Rodent III 0.273 - official release

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

Moderators: hgm, Rebel, chrisw

PK
Posts: 893
Joined: Mon Jan 15, 2007 11:23 am
Location: Warsza

Rodent III 0.273 - official release

Post by PK »

Is available at http://www.pkoziol.cal24.pl/rodent/download.htm

With AppVeyor builds, the notion of official release is a bit blurry, so I guess I should step in to make the testers busy.

This version builds on TCEC experience, which has shown that Rodent needs better multithreaded code. The current edition can use up to 48 threads. I'm not sure if max hash size would be enough for that, so this is the next task, but testing on my limited resources has shown a nice gain from 6 threads onwards.

Beside that, there were some Texel tuning runs, some search improvements and basic multi-pv code (one thread only) has been added.

Please enjoy!
Dokterchen
Posts: 133
Joined: Wed Aug 15, 2007 12:18 pm
Location: Munich

Re: Rodent III 0.273 - official release

Post by Dokterchen »

Good news! Thanks Pawel
User avatar
Graham Banks
Posts: 41428
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Re: Rodent III 0.273 - official release

Post by Graham Banks »

Thanks Pawel.
What is your estimated strength improvement?
gbanksnz at gmail.com
PK
Posts: 893
Joined: Mon Jan 15, 2007 11:23 am
Location: Warsza

Re: Rodent III 0.273 - official release

Post by PK »

I definately expect crossing 3000 CCRL on a single core, perhaps getting even as far as 3025. I'm not sure whether multi-core fixes will be meaningful on 4 threads.
User avatar
Graham Banks
Posts: 41428
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Re: Rodent III 0.273 - official release

Post by Graham Banks »

PK wrote: Sat Sep 01, 2018 10:57 am I definately expect crossing 3000 CCRL on a single core, perhaps getting even as far as 3025. I'm not sure whether multi-core fixes will be meaningful on 4 threads.
Nice! :)
gbanksnz at gmail.com
BrendanJNorman
Posts: 2526
Joined: Mon Feb 08, 2016 12:43 am
Full name: Brendan J Norman

Re: Rodent III 0.273 - official release

Post by BrendanJNorman »

PK wrote: Sat Sep 01, 2018 10:03 am Is available at http://www.pkoziol.cal24.pl/rodent/download.htm

With AppVeyor builds, the notion of official release is a bit blurry, so I guess I should step in to make the testers busy.

This version builds on TCEC experience, which has shown that Rodent needs better multithreaded code. The current edition can use up to 48 threads. I'm not sure if max hash size would be enough for that, so this is the next task, but testing on my limited resources has shown a nice gain from 6 threads onwards.

Beside that, there were some Texel tuning runs, some search improvements and basic multi-pv code (one thread only) has been added.

Please enjoy!
My favorite chess engine (by far) has a new version!

Is it Christmas already? :D

Thanks Pawel, I'll begin trying all of my personalities with this new version now. 8-)
tpoppins
Posts: 919
Joined: Tue Nov 24, 2015 9:11 pm
Location: upstate

Re: Rodent III 0.273 - official release

Post by tpoppins »

I got four time forfeits with v0.273 out of 40 games vs. v0.238. Never seen Rodent lose on time before.

