Matefinder Compiles

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

Moderators: hgm, Harvey Williamson, bob

jhellis3
Posts: 340
Joined: Fri Aug 16, 2013 10:36 pm

Re: Matefinder Compiles

Post by jhellis3 » Sat Nov 25, 2017 7:05 pm

Hint: not stripping the exe does not a debug build make.

jhellis3
Posts: 340
Joined: Fri Aug 16, 2013 10:36 pm

Re: Matefinder Compiles

Post by jhellis3 » Sat Nov 25, 2017 8:43 pm

IpMan has very graciously provided optimized builds for those who want/need the best performance.

You can grab them here:
https://drive.google.com/open?id=1x7PdR ... ONeV38K8oV

I have also added the two fastest compiles to my repo's release page.

ernest
Posts: 1746
Joined: Wed Mar 08, 2006 7:30 pm

Re: Matefinder Compiles

Post by ernest » Mon Nov 27, 2017 1:20 am

jhellis3 wrote:Ok, I have uploaded compiles for non-popcount and bmi2 to the release as well.
Thanks, Joseph ! (and now also Ipman !)

Your matefinder_nopopcnt runs well
* on my old Win7 laptop
* on my old Win XP Pro x64 dualcore

Nice gift to us !

Jouni
Posts: 1782
Joined: Wed Mar 08, 2006 7:15 pm

Re: Matefinder Compiles

Post by Jouni » Mon Nov 27, 2017 10:09 pm

This old 11x seems to be unsolvable by matefinder:

[D]3qk3/8/8/8/8/8/PPPPPPPP/RNBQKBNR w KQ - 0 1

It NEVER EVER considers solution e4! How can 21 plies search be so difficult??
Jouni

jhellis3
Posts: 340
Joined: Fri Aug 16, 2013 10:36 pm

Re: Matefinder Compiles

Post by jhellis3 » Tue Nov 28, 2017 9:02 pm

How can 21 plies search be so difficult??
Think about what you are saying and the position you are posting. There are 18 legal first moves for white and all of them are mate in X. And this trend continues for quite some time. Then black has a queen, and all of its moves are also losing. Thus, you end up with a position with a much higher branching factor than normal.

(high branching factor) ^ 21 is certainly not trivial.

Secondly, can you post the line for mate in 11? I forced e4 and made some modifications to MF and it finds mate in 11 from there, or a mate in 12 from the position you posted (23 ply).

Jouni
Posts: 1782
Joined: Wed Mar 08, 2006 7:15 pm

Re: Matefinder Compiles

Post by Jouni » Tue Nov 28, 2017 9:28 pm

I got some tens of solutions by search here at CCC like

Analysis by Houdini 1.03a:

1.e4 Qd4 2.Qg4 Kd8 3.Nc3 Qd7 4.Qf4 Kc8 5.Ba6+ Kd8 6.Bb5 Qe6 7.Nd5 Qxd5 8.exd5 Kc8 9.Bc6 Kd8 10.d6 Kc8 11.Qf8#
+- (#11) Depth: 29/49 00:11:59 2986mN

BTW matefinder 2016 was fast here. OK I admit this is very special position and may be the most difficult 11 mover :).
Jouni

mibere
Posts: 3
Joined: Sun Nov 12, 2017 3:14 pm
Location: Germany
Contact:

Re: Matefinder Compiles

Post by mibere » Wed Dec 06, 2017 5:35 pm

The new macOS binaries as of Dec 5 are available, https://stockfishchess.de

Vinvin
Posts: 4013
Joined: Thu Mar 09, 2006 8:40 am

Re: Matefinder Compiles

Post by Vinvin » Wed Dec 06, 2017 6:16 pm

jhellis3 wrote:Secondly, can you post the line for mate in 11?
A couple of minutes for McBrain 3.0 on 6 cores :

