: Select all

PolyGlot 1.4 by Fabien Letouzey
MATS001 1    1    1 -  6   0.07     44312  +0.32 d6 Rxc3 Nxf6+ Kf8 dxe7+ Kg7 e8=N+ Rxe8 Nxe8+ Kg8 bxc3 Qd8 Nd4 d2 Re5 Bd7 Rd1 f6 Re2 Bxe8 Rdxd2 Bd7 Ne6
MATS002 1    2    2 -  7   0.38    195596  +0.73 e4 Nd7 O-O Bb7 e5 Qh6 Qc1 Qxc1 Raxc1 c5 f4 Bc6 b3 f6 Rce1
MATS003 1    3    3 -  3   0.00      3570  +3.07 Qg4 Qb5 Qc4 Qd7 Qc7 Qb5 a4 Qxa4 Re4 Rf8 Qxc8 Qxe4 Qxf8+ Kxf8 Rxe4 Bxb2 Ra4 b6 Rh4 Kg8 Re4
MATS004 1    4    4 -  8   0.50    356706  +5.86 Rxc3 Bxb7 Rxg3 Rac1 Rh3 Rc2 Be3 Bc1 Bxc1 Rfxc1 Rxh2+ Qxh2 Nxh2 Rxh2 Qxf4 Rhc2 Qh6+ Kg1 Qe3+ Kh1 Qxe5 Rc7
MATS005 0    4    5 -  6   0.06     50698  +0.10 Bd8 Kf1 Bf6 Nxc5 Rxb2 Rxb2 Bxb2 Ke2 Be5 Nd7 Bxh2 Kd2 Bf4+ Kc3 Kg6 Kxc4 Kf5 Kd5 g6 Nc5
MATS006 1    5    6 - 13  24.59  15774249  +2.12 d6 Rc5 Rxc5 bxc5 Rf5 Nf8 Rxc5 Ne6 Rb5 g6 Rb8+ Kg7 e5 f5 h5 f4 Bh4 g5
MATS007 1    6    7 -  1   0.00         4  +0.43 b4 Bb3 Kc7 Bd1 Kb6 Bc2 Kc5 Bb3 Ne8 Bc2 Nf6 Bd1 Kd6 g4 Kc5 Ke3 Nd7 Kd3 Kd6 Kc4 Nc5 Bc2 g5 hxg5 hxg5
MATS008 0    6    8 - 15  36.39  27465646  +0.43 b4 Nd7 a4 bxa4 Nxa4 Ra7 Rf3 Rc7 Rc3 Rb7 Rb3 Rc7 Ra2 Re8 Bf3 f6 Bh5 g6
MATS009 1    7    9 - 11   0.34    371266  +0.47 Rc5 bxc6 Ke2 Ke7 Ra5 Ra8 Kd3 Kd6 e4 dxe4+ Kxe4 f6 Ke3 Kc7 Ra4 f5 Kf4 Kb6 Rb4+ Kc7 Rd4 Re8 Rd2 Re1
MATS010 0    7   10 - 14  79.53  48679998  +0.00 Rad1 Rf6 Nh5 Rg6 Nf4 Rf6 Rc1 Ng6 Nh5 Rf5 Ng3 Rf6 Nh5
MATS011 0    7   11 - 12  46.95  27081091  +0.72 Rd3 Rc8 Rf3 Qb6 Qh4 Rfd8 Bh6 Qb4 a3 Qb5 Rg3 g6
MATS012 1    8   12 -  7   0.20    122051  +1.03 Be6 Qc2 Be3+ Kh1 O-O Ra4 Kh8 Rxd4 Bxd4 e4 Rd8 f5 Bf7 Rd1 Qg5 Nc5
MATS013 1    9   13 -  6   0.09     58722  +0.41 Nxf5 Bxc3 Nxd6 Qb8 Nxe4 Bf6 Nd2 a6 e4 Bd4 e5 b5 Rad1 Qc7 Nf3 bxc4 Qc2 Qb6 Nxd4 cxd4 Qxc4 Qxb2 Rxd4
MATS014 1   10   14 -  8   0.16    130967  +0.35 c4 bxc4 Re5 Rf1 Rc5 Rb1 Bxc4 a3 Be6 Rb7 a5 Ra7 Bg4 Rb7 Rc3 Ra7 Rxa3 Rxc7+ Kf6 Rb7
MATS015 1   11   15 -  1   0.01       197  +0.99 Qh4 Qc2 Qxc4 Qxc4 dxc4 Bd2 Rfd8 Bg5 Rd7 Rad1 Rxd1 Rxd1 h6 Bd8 Nxe5 Nc7 Rc8 Nxe6
MATS016 1   12   16 -  5   0.02      8860  -0.44 hxg4 fxg4 f5 g5 Bg7 Ne3 f4 Ng4 Ng8 Kf2 Bb7 Kf3 Kc7 Ba5+ Kc8 Bb6 Ne7 Nf6 b4 h5
MATS017 1   13   17 - 17  16.98  15546294  -0.95 Ba7 Bg4 Bg1 Kf3 Bc5 Kg2 Nb6 b4 Nc4 bxc5 Nxa5 c6 Nxc6 dxc6 Bxc6 Kf3 d5 Nb4 dxe4+ Ke3 Bb7 a3 a5
MATS018 1   14   18 - 15  74.46  57210769  +0.62 Na4 Kb1 Nc5 Kc1 g6 Kd2 Kg7 c3 b6 a3 Nxd3 Kxd3 f6 Nb5 Bf5+ Kd2 Rh8 Nxc7 Rxh2+ Ke3
MATS019 0   14   19 - 16  27.36  24415684  +0.55 Ra8 Kf2 Ra2 Ke3 Rb2 Rc7 b5 Rb7 b4 h4 Rb3+ Kf2 Kg7 h5 Bf5 g4 Be4 g5 h6
MATS020 1   15   20 -  9   0.37    322229  -0.59 h5 Rd7 Rh6 Ra7 Rf6 h3 Rf2 g4 h4 g5 Rd8 Rxd8+ Kxd8 g6 fxg6 Rxg7 Rf3 Rxg6 Rxe3 Rg8+ Kd7
MATS021 0   15   21 -  5   0.03     16509  +0.41 Qf3 Qxf3 Nxf3 Be7 Be3 Nxb3 axb3 Rf8 b4 Rf5 Nd4 Nxd4 Bxd4 Bg5 Rfd1 g6 f3 Kf7
MATS022 0   15   22 - 10   5.18   3107531  -0.34 Nb6 Kh1 Kb8 a4 Bb4 cxd5 Bxd2 dxc6 Bb4 cxb7 Kxb7 Rf7+ Ka8 Qf3+ e4 dxe4 Qc6 Qe3
MATS023 1   16   23 -  7   0.08     53608  -0.05 O-O Re8 Qxd8 Rxd8 Rad1 Be6 f3 exf3 Bxf3 c6 Bd6 Rd7 Rd3 Rad8 Rfd1 Bc3 g4 f6 h4
MATS024 0   16   24 -  6   0.03     15637  +0.47 O-O-O Rg1 g6 Nc3 Qe6 Rg3 Kb8 Qe2 h5 Bg5 Be7 Bxe7 Qxe7 O-O-O Nc5 Kb1