[pgn][Event "Rodent III 0.273 64-bit 40/40"] [Site "Dual E5-2690v2"] [Date "2018.09.03"] [Round "10"] [White "Rodent III 0.238 64-bit"] [Black "Rodent III 0.273 64-bit"] [Result "1-0"] [ECO "A20"] [GameDuration "00:40:04"] [GameEndTime "2018-09-03T12:05:08.579 Eastern Daylight Time"] [GameStartTime "2018-09-03T11:25:04.027 Eastern Daylight Time"] [Opening "English Opening"] [PlyCount "79"] [Termination "time forfeit"] [TimeControl "40/1200"] 1. c4 {book} e5 {book} 2. g3 {book} c5 {book} 3. Nc3 {book} Nc6 {book} 4. Bg2 {book} g6 {book} 5. Nf3 {book} Bg7 {book} 6. d3 {book} d6 {book} 7. h4 {book} h6 {book} 8. Nh2 {book} Nf6 {book} 9. Rb1 {book} Bf5 {book} 10. Nf1 {book} O-O {book} 11. Ne3 {+0.33/23 40s} Bd7 {-0.24/23 40s} 12. Bd2 {+0.38/22 40s} a5 {-0.35/23 40s} 13. Ned5 {+0.28/23 40s} Nxd5 {-0.24/23 40s} 14. Bxd5 {+0.18/24 40s} Bg4 {-0.24/22 40s} 15. f3 {+0.40/23 40s} Bf5 {-0.24/23 40s} 16. g4 {+0.24/22 40s} Bc8 {-0.34/25 40s} 17. Qc1 {+0.26/20 40s} Kh7 {-0.27/23 40s} 18. a3 {+0.25/21 40s} Nd4 {-0.34/22 40s} 19. h5 {+0.58/24 40s} g5 {-0.33/25 40s} 20. b4 {+0.66/23 40s} cxb4 {-0.54/23 40s} 21. axb4 {+0.74/23 40s} a4 {-0.43/23 40s} 22. e3 {+0.78/25 40s} Nc6 {-0.76/25 40s} 23. b5 {+1.01/26 40s} Ne7 {-0.76/27 40s} 24. e4 {+1.17/25 40s} Nxd5 {-0.87/23 40s} 25. Nxd5 {+0.96/26 40s} Be6 {-0.87/24 40s} 26. Nc3 {+1.03/23 40s} Qd7 {-0.96/22 40s} 27. Ra1 {+1.08/22 40s} f5 {-0.97/23 40s} 28. gxf5 {+0.96/23 40s} Bxf5 {-1.10/23 40s} 29. exf5 {+1.17/24 40s} Qxf5 {-0.92/25 40s} 30. O-O {+1.09/24 40s} Qxd3 {-1.37/22 40s} 31. Qb1 {+1.32/25 40s} Qxb1 {-1.33/26 40s} 32. Raxb1 {+1.37/26 40s} Rfc8 {-1.22/25 40s} 33. Nd5 {+1.39/26 40s} e4 {-1.24/26 40s} 34. fxe4 {+1.50/26 40s} Rxc4 {-1.44/26 40s} 35. Rf7 {+1.95/24 40s} Kh8 {-1.69/25 40s} 36. Rxb7 {+1.62/26 40s} a3 {-1.89/25 40s} 37. Rxg7 {+1.98/28 40s} a2 {-2.09/28 80s} 38. Ra1 {+1.96/29 40s} Kxg7 {-2.07/28 53s} 39. Bc3+ {+2.04/26 40s} Kf7 {-1.89/26 13s} 40. Rf1+ {+1.82/25 39s, Black loses on time} 1-0 [Event "Rodent III 0.273 64-bit 40/40"] [Site "Dual E5-2690v2"] [Date "2018.09.03"] [Round "11"] [White "Rodent III 0.273 64-bit"] [Black "Rodent III 0.238 64-bit"] [Result "0-1"] [ECO "A10"] [GameDuration "00:39:30"] [GameEndTime "2018-09-03T12:21:18.416 Eastern Daylight Time"] [GameStartTime "2018-09-03T11:41:47.812 Eastern Daylight Time"] [Opening "English Opening"] [PlyCount "78"] [Termination "time forfeit"] [TimeControl "40/1200"] 1. c4 {book} g6 {book} 2. g3 {book} Bg7 {book} 3. Bg2 {book} e5 {book} 4. Nc3 {book} d6 {book} 5. d3 {book} Nd7 {book} 6. e4 {book} Ne7 {book} 7. Be3 {+0.21/22 35s} O-O {-0.08/21 35s} 8. Qd2 {+0.21/22 35s} f5 {-0.17/21 35s} 9. Nge2 {+0.12/21 35s} c6 {-0.24/22 