Code: Select all

McBrain_2017_v30_TCEC-x86-64-modern:
...
 44/26	01:50	 1 910 473k	17 218k	+M12	1.e4 Qg5 2.Nc3 Qg7 3.d4 Qg6 4.Qf3 Qf7 5.Bb5+ Ke7 6.Bg5+ Kf8 7.Bh6+ Kg8 8.Bc4 Kh7 9.Qxf7+ Kxh6 10.Qg8 Kh5 11.Bf7+ Kh6 12.Qg6+
 45/24	02:06	 2 177 444k	17 250k	+M11	1.e4 Qg5 2.d4 Qg6 3.Nc3 Kf7 4.Nf3 Ke6 5.Ne5 Qg7 6.Nd5 Qxe5 7.dxe5 Kxe5 8.Qf3 Kd6 9.Qf6+ Kd7 10.Qe7+ Kc8 11.Qc7+
 46/22	02:14	 2 320 668k	17 292k	+M11	1.e4 Qg5 2.d4 Qg6 3.Nc3 Kf7 4.Nf3 Ke6 5.Ne5 Qg7 6.Nd5 Qxe5 7.dxe5 Kxe5 8.Qf3 Kd6 9.Qf6+ Kd7 10.Qe7+ Kc8 11.Qc7+
 47/22	02:28	 2 591 930k	17 436k	+M11	1.e4 Qd4 2.Qg4 Ke7 3.Nc3 Kf7 4.Nf3 Qf6 5.d4 Qd6 6.Ne5+ Kf8 7.Nd5 Qg6 8.Nxg6+ Kf7 9.Qd7+ Kxg6 10.Qe7 Kh5 11.Qg5+
 48/22	02:40	 2 805 690k	17 430k	+M11	1.e4 Qg5 2.d4 Qg6 3.Nc3 Kf7 4.Nf3 Ke6 5.Ne5 Qg7 6.Nd5 Qxe5 7.dxe5 Kxe5 8.Qf3 Kd6 9.Qf6+ Kd7 10.Qe7+ Kc8 11.Qc7+

McBrain_2017_v30_TCEC-x86-64-modern:
...
 49/24	02:35	 2 736 924k	17 566k	+M12	1.e4 Qd4 2.Qf3 Qg7 3.Nc3 Qf7 4.Qg4 Kd8 5.d4 Qd7 6.Qg8+ Qe8 7.Bg5+ Kd7 8.Bb5+ Kc7 9.Qxe8 Kb7 10.Qd7+ Ka8 11.Bc6+ Kb8 12.Bf4+
 50/24	03:46	 4 030 814k	17 766k	+M11	1.e4 Qd4 2.Qg4 Kd8 3.Nc3 Qd7 4.Qf4 Kc8 5.Ba6+ Kd8 6.Bb5 Qe6 7.Nd5 Qxd5 8.exd5 Kc8 9.Bc6 Kd8 10.d6 Kc8 11.Qf8+
 51/22	04:06	 4 397 971k	17 824k	+M11	1.e4 Qd4 2.Qg4 Kd8 3.Nc3 Qd7 4.Qf4 Kc8 5.Ba6+ Kd8 6.Bb5 Qe6 7.Nd5 Qxd5 8.exd5 Kc8 9.Bc6 Kd8 10.d6 Kc8 11.Qf8+
 52/22	04:33	 4 893 837k	17 869k	+M11	1.e4 Qd4 2.Qg4 Kd8 3.Nc3 Qd7 4.Qf4 Kc8 5.Ba6+ Kd8 6.Bb5 Qe6 7.Nd5 Qxd5 8.exd5 Kc8 9.Bc6 Kd8 10.d6 Kc8 11.Qf8+

McBrain_2017_v30_TCEC-x86-64-modern:
...
 45/24	01:21	 1 392 842k	17 098k	+M12	1.e4 Qd4 2.Nc3 Qg7 3.d4 Qg6 4.Qf3 Qf7 5.Bb5+ Ke7 6.Bg5+ Kf8 7.Bh6+ Kg8 8.Bc4 Kh7 9.Bxf7 Kxh6 10.Qf6+ Kh7 11.Qg6+ Kh8 12.Qh6+
 46/24	02:06	 2 205 313k	17 402k	+M11	1.e4 Qd4 2.Qg4 Kf8 3.Nc3 Qg7 4.Qc8+ Ke7 5.Bc4 Qg5 6.d4 Qa5 7.Bh6 Qe5 8.Bf8+ Kf6 9.dxe5+ Kxe5 10.Nf3+ Kf6 11.Qe6+
 47/24	02:18	 2 422 009k	17 476k	+M11	1.e4 Qg5 2.d4 Qg6 3.Qf3 Qf7 4.Bb5+ Ke7 5.Bg5+ Kf8 6.Bh6+ Ke7 7.Qg3 Qh7 8.Qc7+ Kf6 9.Qxh7 Ke6 10.Bd7+ Kf6 11.e5+
 48/22	02:26	 2 568 166k	17 497k	+M11	1.e4 Qg5 2.d4 Qg6 3.Qf3 Qf7 4.Bb5+ Ke7 5.Bg5+ Kf8 6.Bh6+ Ke7 7.Qg3 Qh7 8.Qc7+ Kf6 9.Qxh7 Ke6 10.Bd7+ Kf6 11.e5+
 49/22	02:41	 2 835 070k	17 525k	+M11	1.e4 Qg5 2.d4 Qg6 3.Qf3 Qf7 4.Bb5+ Ke7 5.Bg5+ Kf8 6.Bh6+ Ke7 7.Qg3 Qh7 8.Qc7+ Kf6 9.Qxh7 Ke6 10.Bd7+ Kf6 11.e5+

Post Reply