Komodo 13 released

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

Moderators: hgm, Rebel, chrisw

User avatar
Ovyron
Posts: 4556
Joined: Tue Jul 03, 2007 4:30 am

Re: Komodo 13 released

Post by Ovyron »

Gabor Szots wrote: Tue May 07, 2019 8:32 amWhere do I set MultiPV under the Shredder GUI?
It seems that you can't set MutiPV under the Shredder GUI (it hides the setting in engine options, and doesn't allow to set it directly; I had to make my own compile of McCain that calls it "MultiPW" just to fix this.)

The best you can do is using the F6 Key to increase MutiPV lines by one, and the F5 Key to decrease them by one, but it's impossible to play MultiPV games in the Shredder GUI (as it defaults to 1 in play mode.)
Your beliefs create your reality, so be careful what you wish for.
mjlef
Posts: 1494
Joined: Thu Mar 30, 2006 2:08 pm

Re: Komodo 13 released

Post by mjlef »

Ovyron wrote: Wed May 08, 2019 10:44 pm
Uri Blass wrote: Wed May 08, 2019 11:45 am I believe that both Komodo and Komodo MCTS share the same code and it is not fair to have them as seperate engines.
This is a common thing for the CCRL, as Fritz 16 is +90% Rybka 4, and they have them as separate engines.
it is up to each person running tournaments to decide what rules they want for entries. We have never claimed all of K MCTS code is original code and when we introduced it we stated what we are doing. Although we use some common code, like move generation and evaluation of positions, there are a lot of changes. For example, move ordering is different, the moves pruned, reduced are different, the MCTS node logic is totally new, how nodes are selected for expansion is different, How the best move is found it totally different, the short search evaluation is transformed into a win probability or policy value. There are hundreds of differences. There are many chess engines that have used code from others, and I think this is fine as long as they have permission, and they clearly declare it so the person running the tournament can decide if the programs are different enough to be allowed to compete. And I assure, I gave myself permission! :-)
Gabor Szots
Posts: 1362
Joined: Sat Jul 21, 2018 7:43 am
Location: Szentendre, Hungary
Full name: Gabor Szots

Re: Komodo 13 released

Post by Gabor Szots »

Ovyron wrote: Wed May 08, 2019 10:48 pm
Gabor Szots wrote: Tue May 07, 2019 8:32 amWhere do I set MultiPV under the Shredder GUI?
It seems that you can't set MutiPV under the Shredder GUI (it hides the setting in engine options, and doesn't allow to set it directly; I had to make my own compile of McCain that calls it "MultiPW" just to fix this.)

The best you can do is using the F6 Key to increase MutiPV lines by one, and the F5 Key to decrease them by one, but it's impossible to play MultiPV games in the Shredder GUI (as it defaults to 1 in play mode.)
Thank you for your reply. I edited my Komodo eng file to include a line MultiPV=6, only I don't know whether it has an effect.
Gabor Szots
CCRL testing group
User avatar
Nordlandia
Posts: 2821
Joined: Fri Sep 25, 2015 9:38 pm
Location: Sortland, Norway

Re: Komodo 13 released

Post by Nordlandia »

Komodo 13 held this as black against latest brainfish. Impressive.

Opening - Danish Gambit - 1. e4 e5 2. d4 exd4 3. c3 dxc3 (3... d3) 4. Bc4 cxb2 5. Bxb2 d5 6. Bxd5 Nf6 7.
Nc3 Nxd5 8. Nxd5 c6 9. Nf6+ gxf6 10. Qxd8+ Kxd8 11. Bxf6+ Kc7 12. Bxh8 *

