Is Glaurung 2.0.1 now the fastest mate solver?

[Jouni]

I mean among playing programs excluding Chest. Program must display
correct mates of course. Fruit comes very close - it's difficult to say sure which one is best. I don't have any King or Hiarcs versions though.

Jouni

[Eelco de Groot]

I mean among playing programs excluding Chest. Program must display
correct mates of course. Fruit comes very close - it's difficult to say sure which one is best. I don't have any King or Hiarcs versions though.

Jouni

How about Deep Fritz 11 on this computer? Assuming it could use all those processors. Glaurung's max is now eight I believe, can Zappa use 32?

http://www.jens.tauchclub-krems.at/dive ... marks.html

HP-Server rp8420-32, 64 GB RAM 32 x 3100 123,93 (speed relative to Pentium III 1GHz) 59329 kilo nps M. Berger;

Screenshot:


Eelco

[Jouni]

Fritz 10 is totally useless as mate solver, it can't find even mate in 1:

[D]5K2/8/2qk4/2nPp3/3r4/6B1/B7/3R4 w - e6

Analysis by Fritz 10:

1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Rc1 Rd2 5.Rxc5+ Kxc5 6.Be6 Rd1
= (-0.11) Depth: 9/21 00:00:00 65kN
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Rc1 Rd2 5.Be6 Rf2+ 6.Ke7 Re2 7.Rxc5+ Kxc5
= (-0.11) Depth: 10/24 00:00:00 123kN
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Rc1 Rd2 5.Be6 Rf2+ 6.Ke7 Re2 7.Rxc5+ Kxc5
= (-0.11) Depth: 11/19 00:00:00 154kN
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Rc1 Rd2 5.Be6 Rf2+ 6.Ke7 Re2 7.Rxc5+ Kxc5
= (-0.11) Depth: 12/25 00:00:00 327kN
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Rc1 Kd6 5.Rxc5 Kxc5 6.Be6 Rd1 7.Kf7 Rf1+ 8.Kg6
= (-0.11) Depth: 13/25 00:00:00 643kN
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Rc1 Kd6 5.Rxc5 Kxc5 6.Kf7 Rd1 7.Bb3 Rf1+ 8.Ke6 Re1+ 9.Kf5
= (-0.11) Depth: 14/25 00:00:01 1259kN
1.Bxe5+ Kxe5 2.Re1+ Kd6 3.dxc6 Kxc6 4.Rc1 Kd6 5.Rxc5 Kxc5 6.Ke7 Rd1 7.Be6 Kd4 8.Kd6 Ke4+ 9.Kc5
= (-0.11) Depth: 15/37 00:00:04 4828kN

BTW is this fixed in F11?

Jouni

[perejaslav]

Analysis by Fritz 11:

1.dxe6#
+- (#1) 7/7 00:00:00 1тП

Yes, fixed

[Mike S.]

This problem must have been introduced with Fritz' GUI 9 or 10. I guess it happens if the GUI ignores the e.p. square in the FEN. It works correctly in ChessProgram8.exe (Fritz 8 GUI) of Jan. 19th 2007.

Does it work in an updated Fritz 9 GUI?

[Anabolic Karpov]

This problem must have been introduced with Fritz' GUI 9 or 10. I guess it happens if the GUI ignores the e.p. square in the FEN. It works correctly in ChessProgram8.exe (Fritz 8 GUI) of Jan. 19th 2007.

Does it work in an updated Fritz 9 GUI?

It is purely engine related and affects F9 and F10/DF10. F8 Sees M1 in GUI8, GUI 9 and GUI 10 as do other CB native engines.

regards,
AK

[Tord Romstad]

It would surprise me very much if Glaurung 2.0 is among the fastest mate solvers. It is not optimized for finding mates, it does not extend mate threats, it extends single reply to checks by only half a ply, and it never searches more than one ply of checks in the quiescence search.

In fact, I think Glaurung 1.2.1 is probably a faster mate solver than Glaurung 2 in most positions (Glaurung 2 is far better at tactics in general, though).

Tord

[Edmund]

Can you actually say what tricks make up a good mate-finder..

Are there any described algorithms?

I mean Null-move or similar isn't the key, so which part of the ordinary alpha beta search can be improved? Are there any good move ordering guidlines?

Edmund

[Jouni]

Hi Tord,

I think 1.2.1 don't show correct mates always? Question about Glaurung: what bitbases it is currently using and can it use Scorpio 5 piece set may be!

Jouni

[Uri Blass]

It would surprise me very much if Glaurung 2.0 is among the fastest mate solvers. It is not optimized for finding mates, it does not extend mate threats, it extends single reply to checks by only half a ply, and it never searches more than one ply of checks in the quiescence search.

In fact, I think Glaurung 1.2.1 is probably a faster mate solver than Glaurung 2 in most positions (Glaurung 2 is far better at tactics in general, though).

Tord

I am not sure if extending single reply to check by more than half ply help programs to be faster mate solvers.

More extensions can help programs to see some long mates very fast but it can also make them slower in case that the mate is not based on single replies.

Uri