Komodo 10.4 released

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

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Werewolf
Posts: 1254
Joined: Thu Sep 18, 2008 8:24 pm

Re: Komodo 10.4 released

Post by Werewolf » Mon Mar 20, 2017 10:50 am

Cumnor wrote:Hi Carl,

If your going to change to K10.4 is there a simple way of changing a remote engine or do you spend days like me changing all the engines by hand.

At the moment I have 120 single core remote engines on my IDeA projects and I am rather reluctant to spend more days changing each one by hand.
120 cores - impressive!

This is the best I can suggest:

Create "dummy folders" for each of your Slave machines. These dummy folders should contain everything particular for each machine, RTHomeServer configured with correct port numbers for each core and so on.
So, for example, if you have a 24 core machine (I assume you have some dual xeons to get to 120 cores) you'd create 1 dummy folder with 24 sub folders, each sub folder with RTHomeServer setup correctly. You'd also have at the bottom an "Auto-start" file, which I can email you if you don't have. It's just a tiny Notepad program which opens up RTHomeserver automatically for each core rather than by doing it by hand.

Anyway do NOT include the engine in the dummy folders, leave it out.

Then when a new engine comes out, copy the dummy folder for each machine and put it somewhere on C:
Then just drop the new engine into each sub-folder. Then you'll need to get each instance of the engine working properly (this will have to be done manually for each one, but should take 10 seconds per core).

Repeat for each Slave machine.

For 120 cores I think it'll still take you a full day to change the engine over - it's still very tedious - but at least you don't have to manually configure the RTHomeserver.

Cumnor
Posts: 63
Joined: Fri Jan 27, 2012 3:23 pm
Location: Cumnor, Oxford, UK
Contact:

Re: Komodo 10.4 released

Post by Cumnor » Mon Mar 20, 2017 11:30 am

Hi Carl,

Your suggestion is more or less what I am doing to change to a new engine.

As for the loading and unloading of the remote engines I use batch files to do this.

If I remember correctly Lucas says he does all these kind of things through powershell when he was setting up the Cluster. So now I am looking into this and if I find a good way to do it I will let you know.

Regards
Kevin
Moderator of Rybka forum
Admin of Infinitychess playing server

Jesse Gersenson
Posts: 575
Joined: Sat Aug 20, 2011 7:43 am
Contact:

Re: Komodo 10.4 released

Post by Jesse Gersenson » Mon Mar 20, 2017 1:21 pm

Cumnor wrote: If your going to change to K10.4 is there a simple way of changing a remote engine or do you spend days like me changing all the engines by hand.
Sounds like a script which'd take 2 minuets to write. If one of you wants to PM me the:
1. batch script which launches the engines
2. two example config files
3. a screenshot of the intended directory structure
I'll write the script.

Just a guess...

Code: Select all

#!/bin/bash
for i in $(seq 1 120);do 
  mkdir engine-$i;
  echo '[main]' > engine-$i/RTHomeServer.ini;
  echo "port `echo 7000 + $i |bc -l`" >> engine-$i/RTHomeServer.ini;
  echo "engine=K10.4.exe" >> engine-$i/RTHomeServer.ini;
  cp "K10.4.exe" engine-$i/
done


User avatar
Nordlandia
Posts: 2588
Joined: Fri Sep 25, 2015 7:38 pm
Location: Sortland, Norway

Re: Komodo 10.4 released

Post by Nordlandia » Mon Mar 20, 2017 1:56 pm

TC: 20m+20s + Ponder

[pgn][Event "DESKTOP-DD1B2QU, Rapid 20m+20s"]
[Site "DESKTOP-DD1B2QU"]
[Date "2017.03.20"]
[Round "1"]
[White "Komodo 10.4 64-bit"]
[Black "McBrain 2.0 64 POPCNT"]
[Result "1-0"]
[SetUp "1"]
[FEN "1rbqnk2/4ppbp/p2p2p1/2p5/2P1PP2/3P2PP/1P1QN1B1/q1BNRRK1 w - - 0 1"]
[PlyCount "71"]

1. Qe3 Qa4 2. Bd2 e6 3. f5 Kg8 4. Bc3 e5 5. Qf2 Qad7 6. Ne3 Nf6 7. Ra1 Qf8 8.
Ra2 Bh6 9. Be1 Qg7 10. Nc3 Bb7 11. Bd2 Qd8 12. fxg6 fxg6 13. Kh2 Bc8 14. Qe1
Rb3 15. Qe2 Kh8 16. Be1 Qdd7 17. Ncd5 Nxd5 18. Nxd5 Qc6 19. Bc3 Bg5 20. h4 Bd8
21. Bh3 Bxh3 22. Kxh3 Rb7 23. Kg2 h6 24. b4 cxb4 25. Bxb4 Rf7 26. Rxf7 Qxf7 27.
Qe3 Kh7 28. Rf2 Qg7 29. Rf3 a5 30. Ba3 Qcd7 31. Qf2 h5 32. Rf8 g5 33. Qf3 Qh3+
34. Kf2 Qh2+ 35. Kf1 Qh3+ 36. Ke2 1-0

