Can Komodo 13. for Android beat Shredder 8 on a P.C ?

[Chessqueen]

Threads 8
Ponder Not Checked
Table memory 64
OwnBook Not checked
Book Moves 1000
Best Book Lines Checked
Log
Use LMR checked
Null Move Pruning Checked
Overhead ms 50
Use Syzygy Unchecked
Syzygy 50 Move Rules checked
King Safety 84
Selectivity 153
Reduction 40
Dynamism 124
Comtempt 16
White Contempt unchecked
Use MCTS unchecked
MCTS Hash 64
Progress Threshold 30
Skill 25

Thanks!

The "Use MCTS" option seems missing from your list. Is that enabled? I had only compared play to regular K13. Perhaps you were using K MCTS?

All the other options I did NOT listed since they are defaults like Variety etc....which is set to 0

Well since Nobody wants to reveal their settings to see if it is any different then my settings I will post this game here : I repeated the same opening played by both up to move 6 and see if Komodo 13 learned from its mistake in the previous games. The game continue the same but I stopped it in move 31 since the result was going to be identical mate in 2 more moves. Probably if somebody here has a different settings that can play stronger for Komodo 13 they should post it here or simply modify my settings. My system is about 4 years old AMD E-2 7110 APU with Radeon R2 Graphics 1.80 GHz installed Memory (RAM) 8.00 GB which is about 3 to 4 times faster then when Shredder 8 was tested back in 2004 with an Athlon 1200 MHz


[pgn][Event "Chess for Android Game"]
[Site "United States"]
[Date "2020.04.23"]
[Round "?"]
[White "Komodo 13"]
[WhiteElo "-"]
[Black "Shredder 8 on my P.C."]
[BlackElo "-"]
[Result "0-1"]
[PlyCount "64"]
[TimeControl "900+10"]
[Device "SM-G955U"]

1. d4 {opening reply} d5
2. e4 {opening reply} e6
3. exd5 {[8/14]0} exd5
4. Be2 {[8/14]-2} Nf6
5. Nf3 {[8/14]0} Bd6
6. O-O {[8/14]1} O-O
7. Bg5 {[8/13]0} Bg4
8. c4 {[9/14]13} dxc4
9. Bxc4 {[8/14]-7} Nc6
10. d5 {[8/14]-94} Ne5
11. Be2 {[8/14]-5} Re8
12. Nbd2 {[8/14]3} Ng6
13. h4 {[8/13]-98} h6
14. Bxf6 {[9/14]-76} Qxf6
15. Bb5 {[8/14]-98} Rec8
16. Qa4 {[8/14]-96} Nf4
17. Qb3 {[7/14]-5} Qg6
18. Ne1 {[7/12]-87} a6
19. Bc4 {[8/13]-97} Re8
20. Qxb7 {[8/14]-200} Bd7
21. Qb3 {[8/14]-296} Rxe1
22. g3 {[8/14]-308} Ne2+
23. Bxe2 {[8/14]-309} Rxe2
24. Nc4 {[8/14]-330} Bh3
25. Nxd6 {[8/14]-319} cxd6
26. Qb7 {[8/14]-416} Qe4
27. Qxa8+ {[9/14]-493} Kh7
28. f3 {black mates in #5} Qe3+
29. Kh1 {black mates in #4} Bg2+
30. Kh2 {[1/3]-204} Bxf1+
31.[/pgn]

[MonteCarlo]

Ok, those depth outputs are not like anything K13 displays on Android.

With the MCTS version, you'll usually see something like 7/0, 8/1, etc., while with the non-MCTS, you'll see 15/15, 16/16, etc.

The depths you're showing are just way too low to be the AB K13 on a high-end Android, and they're the wrong relative numbers to be K13 MCTS (or AB, for that matter) on a high-end Android.

Those depths in your PGN, however, DO look a lot like what is output by the engine that comes with Chess for Android, though.

Some of those moves, too, are just absolute howlers that K13 in either AB or MCTS mode rejects instantly even on my slow Moto G6.

It now seems most likely that you're not actually running K13 in these tests at all.

Cheers!

[Chessqueen]

Ok, those depth outputs are not like anything K13 displays on Android.

With the MCTS version, you'll usually see something like 7/0, 8/1, etc., while with the non-MCTS, you'll see 15/15, 16/16, etc.