[pgn][Event "?"] [Site "i7-5960X 4.5GHz"] [Date "2019.05.09"] [Round "?"] [White "BrainFish_190505_x64_bmi2"] [Black "komodo-13.01-64bit-bmi2"] [Result "1/2-1/2"] [FEN "rnb2b1B/ppk2p1p/2p5/8/4P3/8/P4PPP/R3K1NR b KQ - 0 12"] [GameDuration "01:44:15"] [GameEndTime "2019-05-09T18:31:42.471 Vest-Europa (sommertid)"] [GameStartTime "2019-05-09T16:47:27.055 Vest-Europa (sommertid)"] [PlyCount "104"] [SetUp "1"] [Termination "adjudication"] [TimeControl "1800+30"] 1... Bb4+ {-1.15/34 142s} 2. Ke2 {+1.46/38 154s} b5 {-1.06/32 71s} 3. a3 {+1.61/32 118s} Bf8 {-0.93/32 72s} 4. Nh3 {+1.02/33 172s} Be6 {-0.92/34 121s} 5. Kf3 {+1.32/32 79s} Nd7 {-0.47/31 62s} 6. Ng5 {+1.46/34 35s} Bb3 {-0.86/34 76s} 7. Rhb1 {+1.46/35 41s} Bc4 {-0.97/36 79s} 8. Nxh7 {+1.55/36 53s} Bd6 {-1.07/34 95s} 9. Bc3 {+1.46/38 84s} Ne5+ {-1.05/35 60s} 10. Kf4 {+1.46/37 33s} Nd3+ {-1.00/36 108s} 11. Ke3 {+1.69/38 35s} f5 {-1.12/38 86s} 12. g3 {+1.68/37 68s} fxe4 {-1.10/38 114s} 13. Nf6 {+1.62/40 110s} Bc5+ {-1.16/36 83s} 14. Kxe4 {+1.59/40 55s} Nxf2+ {-1.03/36 54s} 15. Kf3 {+1.68/41 92s} Kb7 {-1.17/37 83s} 16. Kg2 {+1.83/38 122s} Rd8 {-1.12/38 134s} 17. Re1 {+1.84/36 22s} Rd3 {-1.17/35 54s} 18. Rac1 {+1.48/41 92s} Rd8 {-1.14/37 53s} 19. Ra1 {+1.92/41 141s} Rd3 {0.00/42 45s} 20. Rac1 {+1.49/41 68s} Rd8 {-1.16/37 99s} 21. Bb4 {+1.84/40 34s} Bd4 {-1.18/36 48s} 22. Re7+ {+1.93/37 46s} Ka6 {-1.15/39 64s} 23. Bc3 {+2.00/37 42s} Bxc3 {-1.20/36 55s} 24. Rxc3 {+2.06/37 73s} Rd2 {-1.17/36 58s} 25. Kf3 {+2.14/38 45s} Nd1 {-1.10/35 51s} 26. Rc1 {+2.24/36 36s} Rf2+ {-1.15/35 59s} 27. Kg4 {+2.24/35 33s} Nb2 {-1.16/37 50s} 28. Nd7 {+2.18/36 47s} Rxh2 {-1.18/35 80s} 29. Ne5 {+2.14/35 45s} Kb6 {-1.42/34 81s} 30. Kf5 {+2.13/33 62s} Rf2+ {-1.39/32 62s} 31. Kg6 {+2.15/34 38s} Rh2 {-1.49/32 63s} 32. Rc3 {+2.42/36 97s} a5 {-1.47/37 63s} 33. g4 {+2.00/40 65s} Kc5 {-1.47/37 34s} 34. g5 {+2.00/41 49s} b4 {-1.69/36 49s} 35. axb4+ {+2.00/43 34s} axb4 {-1.70/38 30s} 36. Rg3 {+2.33/45 159s} Bd5 {-1.71/38 35s} 37. Rb7 {+2.49/39 22s} Nc4 {-1.70/39 33s} 38. Nd7+ {+2.00/45 74s} Kd6 {-1.71/40 34s} 39. Nf6 {+2.00/49 42s} Ne5+ {-1.72/39 38s} 40. Kg7 {+2.00/50 38s} c5 {-1.74/40 37s} 41. Nxd5 {+2.00/51 34s} Kxd5 {-1.74/38 22s} 42. g6 {+2.00/50 31s} Kd4 {-1.74/40 86s} 43. Rb6 {+2.00/52 24s} Rh1 {-1.74/35 41s} 44. Rd6+ {+2.78/42 23s} Kc4 {-1.76/38 55s} 45. Kf6 {+2.78/48 28s} Nxg6 {-1.94/39 34s} 46. Kxg6 {+2.78/53 31s} Rc1 {-1.96/38 27s} 47. Kf5 {+2.78/57 24s} Rc3 {-1.97/41 34s} 48. Rg1 {+2.78/60 39s} Kb3 {-1.86/41 69s} 49. Ke4 {+2.78/62 35s} Kb2 {-1.88/42 38s} 50. Rdd1 {+2.78/62 29s} b3 {-1.90/42 30s} 51. Kd5 {+2.78/65 36s} Rc2 {-1.92/44 36s} 52. Ra1 {+2.78/67 26s} Rh2 {-1.94/43 28s} 53. Kxc5 {+2.78/68 23s, Draw by adjudication: SyzygyTB} 1/2-1/2 [/pgn]
mjlef
Posts: 1494
Joined: Thu Mar 30, 2006 2:08 pm

Re: Komodo 13 released

Post by mjlef »

Gabor Szots wrote: Thu May 09, 2019 9:58 am
Ovyron wrote: Wed May 08, 2019 10:48 pm
Gabor Szots wrote: Tue May 07, 2019 8:32 amWhere do I set MultiPV under the Shredder GUI?
It seems that you can't set MutiPV under the Shredder GUI (it hides the setting in engine options, and doesn't allow to set it directly; I had to make my own compile of McCain that calls it "MultiPW" just to fix this.)