[/pgn]

https://nb.lichess.org/2h2t0o2A

http://www.viewchess.com/cbreader/2017/ ... 50765.html

mjlef
Posts: 1448
Joined: Thu Mar 30, 2006 12:08 pm
Contact:

Re: Komodo 10.4 released

Post by mjlef » Mon Mar 20, 2017 2:14 pm

Nordlandia wrote:Do principle of "Major Piece Redudancy" affects Komodo evaluation if there is two queens for one side, and many pieces remaining elsewhere on the board?

Penalty for the Queen pair (Larry Kaufman called it "principle of redundancy" although how much of penalty implied is unknown. In theory rook pair, knights and queens suffers from this.

Viktor Korthnjoi --->

Hypothesis "Having two Qs may be clearly worse than having a Q and 2 Rs. That would definitely be a special case.
During the analysis, I discovered something very remarkable: the board is simply too small for two Queens of the same color. They only get in each other's way. I realize that this might sound stupid, but I fully mean it. The advantage is much less than one would expect by counting material.
[d]1rbqnk2/4ppbp/p2p2p1/2p5/2P1PP2/3P2PP/1P1QN1B1/q1BNRRK1 w - - 0 1
Komodo's evaluation does indeed include a measure of queen redundancy.

carldaman
Posts: 1835
Joined: Sat Jun 02, 2012 12:13 am

Re: Komodo 10.4 released

Post by carldaman » Mon Mar 20, 2017 6:37 pm

Nordlandia wrote:Do principle of "Major Piece Redudancy" affects Komodo evaluation if there is two queens for one side, and many pieces remaining elsewhere on the board?

Penalty for the Queen pair (Larry Kaufman called it "principle of redundancy" although how much of penalty implied is unknown. In theory rook pair, knights and queens suffers from this.

Viktor Korthnjoi --->

Hypothesis "Having two Qs may be clearly worse than having a Q and 2 Rs. That would definitely be a special case.
During the analysis, I discovered something very remarkable: the board is simply too small for two Queens of the same color. They only get in each other's way. I realize that this might sound stupid, but I fully mean it. The advantage is much less than one would expect by counting material.
[d]1rbqnk2/4ppbp/p2p2p1/2p5/2P1PP2/3P2PP/1P1QN1B1/q1BNRRK1 w - - 0 1
Besides redundancy, a simplistic argument would be that the second Queen is only worth a Q-1 since the promoted pawn goes off the board. Add to that the fact that 2R = Q+1 and you already have a 2 point deficit when comparing a 2nd Q to 2R.

User avatar
Nordlandia
Posts: 2588
Joined: Fri Sep 25, 2015 7:38 pm
Location: Sortland, Norway

Re: Komodo 10.4 released

Post by Nordlandia » Mon Mar 20, 2017 7:26 pm

1 minute of infinite search à la 4096 Mb and default settings.

Komodo 10.4 and Chiron 4 is the two engines that stands out.


Analysis by Houdini 5.01 Pro x64-pext:

1.Qe3 Bb7
-/+ (-1.33 --) Depth: 25/50 00:01:13 1372MN
(, 20.03.2017)

Analysis by Stockfish 180317 64 BMI2:

1.Qe3 Qa4
-/+ (-0.91 --) Depth: 31/44 00:01:01 883MN
(, 20.03.2017)

Analysis by Houdini 4 Pro x64:

1.Qe3 Kg8 2.Bd2 Qa4 3.Bc3 Bb7 4.f5 Bxc3 5.Nexc3 Qad7 6.Rf2 Nf6 7.Ref1 Bc6 8.g4 Qf8
-/+ (-0.94) Depth: 26/63 00:01:55 2313MN
(, 20.03.2017)

Analysis by Komodo 10.4 64-bit:

1.Qe3 Qaa5
= (-0.06 --) Depth: 30 00:01:02 781MN, tb=30
(, 20.03.2017)

Analysis by Chiron 4 64bit:

1.Ndc3 Bd7 2.f5 Nf6 3.Qf4 Kg8 4.Qf2 Qaa5 5.g4 Qe8 6.g5 Nh5 7.Bf3 Be5 8.Bxh5 gxh5 9.Qh4 Qad8 10.Qxh5 Rb3 11.Nf4 Bd4+ 12.Kh2 e6 13.fxe6 fxe6 14.g6
= (-0.10) Depth: 26/50 00:02:09 1718MN
(, 20.03.2017)