The depths you're showing are just way too low to be the AB K13 on a high-end Android, and they're the wrong relative numbers to be K13 MCTS (or AB, for that matter) on a high-end Android.

Those depths in your PGN, however, DO look a lot like what is output by the engine that comes with Chess for Android, though.

Some of those moves, too, are just absolute howlers that K13 in either AB or MCTS mode rejects instantly even on my slow Moto G6.

It now seems most likely that you're not actually running K13 in these tests at all.

Cheers!

I am using the Chess for Android GUI, but Now I configure Komodo 13 with all the maximum Settings and I am getting a dept of 8//14, 9/14 with the Komodo regular Non MCTC. What GUI do you recommend me to use?

[MonteCarlo]

Chess for Android is what I tested with to confirm that the depths in your PGN didn't line up with K13's, so Chess for Android works for me.

You're definitely long pressing, going to "UCI and XBoard"->"Import engine" and selecting Komodo before playing?

It should show "komodo13 :ready" in the bottom-left of the notation area when that's done.

Cheers!

[Chessqueen]

Chess for Android is what I tested with to confirm that the depths in your PGN didn't line up with K13's, so Chess for Android works for me.

You're definitely long pressing, going to "UCI and XBoard"->"Import engine" and selecting Komodo before playing?

It should show "komodo13 :ready" in the bottom-left of the notation area when that's done.

Cheers!

Thanks, Now I am getting twice the Depth

[MonteCarlo]

Glad to help!

That means you were indeed using Chess for Android's built-in java engine before.

Games against Shredder 8 should be more interesting now

[Chessqueen]

Glad to help!

That means you were indeed using Chess for Android's built-in java engine before.

Games against Shredder 8 should be more interesting now

Well before I was just clicking on the Komodo 13 icon from my Samsung Galaxy S8+, then it took me to a screen saying that i could select from 5 or 6 different GU'sI and I always selected the first one Chess for android, but without going first to Chess for Android and selecting UCI and then import Komodo 13 like to said . Now I am getting close to the Depth that you stated previously. Tomorrow I will have another match game in 15 plus 10s it should be interesting.

PS: But remember my Dell system is about 4 times faster than Athlon 256 1200 MHz that Shredder 8 was tested back in 2004

[MonteCarlo]

Sounds good.

Just so I have some idea, what kn/s are you getting?

I get about 1200 kn/s from AB K13 when using all 8 of my phone's cores.

I'd expect yours to be substantially faster.

Cheers!

[Chessqueen]

Sounds good.

Just so I have some idea, what kn/s are you getting?

I get about 1200 kn/s from AB K13 when using all 8 of my phone's cores.

I'd expect yours to be substantially faster.

Cheers!

I saw it reaching 4830 kn/s, but here is the Result. I forced them to start with 1.h3 The Clemenz Opening which NoT too many programs knows the exact lines or variations Shredder 8 took White and Komodo 13 Black.



[Event "Chess for Android Game"]
[Site "United States"]
[Date "2020.04.23"]
[Round "?"]
[White "Shredder 8"]
[WhiteElo "-"]
[Black "Komodo 13.1 64-bit (terminated)"]
[BlackElo "-"]
[Result "1-0"]
[PlyCount "67"]
[TimeControl "900+10"]
[Device "SM-G955U"]

1. h3 e5 {opening reply}
2. e4 Nf6 {[15/4]24/b1c3}
3. Nc3 Bb4 {[16/4]19/g1f3}
4. Nge2 O-O {[17/6]20/g2g3}
5. a3 Bc5 {[14/4]26/b2b3}
6. b4 Bb6 {[14/4]43/g2g3}
7. Bb2 a5 {[16/4]65/g2g4}
8. b5 c6 {[15/4]67/e2g3}
9. Ng3 a4 {[16/3]73/g3e2}
10. bxc6 dxc6 {[17/5]65/d1c1}
11. Qf3 Be6 {[17/3]82/c3a2}
12. Be2 g6 {[17/3]95/c3b1}
13. Nd1 Nbd7 {[25]121/a1b1}
14. Ne3 h5 {[24]131/e2d3}
15. O-O Nh7 {[8/14]30}
16. Bc4 Ng5 {[9/14]119}
17. Qe2 Bxe3 {[9/14]118}
18. fxe3 h4 {[10/14]28}
19. Nh1 Nxe4 {[10/14]24}
20. Bxe6 fxe6 {[10/14]12}
21. Qg4 Ng5 {[10/14]14}
22. Rxf8+ Kxf8 {[9/14]37}
23. Rf1+ Kg8 {[8/14]24}
24. Nf2 Kg7 {[8/14]4}
25. Nd3 Ra5 {[8/14]-80}
26. Qf4 Rb5 {[31]-526/b2c3}
27. Bc3 Qe7 {[34]-545/d3e5}
28. Nxe5 Nxe5 {[36]-545/c3e5}
29. Bxe5+ Kg8 {[37]-548/e5f6}
30. Bf6 Qf8 {[36]-574/f4h4}
31. Bxg5 Qxf4 {[31]-484/g5f4}
32. exf4 Rb2 {[32]-495/f1c1}
33. c4 b5 {[31]-460/f1c1}
34. Bxh4 1-0
{black resigned:engine terminates}