The best you can do is using the F6 Key to increase MutiPV lines by one, and the F5 Key to decrease them by one, but it's impossible to play MultiPV games in the Shredder GUI (as it defaults to 1 in play mode.)
Thank you for your reply. I edited my Komodo eng file to include a line MultiPV=6, only I don't know whether it has an effect.
MultiPV is used by most people to see multiple lines. What they do not realize is the higher MultiPV is set, the worse the overall play since it takes longer in alpha-beta programs to find multiple bes moves. You have to basically do multiple searches excluding the the previously discovered lines, an that slows things down. In K MCTS, there is almost no cost to MultiPV, so you still get the best lines, but playing the best line is still as strong.

Different GUIs handle this differently in terms of what you can set in the engine in different modes.

Mark
User avatar
Laskos
Posts: 10948
Joined: Wed Jul 26, 2006 10:21 pm
Full name: Kai Laskos

Re: Komodo 13 released

Post by Laskos »

mjlef wrote: Thu May 09, 2019 8:02 pm
Gabor Szots wrote: Thu May 09, 2019 9:58 am
Ovyron wrote: Wed May 08, 2019 10:48 pm
Gabor Szots wrote: Tue May 07, 2019 8:32 amWhere do I set MultiPV under the Shredder GUI?
It seems that you can't set MutiPV under the Shredder GUI (it hides the setting in engine options, and doesn't allow to set it directly; I had to make my own compile of McCain that calls it "MultiPW" just to fix this.)

The best you can do is using the F6 Key to increase MutiPV lines by one, and the F5 Key to decrease them by one, but it's impossible to play MultiPV games in the Shredder GUI (as it defaults to 1 in play mode.)
Thank you for your reply. I edited my Komodo eng file to include a line MultiPV=6, only I don't know whether it has an effect.
MultiPV is used by most people to see multiple lines. What they do not realize is the higher MultiPV is set, the worse the overall play since it takes longer in alpha-beta programs to find multiple bes moves. You have to basically do multiple searches excluding the the previously discovered lines, an that slows things down. In K MCTS, there is almost no cost to MultiPV, so you still get the best lines, but playing the best line is still as strong.

Different GUIs handle this differently in terms of what you can set in the engine in different modes.

Mark
I don't totally understand. Isn't it just a "partial MultiPV"? The second, third etc. lines being progressively less explored? Am I wrong? And isn't this "partial MultiPV" doable with AB engines in a similar fashion using the regular search for the best move?
chrisw
Posts: 4313
Joined: Tue Apr 03, 2012 4:28 pm

Re: Komodo 13 released

Post by chrisw »

Laskos wrote: Thu May 09, 2019 9:08 pm
mjlef wrote: Thu May 09, 2019 8:02 pm
Gabor Szots wrote: Thu May 09, 2019 9:58 am
Ovyron wrote: Wed May 08, 2019 10:48 pm
Gabor Szots wrote: Tue May 07, 2019 8:32 amWhere do I set MultiPV under the Shredder GUI?
It seems that you can't set MutiPV under the Shredder GUI (it hides the setting in engine options, and doesn't allow to set it directly; I had to make my own compile of McCain that calls it "MultiPW" just to fix this.)

The best you can do is using the F6 Key to increase MutiPV lines by one, and the F5 Key to decrease them by one, but it's impossible to play MultiPV games in the Shredder GUI (as it defaults to 1 in play mode.)
Thank you for your reply. I edited my Komodo eng file to include a line MultiPV=6, only I don't know whether it has an effect.
MultiPV is used by most people to see multiple lines. What they do not realize is the higher MultiPV is set, the worse the overall play since it takes longer in alpha-beta programs to find multiple bes moves. You have to basically do multiple searches excluding the the previously discovered lines, an that slows things down. In K MCTS, there is almost no cost to MultiPV, so you still get the best lines, but playing the best line is still as strong.

Different GUIs handle this differently in terms of what you can set in the engine in different modes.

Mark
I don't totally understand. Isn't it just a "partial MultiPV"? The second, third etc. lines being progressively less explored?
Yes, and by definition, the line treated as “first” is the one that has been most explored.
Am I wrong? And isn't this "partial MultiPV" doable with AB engines in a similar fashion using the regular search for the best move?
No, because the criteria for pruning in AB and the criteria for exploration in MCTS PUCT are radically different. AB fundamentally prunes (well, anti-explores!) on the values alpha beta which are evaluation type things, and got from the PV search, backed up to the root. PUCT explores (anti-prunes) based on winning lines and visit count (which are win rate and popularity type things) backed up from the line/move it is considering.