Analysis by Fritz 15:

1.f5 Kg8 2.Ne3 Bxb2 3.Ng4 Nf6 4.Bxb2 Qxb2 5.Qxb2 Rxb2 6.Nxf6+ exf6 7.Nf4 Bb7 8.fxg6 fxg6 9.Rf2 Rxf2 10.Kxf2 Qa5 11.Bf1 Qc3 12.Ne6 Qd2+ 13.Be2 Bc8
-+ (-2.27) Depth: 21 00:01:16 482MN
(, 20.03.2017)

Analysis by Deep Rybka 4.1 SSE42 x64:

1.d4 Qb6 2.e5[] Bb7 3.Bxb7 Rxb7 4.Ndc3 Qaa5 5.Kh2 cxd4 6.Nxd4 Qbc5 7.Qd3 Qc8 8.e6 f5 9.b3 Kg8 10.Nd5 Nf6 11.Bd2
-+ (-2.18) Depth: 17 00:01:46 64897kN
(, 20.03.2017)

Average evaluation = 1,12


Standard deviation = 0.82

S.Taylor
Posts: 8399
Joined: Thu Mar 09, 2006 2:25 am
Location: Jerusalem Israel

Re: Komodo 10.4 released

Post by S.Taylor » Mon Mar 20, 2017 8:16 pm

lkaufman wrote:We have just released Komodo 10.4 at komodochess.com. It is about 20 elo stronger than Komodo 10.3 (at blitz levels) and has one new UCI option, "variety". Improvements were in both search and eval. Linux users will see an even larger elo gain. In my testing it leads by 13 elo against Houdini 5.01 at 2' + 1" after 1350 games on four threads. This will probably be the last of the 10.x series. As usual, it is free for subscribers and available for a 20% discount for buyers of Komodo 9 or later version.
I was very excited to see this headline about a Komodo 10.4. But if it is only talking blitz, it's not yet what interests me.
UNLESS it contributes towards success in LTC.

I don't think it is much to be proud of to merely be a blitz computer champion. For humans playing blitz, it is an interesting challenge, but for computers, the only interesting thing is really at long time controls.

mjlef
Posts: 1448
Joined: Thu Mar 30, 2006 12:08 pm
Contact:

Re: Komodo 10.4 released

Post by mjlef » Mon Mar 20, 2017 8:18 pm

S.Taylor wrote:
lkaufman wrote:We have just released Komodo 10.4 at komodochess.com. It is about 20 elo stronger than Komodo 10.3 (at blitz levels) and has one new UCI option, "variety". Improvements were in both search and eval. Linux users will see an even larger elo gain. In my testing it leads by 13 elo against Houdini 5.01 at 2' + 1" after 1350 games on four threads. This will probably be the last of the 10.x series. As usual, it is free for subscribers and available for a 20% discount for buyers of Komodo 9 or later version.
I was very excited to see this headline about a Komodo 10.4. But if it is only talking blitz, it's not yet what interests me.
UNLESS it contributes towards success in LTC.

I don't think it is much to be proud of to merely be a blitz computer champion. For humans playing blitz, it is an interesting challenge, but for computers, the only interesting thing is really at long time controls.
We will have to await more LTC games by the various testing groups, but I am seeing very nice results so far at LTC. But not enough games to know with much certainly.

lkaufman
Posts: 4069
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: Komodo 10.4 released

Post by lkaufman » Mon Mar 20, 2017 8:28 pm

carldaman wrote:
Nordlandia wrote:Do principle of "Major Piece Redudancy" affects Komodo evaluation if there is two queens for one side, and many pieces remaining elsewhere on the board?

Penalty for the Queen pair (Larry Kaufman called it "principle of redundancy" although how much of penalty implied is unknown. In theory rook pair, knights and queens suffers from this.

Viktor Korthnjoi --->

Hypothesis "Having two Qs may be clearly worse than having a Q and 2 Rs. That would definitely be a special case.
During the analysis, I discovered something very remarkable: the board is simply too small for two Queens of the same color. They only get in each other's way. I realize that this might sound stupid, but I fully mean it. The advantage is much less than one would expect by counting material.
[d]1rbqnk2/4ppbp/p2p2p1/2p5/2P1PP2/3P2PP/1P1QN1B1/q1BNRRK1 w - - 0 1
Besides redundancy, a simplistic argument would be that the second Queen is only worth a Q-1 since the promoted pawn goes off the board. Add to that the fact that 2R = Q+1 and you already have a 2 point deficit when comparing a 2nd Q to 2R.
The diagram shows R + N for the queen, not two rooks.
Komodo rules!

Post Reply