35s} 10. O-O {+0.21/22 35s} Nf6 {-0.22/20 35s} 11. Bh6 {+0.17/21 35s} Be6 {-0.07/19 35s} 12. b3 {+0.19/22 35s} Bxh6 {-0.18/19 35s} 13. Qxh6 {+0.20/20 35s} Qa5 {-0.06/19 35s} 14. a4 {+0.29/18 35s} Rae8 {-0.06/20 35s} 15. Qd2 {+0.24/19 35s} fxe4 {+0.08/20 35s} 16. dxe4 {+0.39/21 35s} Rd8 {0.00/20 35s} 17. Qb2 {+0.38/21 35s} Bg4 {-0.15/22 35s} 18. h3 {+0.59/23 35s} Be6 {-0.27/22 35s} 19. Rfd1 {+0.50/20 35s} g5 {-0.32/21 35s} 20. Qa3 {+0.51/22 35s} Nc8 {-0.12/21 35s} 21. Rf1 {+0.53/22 35s} g4 {-0.09/21 35s} 22. h4 {+0.46/21 35s} Rd7 {-0.05/22 35s} 23. Nd1 {+0.58/20 35s} d5 {-0.23/23 35s} 24. b4 {+0.55/21 35s} Qd8 {-0.32/20 35s} 25. exd5 {+0.58/22 35s} cxd5 {-0.49/23 35s} 26. cxd5 {+0.66/23 35s} Nxd5 {-0.33/23 35s} 27. Nec3 {+0.71/21 35s} Nxc3 {-0.38/21 35s} 28. Qxc3 {+0.79/22 35s} Rd4 {-0.50/22 35s} 29. Ne3 {+0.93/23 35s} Qd6 {-0.50/23 35s} 30. b5 {+1.03/22 35s} Nb6 {-0.70/20 35s} 31. Qa5 {+1.02/24 35s} Nc4 {-0.85/22 35s} 32. Qxa7 {+1.08/23 35s} Nd2 {-0.83/23 35s} 33. Rfe1 {+1.24/22 35s} Nb3 {-0.87/23 35s} 34. Rad1 {+1.46/22 35s} b6 {-0.88/24 35s} 35. h5 {+1.64/22 35s} Rd8 {-1.25/20 35s} 36. Bc6 {+2.25/23 35s} Nd2 {-1.58/21 35s} 37. h6 {+3.45/24 35s} Qf8 {-3.20/23 35s} 38. Nf1 {+3.73/24 35s} Nb3 {-3.57/23 35s} 39. Rxd4 {+4.26/23 35s} Nxd4 {-3.65/24 35s, White loses on time} 0-1 [Event "Rodent III 0.273 64-bit 40/40"] [Site "Dual E5-2690v2"] [Date "2018.09.03"] [Round "18"] [White "Rodent III 0.238 64-bit"] [Black "Rodent III 0.273 64-bit"] [Result "1-0"] [ECO "A00"] [GameDuration "00:40:04"] [GameEndTime "2018-09-03T13:13:51.442 Eastern Daylight Time"] [GameStartTime "2018-09-03T12:33:47.005 Eastern Daylight Time"] [Opening "Van't Kruijs Opening"] [PlyCount "79"] [Termination "time forfeit"] [TimeControl "40/1200"] 1. e3 {book} Nf6 {book} 2. Nf3 {book} g6 {book} 3. b3 {book} Bg7 {book} 4. Bb2 {book} d6 {book} 5. d4 {book} Nbd7 {book} 6. Be2 {book} O-O {book} 7. c4 {book} e5 {book} 8. O-O {book} e4 {book} 9. Nfd2 {+0.13/23 37s} Qe7 {-0.23/21 38s} 10. Nc3 {+0.09/20 37s} Re8 {-0.24/23 37s} 11. Re1 {+0.13/21 38s} c6 {-0.21/23 37s} 12. Qc2 {0.00/19 38s} a6 {-0.09/21 37s} 13. Rac1 {0.00/20 37s} b6 {-0.10/22 37s} 14. a4 {+0.10/21 38s} d5 {-0.14/22 37s} 15. f3 {0.00/20 37s} Bh6 {+0.17/22 38s} 16. f4 {+0.37/21 37s} Bb7 {+0.02/22 38s} 17. g4 {+0.12/23 37s} g5 {+0.29/21 37s} 18. h4 {0.00/22 38s} gxf4 {+0.93/20 37s} 19. g5 {-0.54/20 37s} Qd6 {+0.82/22 75s} 20. gxh6 {-0.83/20 37s} Kh8 {+1.56/21 36s} 21. Kh1 {-1.01/19 38s} Rg8 {+1.84/23 71s} 22. Ndxe4 {-1.01/21 37s} Nxe4 {+1.10/24 34s} 23. Nxe4 {-1.53/24 38s} Qxh6 {+1.65/23 34s} 24. Bf3 {-2.85/24 37s} dxe4 {+3.28/23 34s} 25. Bxe4 {-2.84/25 38s} fxe3 {+3.40/26 68s} 26. Bf5 {-3.11/25 37s} Nf6 {+2.97/23 32s} 27. Rg1 {-3.14/23 37s} Rae8 {+3.57/26 63s} 28. Qh2 {-3.10/24 37s} e2 {+3.75/25 58s} 29. Qf2 {-3.61/21 38s} Nh5 {+4.08/24 53s} 30. Be6 {-3.88/22 38s} Rxe6 {+3.86/20 24s} 31. d5+ {-3.69/24 37s} Rf6 {+3.93/23 24s} 32. Kh2 {-3.99/22 37s} cxd5 {+4.36/22 24s} 33. cxd5 {-4.21/23 38s} Bxd5 {+4.53/23 24s} 34. a5 {-4.46/22 37s} bxa5 {+4.94/21 24s} 35. Be5 {-4.34/22 38s} Rxg1 {+5.29/24 24s} 36. Rxg1 {-5.35/27 38s} Qg7 {+5.85/29 24s} 37. Bxf6 {-6.33/30 37s} Qxf6 {+6.34/32 49s} 38. Qxf6+ {-6.82/31 38s} Nxf6 {+6.76/34 32s} 39. Re1 {-7.08/34 37s} Bxb3 {+6.77/32 8.1s} 40. Rxe2 {-7.31/33 36s, Black loses on time} 1-0 [Event "Rodent III 0.273 64-bit 40/40"] [Site "Dual E5-2690v2"] [Date "2018.09.03"] [Round "19"] [White "Rodent III 0.273 64-bit"] [Black "Rodent III 0.238 64-bit"] [Result "0-1"] [ECO "B23"] [GameDuration "00:39:30"] [GameEndTime "2018-09-03T13:16:22.637 Eastern Daylight Time"] [GameStartTime "2018-09-03T12:36:52.012 Eastern Daylight Time"] [Opening "Sicilian"] [PlyCount "78"] [Termination "time forfeit"] [TimeControl "40/1200"] [Variation "Closed, 2...Nc6"] 1. e4 {book} c5 {book} 2. Nc3 {book} Nc6 {book} 3. Bb5 {book} Nf6 {book} 4. Bxc6 {book} dxc6 {book} 5. d3 {book} e5 {book} 6. f4 {book} exf4 {book} 7. Bxf4 {-0.16/22 35s} c4 {+0.10/20 35s} 8. Nge2 {-0.13/21 35s} cxd3 {+0.11/22 35s} 9. Qxd3 {-0.19/22 35s} Qxd3 {+0.10/21 35s} 10. cxd3 {-0.13/22 35s} Be6 {+0.08/23 35s} 11. Rd1 {-0.07/23 35s} Bb4 {+0.08/23 35s} 12. O-O {-0.19/22 35s} O-O {0.00/24 35s} 13. h3 {-0.16/23 35s} a5 {+0.24/23 35s} 14. Be3 {-0.24/24 35s} a4 {+0.16/26 35s} 15. a3 {-0.17/25 35s} Ba5 {+0.26/26 35s} 16. Nd4 {-0.24/26 35s} Bb6 {+0.24/25 35s} 17. Nce2 {-0.30/25 35s} Nd7 {+0.12/23 35s} 18. Bf2 {-0.29/25 35s} Bxd4 {+0.28/25 35s} 19. Nxd4 {-0.19/26 35s} Rfe8 {+0.20/27 35s} 20. Rc1 {-0.12/25 35s} Ne5 {+0.12/23 35s} 21. Rfd1 {-0.16/24 35s} f6 {+0.20/25 35s} 22. Nxe6 {-0.21/26 35s} Rxe6 {+0.19/23 35s} 23. d4 {-0.05/25 35s} Ng6 {+0.11/27 35s} 24. d5 {-0.01/28 35s} Rxe4 {+0.08/27 35s} 25. dxc6 {0.00/27 35s} bxc6 {0.00/27 35s} 26. Rxc6 {-0.13/27 35s} Ne5 {+0.08/27 35s} 27. Rc7 {0.00/30 35s} Rc4 {0.00/28 35s} 28. Rb7 {0.00/30 35s} Rac8 {0.00/31 35s} 29. Rd6 {0.00/28 35s} Rc1+ {0.00/26 35s} 30. Kh2 {0.00/29 35s} R8c7 {0.00/30 35s} 31. Rxc7 {0.00/28 35s} Rxc7 {0.00/30 35s} 32. Bd4 {0.00/30 35s} Rc2 {0.00/30 35s} 33. Kg3 {0.00/28 35s} Nd3 {0.00/27 35s} 34. Bxf6 {0.00/28 35s} gxf6 {0.00/31 35s} 35. Rxd3 {0.00/31 35s} Rxb2 {0.00/32 35s} 36. Kf3 {0.00/30 35s} Kf7 {0.00/31 35s} 37. g4 {0.00/30 35s} Rb3 {0.00/34 35s} 38. Ke4 {0.00/33 35s} Rxd3 {0.00/38 35s} 39. Kxd3 {0.00/39 35s} Kg6 {0.00/40 35s, White loses on time} 0-1 [/pgn]