AB is “efficient” because it cuts away much of the tree on the basis that you only need to know the best move, not to rank all moves into order (which would involve, in minimax, putting equal effort into all root moves). MCTS does rank all moves into order, but it prunes away much of the tree by a different process, a kind of incrementally calculated popularity.
rcmaddox
Posts: 181
Joined: Mon Oct 14, 2013 8:59 pm
Location: Winder, GA
Full name: Robert C. Maddox

Re: Komodo 13 released

Post by rcmaddox »

lkaufman wrote: Tue May 07, 2019 1:05 am
<snip>

While the default mode is still the strongest one for engine vs engine play, the MCTS mode is now stronger whenever MultiPV is set to 2 or more. Since most people use MultiPV for analysis and since the MCTS mode tends to produce more "human" moves because it does not assume that the opponent will play perfectly (by its criteria), we feel that the MCTS mode is now the more useful one for the majority of users, except perhaps for correspondence players. According to my tests, Komodo 13.01 MCTS can now defeat any CPU based engine, including recent Stockfish, if both are set to use MultiPV of 6 or more. This is because Komodo MCTS pays no performance penalty for obtaining this information, while normal alpha-beta engines pay a very heavy price to give analysis for several moves. For users who don't have a good GPU and who use high values of MultiPV, Komodo 13 MCTS should be the strongest engine available regardless of price.

<snip>
This makes Komodo 13.01 MCTS the preferred engine for the Tarrasch GUI. Why? Because Tarrasch is set to MultiPV 4 and you can't turn it off.
lkaufman
Posts: 5960
Joined: Sun Jan 10, 2010 6:15 am
Location: Maryland USA

Re: Komodo 13 released

Post by lkaufman »

Gabor Szots wrote: Thu May 09, 2019 9:58 am
Ovyron wrote: Wed May 08, 2019 10:48 pm
Gabor Szots wrote: Tue May 07, 2019 8:32 amWhere do I set MultiPV under the Shredder GUI?
It seems that you can't set MutiPV under the Shredder GUI (it hides the setting in engine options, and doesn't allow to set it directly; I had to make my own compile of McCain that calls it "MultiPW" just to fix this.)

The best you can do is using the F6 Key to increase MutiPV lines by one, and the F5 Key to decrease them by one, but it's impossible to play MultiPV games in the Shredder GUI (as it defaults to 1 in play mode.)
Thank you for your reply. I edited my Komodo eng file to include a line MultiPV=6, only I don't know whether it has an effect.
It should only affect the display, not the actual move choice. In theory it will be a little slower since it has to do some extra work to display the extra info, but I think that the slowdown is a tiny fraction of one percent so just a tiny fraction of an elo point. But if you set Alpha-Beta engines to MultiPV=6, they will lose something like 200 elo points at the faster time controls.
Komodo rules!
lkaufman
Posts: 5960
Joined: Sun Jan 10, 2010 6:15 am
Location: Maryland USA

Re: Komodo 13 released

Post by lkaufman »

rcmaddox wrote: Thu May 09, 2019 11:05 pm
lkaufman wrote: Tue May 07, 2019 1:05 am
<snip>

While the default mode is still the strongest one for engine vs engine play, the MCTS mode is now stronger whenever MultiPV is set to 2 or more. Since most people use MultiPV for analysis and since the MCTS mode tends to produce more "human" moves because it does not assume that the opponent will play perfectly (by its criteria), we feel that the MCTS mode is now the more useful one for the majority of users, except perhaps for correspondence players. According to my tests, Komodo 13.01 MCTS can now defeat any CPU based engine, including recent Stockfish, if both are set to use MultiPV of 6 or more. This is because Komodo MCTS pays no performance penalty for obtaining this information, while normal alpha-beta engines pay a very heavy price to give analysis for several moves. For users who don't have a good GPU and who use high values of MultiPV, Komodo 13 MCTS should be the strongest engine available regardless of price.

<snip>
This makes Komodo 13.01 MCTS the preferred engine for the Tarrasch GUI. Why? Because Tarrasch is set to MultiPV 4 and you can't turn it off.
That's interesting! It means that any engine vs. engine tests that have been run using this GUI are totally invalid; I wonder if any of the testers are making this mistake without realizing it? They are obviously not using it to test Komodo MCTS since if they were the ratings would be 150 or more elo higher than they are.
Komodo rules!