Komodo 5.1r2 64-bit 'bug'

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.
Post Reply
leavenfish
Posts: 282
Joined: Mon Sep 02, 2013 6:23 am

Komodo 5.1r2 64-bit 'bug'

Post by leavenfish » Wed Sep 04, 2013 11:58 pm

New to the forum.

No, it's not a bug in evaluation or playing...but unlike the previous versions, it does not work properly in Chessbase 12.

The main reason for this version (5.1) was for multi-core operation. However, when you use multiple CPU's in analysis (the reason I have the last 3 versions of Komodo), the 'Depth' registered goes crazy so you never know exactly what depth the output is for....which makes it kind of useless as an analysis tool. Depth displays find on single core, just not multi-core.

I've not seen this bug mentioned in this forum though I've looked for the past month or more. Has anyone else noticed it?

gordonr
Posts: 137
Joined: Thu Aug 06, 2009 6:04 pm
Location: UK

Re: Komodo 5.1r2 64-bit 'bug'

Post by gordonr » Thu Sep 05, 2013 8:09 pm

Hi,

FWIW, I don't see any issues using Komodo 5.1r2 (or other versions) in CB 12 GUI. I'm using "6 CPUs" to use CB GUI terms.

e.g.

r1bqk2r/2ppbppp/p1n2n2/1p2p3/B3P3/2N2N2/PPPP1PPP/R1BQ1RK1 w kq - 0 1

Analysis by Komodo 5.1r2 64-bit:

7.Bb3
= (0.19 ++) Depth: 7 00:00:01 33kN
7.Bb3 Bb7 8.d3 0-0 9.a3 h6 10.Be3
= (0.20) Depth: 7 00:00:01 40kN
7.Bb3 Bb7 8.d3 0-0 9.Nd5 Nxd5 10.Bxd5 Bc5 11.Bxc6 Bxc6
= (0.27) Depth: 8 00:00:01 56kN
7.Bb3 Bb7 8.d3 0-0 9.Nd5 h6 10.Nxe7+ Qxe7 11.h3 Nd4 12.Nxd4
= (0.23) Depth: 9 00:00:01 78kN
7.Bb3 Bb7 8.d3 0-0 9.Nd5 h6 10.Nxf6+ Bxf6 11.Be3 Nd4 12.Nxd4 exd4
= (0.23) Depth: 10 00:00:01 131kN
7.Bb3 Bb7 8.d3 0-0 9.Nd5 h6 10.h3 Bc5 11.Be3 Bxe3 12.Nxe3
+/= (0.32) Depth: 11 00:00:01 338kN
7.Bb3 Bb7 8.d3 0-0 9.Nd5 h6 10.Nxe7+ Qxe7 11.c3 Na5 12.Bc2 d6 13.Be3 c5
+/= (0.33) Depth: 12 00:00:01 494kN
7.Bb3 d6 8.Nd5 Na5 9.Nxe7 Qxe7 10.d3 Nxb3 11.axb3 0-0 12.h3 h6 13.Bd2 Be6 14.b4 Rfe8 15.Be3
= (0.19) Depth: 13 00:00:01 1339kN
7.Bb3 d6 8.Nd5 Na5 9.Nxe7 Qxe7 10.d3 0-0 11.h3 Nxb3 12.axb3 h6 13.Bd2 Be6 14.b4 Rfd8 15.Be3
= (0.19) Depth: 14 00:00:01 1626kN
7.Bb3 0-0 8.d3 d6 9.Nd5 Na5 10.Bd2 Nxb3 11.Nxe7+ Qxe7 12.axb3 h6 13.h3 c5 14.b4 c4 15.Qe2 cxd3 16.Qxd3 Bb7
= (0.27) Depth: 15 00:00:01 3699kN
7.Bb3 0-0 8.d3 d6 9.Nd5 Na5 10.Bd2 Nxb3 11.Nxe7+ Qxe7 12.axb3 h6 13.b4 Be6 14.Qe2 Rfc8 15.h3 c5 16.bxc5 Rxc5 17.Be3
= (0.19) Depth: 16 00:00:01 6467kN
7.Bb3 d6 8.Nd5 Na5 9.Nxe7 Qxe7 10.d3 0-0 11.Bd2 Nxb3 12.axb3 c5 13.b4 c4 14.Bg5 Be6 15.Bxf6 Qxf6 16.h3 h6 17.d4 Qf4 18.dxe5
= (0.21) Depth: 17 00:00:02 11228kN
7.Bb3 d6 8.Nd5 Na5 9.Nxe7 Qxe7 10.d3 Nxb3 11.axb3 c5 12.Ra5 h6 13.h3 0-0 14.Bd2 b4 15.Be3 Qc7 16.Qa1 Bb7 17.Nd2
= (0.18) Depth: 18 00:00:03 15074kN
7.Bb3 d6 8.Nd5 Na5 9.Nxe7 Qxe7 10.d3 Nxb3 11.axb3 c5 12.Bd2 h6 13.h3 0-0 14.Qe2 Bb7 15.Ra5 Rac8 16.Rfa1 b4 17.R5a4 Rce8
= (0.21) Depth: 19 00:00:04 24044kN
7.Bb3 d6 8.Nd5 Na5 9.Nxe7 Qxe7 10.d3 h6 11.Bd2 Nxb3 12.axb3 c5 13.Re1 0-0 14.h3 Be6 15.b4 c4 16.d4 exd4 17.Nxd4 Bd7 18.f3 Qe5 19.c3
= (0.17) Depth: 20 00:00:05 33785kN
7.Bb3 d6 8.Nd5 Na5 9.Nxe7 Qxe7 10.d3 h6 11.Re1 0-0 12.Bd2 Nxb3 13.axb3 c5 14.h3 b4 15.Be3 Be6 16.Nd2 a5 17.Qf3 d5 18.exd5 Bxd5 19.Ne4 Nxe4
= (0.17) Depth: 21 00:00:08 55911kN
7.Bb3 d6 8.Nd5 Na5 9.Nxe7 Qxe7 10.d3 h6 11.Bd2 Nxb3 12.axb3 c5 13.h3 0-0 14.Re1 Be6 15.b4 Qc7 16.bxc5 Qxc5 17.b4 Qc7 18.Ra5 Rfc8 19.c3 Qe7 20.Re3
= (0.21) Depth: 22 00:00:15 108MN