[pgn][Event "Chess for Android Game"]
[Site "United States"]
[Date "2020.04.23"]
[Round "?"]
[White "Shredder 8"]
[WhiteElo "-"]
[Black "Komodo 13.1 64-bit (terminated)"]
[BlackElo "-"]
[Result "1-0"]
[PlyCount "67"]
[TimeControl "900+10"]
[Device "SM-G955U"]

1. h3 e5 {opening reply}
2. e4 Nf6 {[15/4]24/b1c3}
3. Nc3 Bb4 {[16/4]19/g1f3}
4. Nge2 O-O {[17/6]20/g2g3}
5. a3 Bc5 {[14/4]26/b2b3}
6. b4 Bb6 {[14/4]43/g2g3}
7. Bb2 a5 {[16/4]65/g2g4}
8. b5 c6 {[15/4]67/e2g3}
9. Ng3 a4 {[16/3]73/g3e2}
10. bxc6 dxc6 {[17/5]65/d1c1}
11. Qf3 Be6 {[17/3]82/c3a2}
12. Be2 g6 {[17/3]95/c3b1}
13. Nd1 Nbd7 {[25]121/a1b1}
14. Ne3 h5 {[24]131/e2d3}
15. O-O Nh7 {[8/14]30}
16. Bc4 Ng5 {[9/14]119}
17. Qe2 Bxe3 {[9/14]118}
18. fxe3 h4 {[10/14]28}
19. Nh1 Nxe4 {[10/14]24}
20. Bxe6 fxe6 {[10/14]12}
21. Qg4 Ng5 {[10/14]14}
22. Rxf8+ Kxf8 {[9/14]37}
23. Rf1+ Kg8 {[8/14]24}
24. Nf2 Kg7 {[8/14]4}
25. Nd3 Ra5 {[8/14]-80}
26. Qf4 Rb5 {[31]-526/b2c3}
27. Bc3 Qe7 {[34]-545/d3e5}
28. Nxe5 Nxe5 {[36]-545/c3e5}
29. Bxe5+ Kg8 {[37]-548/e5f6}
30. Bf6 Qf8 {[36]-574/f4h4}
31. Bxg5 Qxf4 {[31]-484/g5f4}
32. exf4 Rb2 {[32]-495/f1c1}
33. c4 b5 {[31]-460/f1c1}
34. Bxh4 1-0
{black resigned:engine terminates}[/pgn]

[MonteCarlo]

Now that's even weirder.

Some of the moves are again quickly identified by K13 AB on my phone as giant mistakes.

The depth output makes it look like the engine was being randomly switched between K13 MCTS, K13 non-MCTS, and the built-in Java engine (most of the depth outputs look like what K13 MCTS would show, but others look like K13 AB, and many others again look like the built-in Java engine).

I'll step through the game to see if the bad moves match K13 MCTS or built-in Java choices at the reported depths.

In the meantime, I'd definitely make sure you look at the "Use MCTS" option when you import K13 into Chess for Android.

Also, was this game played all at once, or in multiple sessions? The only way I can imagine the engine changing at different points was if the game was played in multiple sessions and different engines were loaded at different times.

Cheers!

EDIT:

Ok, yes, I've gotten over halfway through the game, and the evals and move choices have matched up very nicely with the different engines I expected from the depth outputs, with the bad moves coming from the built-in Java engine.

It seems almost certain that you were switching engines throughout the game, then, for some reason. Again, I'd guess most likely you did this in multiple sessions and/or were closing and re-opening the Chess for Android app at some moments.