an early match between current Crafty and Fritz 15

[MikeB]

Code: Select all

ResultSet-EloRating>details 
   1 Fritz 15 x86-32      123 1055.0 (801.5 : 253.5)
                              1055.0 (801.5 : 253.5) Crafty_Current       -61
   2 Crafty_Current       -61 4055.0 (1757.0 : 2298.0)
                              1055.0 (253.5 : 801.5) Fritz 15 x86-32      123
                              3000.0 (1503.5 : 1496.5) Crafty 25.0 Legacy   -62
   3 Crafty 25.0 Legacy   -62 3000.0 (1496.5 : 1503.5)
                              3000.0 (1496.5 : 1503.5) Crafty_Current       -61

TC 1/1 , single core/ ponder off / random 1500 position opening book/white and black

"Crafty 25.0 Legacy" was from about two weeks ago - prior to Bob making a major change to LMR - which I believe he will abandon. Crafty current , was taking that new Crafty with one little tweak, which can be applied to the legacy Crafty. Except for this version , which was in a virtual deadheat with legacy, unless you are measuring in nano seconds, all other versions of the new crafty was close but not stronger than the legacy Crafty from two weeks ago.

As far as Fritz goes, this does show about 40 point ELO improvement over Rybka (CCLR 40/4) - would place Fritz 15 4CPU near ~3240, Crafty 4CPU near 3080 - subtract say ~100 ELO points for the 1CPU version - Fritz 1CPU ~3100 , Crafty 2940 1CPU or about 100 points better than Crafty 23.8 (2826) on the CCLR 40/4 listing

[MikeB]

an early game , this one is a win by Crafty

[pgn][Event "Computer Chess Game"]
[Site "Mac-Pro.local"]
[Date "2015.11.25"]
[Round "83"]
[White "Fritz 15 x86-32"]
[Black "Crafty_Current"]
[Result "0-1"]
[TimeControl "60+1"]
[FEN "1rbq1rk1/1pn1ppbp/p2p1np1/2pP4/P3PP2/2NB1N2/1PP1Q1PP/R1B2RK1 w - - 0 1"]
[SetUp "1"]

{--------------
. r b q . r k .
. p n . p p b p
p . . p . n p .
. . p P . . . .
P . . . P P . .
. . N B . N . .
. P P . Q . P P
R . B . . R K .
white to play
--------------}
1. Bd2 {+0.41/12} b5 {+0.07/17 4} 2. axb5 {+0.06/13 4} axb5 {+0.02/17 1.9}
3. b3 {+0.02/14 4} b4 {-0.09/16 1.9} 4. Na4 {+0.04/15 2.9} Nh5 {-0.16/16 4}
5. Ra2 {+0.09/14 4} Bg4 {-0.18/15 2.4} 6. Be3 {-0.05/13 6} f5
{-0.67/16 2.2} 7. exf5 {-0.05/13 3} Nxd5 {-0.69/16 1.7} 8. fxg6
{-0.22/13 1.8} Nhxf4 {-0.84/17 2.9} 9. gxh7+ {-0.30/14 6} Kh8
{-0.96/18 1.7} 10. Bxf4 {-0.41/15 3} Nxf4 {-0.99/17 1.7} 11. Qe4
{-0.42/14 3} Bxf3 {-1.01/18 2.2} 12. Qxf3 {-0.71/13 2.0} Bd4+
{-1.43/17 1.6} 13. Kh1 {-0.82/15 2.9} Nxd3 {-1.44/17 1.6} 14. Qxd3
{-1.04/18 2.8} e5 {-1.44/15 1.6} 15. Re1 {-0.76/11 2.1} Qh4 {-2.45/16 2.9}
16. Qd1 {-1.58/12 1.8} e4 {-2.49/16 1.5} 17. h3 {-1.72/12 1.0} Be5
{-7.35/19 1.5} 18. Re3 {-6.92/15 2.2} Qf4 19. Qg1 {-7.28/15 2.3} Bd4
{-7.35/20 1.3} 20. Ra1 {-7.28/14 1.4} Bxe3 {-7.35/18 1.2} 21. Qe1
{-9.61/14 4} Bd4 {-10.40/18 4} 22. Rb1 {-9.61/14 0.9} e3 {-10.41/18 1.2}
23. c3 {-11.18/12 2.4} Be5 {-17.87/18 1.8} 24. Qg1 {-16.26/13 2.3} e2
{-19.02/18 1.1} 25. cxb4 {-16.41/12 1.8} Bd4 {-26.10/16 1.7} 26. Nc3
{-18.25/15 0.9} Bxc3 {-327.44/16 1.1} 27. Re1 {-99.82/11 1.9} Bxe1
{-327.56/15 0.1} 28. g3 {-99.90/15 1.1} Bxg3 {-327.58/13} 29. Qa1+
{-99.92/23 1.6} Rf6 {-327.60/11} 30. Qxf6+ {-99.94/25 0.9} Qxf6 {-327.62/9}
31. Kg2 {-99.96/26 1.2} Qf2+ {-327.64/7} 32. Kh1 {-99.98/23} e1=Q#
{-327.66/5}
{Xboard adjudication: Checkmate} 0-1[/pgn]