leavenfish
Posts: 282
Joined: Mon Sep 02, 2013 6:23 am

Re: Komodo 5.1r2 64-bit 'bug'

Post by leavenfish » Fri Sep 06, 2013 1:01 am

If you can tell me how to copy and paste...or otherwise get the output into the post, I'll do it. I'm staring at a depth of 20438 at the moment....

If I right click and 'clip analysis', all I get is this...which does not show the depth:


751: Smith,B - Zhang,F, Tenneessee Open
r1bqk2r/pp3pbp/2n1pnp1/2pP4/2B5/1QN1PN2/PP3PPP/R1B2RK1 b kq - 0 1

Analysis by Komodo 5.1r2 64-bit:

1. = (0.04): 9...exd5 10.Bxd5 Nxd5 11.Nxd5 Be6 12.e4 0-0 13.Be3 b6 14.Rad1 Re8 15.Qc2 Qc8 16.h3 Nb4 17.Qb3 Nxd5 18.exd5 Bf5 19.Rfe1 Qd7 20.Bf4 Rad8
2. = (0.10): 9...Nxd5 10.Bxd5 exd5 11.Nxd5 Be6 12.e4 0-0 13.Be3 b6 14.Rad1 Re8 15.h3 Qd6 16.Ng5 Nd4 17.Bxd4 cxd4 18.Qf3 Bxd5 19.exd5 Qe7 20.h4 Rad8
3. +/= (0.49): 9...Na5 10.Bb5+ Kf8 11.Qa3 exd5 12.Qxc5+ Kg8 13.Rd1 Bf8 14.Qd4 a6 15.Ba4 Nc6 16.Bxc6 bxc6 17.e4 c5 18.Qd3 d4 19.Bg5 Be7 20.Bxf6 Bxf6 21.Nd5 Be6 22.Qc4 Bxd5
4. +- (1.74): 9...Ne7 10.dxe6 Bxe6 11.Bxe6 fxe6 12.Ng5 Qb6 13.Nb5 Ned5 14.e4 0-0 15.exd5 exd5 16.Nc3 Qxb3 17.axb3 Rfe8 18.Nb5 Re7 19.Bf4 h6 20.Nf3 a6 21.Nc3 d4 22.Na4 Nd7 23.h3
5. +- (1.93): 9...Nb4 10.dxe6 fxe6 11.Ng5 0-0 12.Bxe6+ Kh8 13.Bxc8 Rxc8 14.Nf7+ Rxf7 15.Qxf7 Rc7 16.Qc4 Nd3 17.Rd1 Rd7 18.f3 a6 19.a4 Bh6 20.Qe6 Bg5 21.b3

(Smith, 05.09.2013)
[/img]

User avatar
Mike S.
Posts: 1480
Joined: Thu Mar 09, 2006 4:33 am

Re: Komodo 5.1r2 64-bit 'bug'

Post by Mike S. » Fri Sep 06, 2013 11:31 am

That is due to multivariation analysis. The depth is included in the copyable engine output, in singlevariation analysis only (unfortunately, in ChessBase interfaces).

Also, note that Fritz replaces the original move number with "1", in the FEN the copied output starts with. I don't know if the database program does the same.
Regards, Mike

leavenfish
Posts: 282
Joined: Mon Sep 02, 2013 6:23 am

Re: Komodo 5.1r2 64-bit 'bug'

Post by leavenfish » Sat Sep 07, 2013 5:12 am

Mike S. wrote:That is due to multivariation analysis. The depth is included in the copyable engine output, in singlevariation analysis only (unfortunately, in ChessBase interfaces).
Yes, that does appear to be the case - the bug manifests itself in multi-variation analysis, not single variation. If I have, say the top 2 or 3 lines shown in the analysis window, at multi-core, the depth goes crazy. But if I reduced the analysis to 1 line, it shows me the true depth. I had not realized that, thanks!

As I use this engine for analysis only, it sure would be nice if it worked properly in multi-variation mode - many people own Chessbase...problably more than own any other database. Komodo is now far less useful than it could be.

Larry knows of the problem, but was unable to replicate it to fix it. Perhaps with this knowledge he can do something about it.

Post Reply