16/24 found

other command line options are: -min-depth, -max-depth - I have not used these

Alessandro Scotti and Tord Romstad helped me with the syntax

Shaun

[Tord Romstad]

other command line options are: -min-depth, -max-depth - I have not used these

Another useful option is -depth-delta, which is the minimum number of consecutive iterations the program must keep the correct best move before it is accepted as a solution. The default value is 3.

Tord

[Uri Blass]

Hi Uri,

Glad you spotted the space issue - always annoying when that happens - one tends to look for bigger problems that are not there .

How is Rybka behaving during epd analysis - I have been using Polyglot for epd analysis - this works fine for me with Toga, Glaurung, Spike etc but does not work with Rybka - it does not seem to obey the time limit?

If you are having more luck with epd2wb - I will have to investigate

Thanks

Shaun

Yes

I have luck with epd2wb

Rybka solves 281 out of 300 of WAC in 1 second per position on my A3000.

failures:
2 71 80 92 100 141 157 163 196 222 229 230 235 265 272
275 287 293 297

For comparison
knightdreamer3.1 solve in the same time 294 out of 300

Failure 100 163 213 230 243 269
Uri

more results

at 1 second per position Fruit solve 290

Failure 2 92 100 141 163 183 222 230 235 243

I tried 10 seconds per position

results were

1.Knightdreamer3.1 298
Failure 100 230
2.Fruit2.2 297
Failure 38 123 230
3.Rybka1.2f 292
Failure 2 71 86 123 163 230 235 287

Hash that was used in all the tests were 32 mbytes for rybka and fruit and 16 mbytes for knightdreamer

[Uri Blass]

checking 123 show a bug in the test because both rybka and fruit likes Re6 that is also winning but not mentioned.

Re6 is included in wacnew but I used the wac that is in arasan site.

6k1/1b2rp2/1p4p1/3P4/PQ4P1/2N2q2/5P2/3R2K1 b - - bm Bxd5 Rc7; id "WAC.123";