Twice, Crafty was able to play 18. ... Qf4 and then 23...Be5 to where Fritz 15 was forced to played Qg1 in response to prevent mate. Late in the game , Crafty missed a cute mate-in-4 that would have started with Qg3, but game was over by then anyway. ( I think it was 26. ..Qg3 )

[MikeB]

forgot to adjust for the 64 bit to 32 bit conversion - so adjust Crafty down by another 20 points or so - to account for the fact that it was playing a 32 bit version for Fritz 15.

[MikeB]

Code: Select all

ResultSet-EloRating>details
   1 Fritz 15 x86-32      122 3000.0 (2276.0 : 724.0)
                              1500.0 (1128.0 : 372.0) Crafty 25.0 Legacy   -60
                              1500.0 (1148.0 : 352.0) Crafty_Current       -61
   2 Crafty 25.0 Legacy   -60 4500.0 (1868.5 : 2631.5)
                              1500.0 (372.0 : 1128.0) Fritz 15 x86-32      122
                              3000.0 (1496.5 : 1503.5) Crafty_Current       -61
   3 Crafty_Current       -61 4500.0 (1855.5 : 2644.5)
                              1500.0 (352.0 : 1148.0) Fritz 15 x86-32      122
                              3000.0 (1503.5 : 1496.5) Crafty 25.0 Legacy   -60

A good showing for Fritz 15 and Crafty. T/C 1/1 Single Core, Ponder off X5690 3.46 Ghz - The Crafty Legacy version here is only about two weeks old and it's the one Bob is going back to as the master. Certainly played better against Fritz 15 then the experimental version. If we say Fritz 15x32 is at least equal to Rybka 4.1 64 bit , and it probably is, this may put current Crafty on single core near 2940-2950 or ~ 100 + Elo over Crafty 23.8.

[mhull]

Code: Select all

ResultSet-EloRating>details
   1 Fritz 15 x86-32      122 3000.0 (2276.0 : 724.0)
                              1500.0 (1128.0 : 372.0) Crafty 25.0 Legacy   -60
                              1500.0 (1148.0 : 352.0) Crafty_Current       -61
   2 Crafty 25.0 Legacy   -60 4500.0 (1868.5 : 2631.5)
                              1500.0 (372.0 : 1128.0) Fritz 15 x86-32      122
                              3000.0 (1496.5 : 1503.5) Crafty_Current       -61
   3 Crafty_Current       -61 4500.0 (1855.5 : 2644.5)
                              1500.0 (352.0 : 1148.0) Fritz 15 x86-32      122
                              3000.0 (1503.5 : 1496.5) Crafty 25.0 Legacy   -60

A good showing for Fritz 15 and Crafty. T/C 1/1 Single Core, Ponder off X5690 3.46 Ghz - The Crafty Legacy version here is only about two weeks old and it's the one Bob is going back to as the master. Certainly played better against Fritz 15 then the experimental version. If we say Fritz 15x32 is at least equal to Rybka 4.1 64 bit , and it probably is, this may put current Crafty on single core near 2940-2950 or ~ 100 + Elo over Crafty 23.8.

Is the next official release (after 24.1) pending the conversion to DTS?