The score is

Code: Select all

Score of Rodent III 0.273 64-bit vs Rodent III 0.238 64-bit: 3 - 10 - 27 [0.412]
Elo difference: -61.43 +/- 60.85

40 of 40 games finished.
Tirsa Poppins
CCRL
MikeGL
Posts: 1010
Joined: Thu Sep 01, 2011 2:49 pm

Re: Rodent III 0.273 - official release

Post by MikeGL »

Thanks for this new release Pawel!

PK wrote: Sat Sep 01, 2018 10:03 am Is available at http://www.pkoziol.cal24.pl/rodent/download.htm

With AppVeyor builds, the notion of official release is a bit blurry, so I guess I should step in to make the testers busy.

This version builds on TCEC experience, which has shown that Rodent needs better multithreaded code. The current edition can use up to 48 threads. I'm not sure if max hash size would be enough for that, so this is the next task, but testing on my limited resources has shown a nice gain from 6 threads onwards.

Beside that, there were some Texel tuning runs, some search improvements and basic multi-pv code (one thread only) has been added.

Please enjoy!
I told my wife that a husband is like a fine wine; he gets better with age. The next day, she locked me in the cellar.
fastgm
Posts: 818
Joined: Mon Aug 19, 2013 6:57 pm

Re: Rodent III 0.273 - official release

Post by fastgm »

tpoppins wrote: Mon Sep 03, 2018 8:35 pm I got four time forfeits with v0.273 out of 40 games vs. v0.238. Never seen Rodent lose on time before.
option name TimeBuffer type spin default 10 min 0 max 1000

I got as well time forfeits and increased this value to 100.
The time forfeits are still there, but much less. My time control is 60 sec + 0.6 sec.

Andreas
tpoppins
Posts: 919
Joined: Tue Nov 24, 2015 9:11 pm
Location: upstate

Re: Rodent III 0.273 - official release

Post by tpoppins »

Thank you, Andreas, I'll follow your recommendation.
The odd thing is I'd run thousands of games with previous versions, starting with v0.210, at default settings and never had a forfeit.

After replaying those four games the score is

Code: Select all

Rank Name                      Elo    +    - games score oppo. draws
   1 Rodent III 0.238 64-bit    20   23   25    40   56%   -20   73%
   2 Rodent III 0.273 64-bit   -20   25   23    40   44%    20   73%
Tirsa Poppins
CCRL