Page 3 of 18

Re: Houdini 1.5 is released

Posted: Wed Dec 15, 2010 8:33 pm
by Houdini
Dann Corbit wrote:You'll also need to contact :
http://www.linkedin.com/pub/andrew-kadatch/3/74b/66
Thanks, a LinkedIn profile can simplify matters ;).

Re: Houdini 1.5 is released

Posted: Wed Dec 15, 2010 8:34 pm
by beachknight

Re: Houdini 1.5 is released

Posted: Wed Dec 15, 2010 8:36 pm
by Houdini
eberle wrote:Thanks Robert!
Does version 1.5 use large parges? I see no UCI option.
Regards Dieter
Yes, Houdini 1.5 will automatically use large pages if Windows can provide them.

Re: Houdini 1.5 is released

Posted: Wed Dec 15, 2010 8:45 pm
by eberle
Fine Robert, thank you!
I couldn't access your server for a while. So I couldn't read your informations there.
Regards Dieter

Re: Houdini 1.5 is released

Posted: Wed Dec 15, 2010 8:51 pm
by Houdini
eberle wrote:Fine Robert, thank you!
I couldn't access your server for a while. So I couldn't read your informations there.
Regards Dieter
Sorry for that, the last hour the web server has been completely overwhelmed by the traffic.
For the next release I'll make sure to use a provider with more bandwidth ;).

Re: Houdini 1.5 is released

Posted: Wed Dec 15, 2010 8:57 pm
by Robert Flesher
Houdini wrote:The new Houdini 1.5 is available.

Main features are improved evaluation and search (by about 35 Elo), and Gaviota EGTB support.
Detailed information and download can be found at http://www.cruxis.com/chess/houdini.htm .

I've created a mailbox for problem reports, suggestions or requests, you'll find the address on the web page.
If you enjoy Houdini, feel free to donate to the development of Houdini. Donations will be used for purchasing hardware to improve the testing of Houdini and will contribute directly to future versions.

Cheers,
Robert

Well, if first results are any indication it would seem Houndini has a new aggression to its play, when you push it pushes back! Also, it's currently mangling Rybka 4.0 SSE42 x64 on my machine. But clearly, these are still very early results, so we'll see if Rybka can turn things around. Great engine Robert! Thank-you very much :D

Re: Houdini 1.5 is released

Posted: Wed Dec 15, 2010 9:05 pm
by Nimzovik
Hey Mr. Taner thanks for the alternative site!

Re: Houdini 1.5 is released

Posted: Wed Dec 15, 2010 9:36 pm
by mwyoung
Houdini wrote:The new Houdini 1.5 is available.

Main features are improved evaluation and search (by about 35 Elo), and Gaviota EGTB support.
Detailed information and download can be found at http://www.cruxis.com/chess/houdini.htm .

I've created a mailbox for problem reports, suggestions or requests, you'll find the address on the web page.
If you enjoy Houdini, feel free to donate to the development of Houdini. Donations will be used for purchasing hardware to improve the testing of Houdini and will contribute directly to future versions.

Cheers,
Robert

Thank You Robert for Houdini 1.5,

First thing I see is the KPS are way up on my Q6600 32bit. Search speed has gone form 1.03a 2,900-3300 Kns to Version 1.5 4300-4700. Kns

First test game won by Houdini...

[Event "MWYOUNG-PC, 5m/40+5m/40+5m/40"]
[Site ""]
[Date "2010.12.15"]
[Round "1"]
[White "Houdini 1.5 w32"]
[Black "Deep Rybka 4 w32"]
[Result "1-0"]
[ECO "B90"]
[WhiteElo "2500"]
[BlackElo "2500"]
[Annotator "0.00;0.15"]
[PlyCount "145"]
[TimeControl "40/300:40/300:40/300"]

{Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz 2394 MHz W=17.7 plies; 4,
315kN/s; 11-8-2010.ctg B=13.3 plies; 168kN/s; 85 TBAs; 11-8-2010.ctg} 1. e4 {
B 0} c5 {B 0} 2. Nf3 {B 0} d6 {B 0} 3. d4 {B 0} cxd4 {B 0} 4. Nxd4 {B 0} Nf6 {
B 0} 5. Nc3 {B 0} a6 {B 0} 6. Be3 {B 0} e5 {B 0} 7. Nb3 {B 0} Be6 {B 0} 8. f3 {
B 0} Nbd7 {B 0} 9. Qd2 {B 0 Both last book move} Be7 {0.15/15 11} 10. Nd5 {0.
00/18 11 (g4)} Nxd5 {-0.01/15 16} 11. exd5 {0.01/16 0} Bf5 {-0.01/16 2} 12. Be2
{0.00/19 8} O-O {0.00/15 14 (Qc7)} 13. O-O {0.00/20 15} Bg6 {0.00/14 10 (Qc7)}
14. Rac1 {0.00/18 25} Qc7 {-0.04/14 11} 15. a4 {0.03/18 10 (c4)} f5 {-0.11/13
12 (Nf6)} 16. c4 {0.04/18 13 (a5)} a5 {-0.06/13 12} 17. Rcd1 {-0.04/18 8 (Bd3)}
Rfc8 {-0.08/13 15 (Rfd8)} 18. Kh1 {-0.02/19 13 (Bf2)} f4 {-0.16/15 7} 19. Bf2 {
-0.02/20 8} Bg5 {-0.14/14 5 (b6)} 20. Rc1 {0.13/19 10 (Qc3)} e4 {0.00/14 8} 21.
fxe4 {0.13/17 0} Nf6 {0.00/15 10} 22. Bd3 {0.14/20 17} Nxe4 {0.00/15 4} 23. Qc2
{0.14/18 0} Re8 {0.00/15 4 (Nxf2+)} 24. Nd4 {-0.04/21 34 (Bd4)} f3 {-0.08/13 5}
25. Rcd1 {-0.04/20 6} fxg2+ {-0.13/14 4} 26. Kxg2 {-0.04/19 0} Bf6 {-0.13/15 5}
27. Be3 {-0.04/19 7 (Bg1)} Qd7 {0.00/14 8 (Bxd4)} 28. Ne6 {0.11/17 8} Rxe6 {0.
00/14 2} 29. dxe6 {0.11/16 0} Qxe6 {0.00/15 4} 30. Kh1 {0.18/19 8 (Kg1)} Re8 {
0.03/14 10} 31. Qg2 {0.22/19 12} Kh8 {0.18/13 10 (Qc8)} 32. Bg1 {0.48/18 10
(Ba7)} d5 {0.55/12 15} 33. cxd5 {0.56/20 7} Qxd5 {0.55/14 3} 34. Bb5 {0.73/21
18} Qh5 {0.83/13 2} 35. Rd3 {0.82/21 21} Rb8 {0.81/14 18 (Rc8)} 36. Rf4 {0.72/
19 5 (Re3)} Ng5 {0.83/13 5} 37. Rd5 {0.97/20 9} Bf7 {1.22/14 10} 38. Rdf5 {1.
03/20 5} Be6 {1.27/14 7 (Bg6)} 39. Be2 {1.21/19 8 (Rxf6)} Qg6 {1.22/14 14} 40.
Rxa5 {1.21/17 0} Qc2 {1.19/15 18} 41. Bc4 {1.59/18 7 (h4)} Qc1 {1.02/12 3} 42.
Qg3 {1.63/19 6 (Rh4)} Rc8 {1.72/14 12 (Re8)} 43. Bxe6 {1.68/18 6} Nxe6 {1.72/
15 1} 44. Re4 {1.64/17 0} Qc6 {1.79/14 7} 45. Qg2 {1.96/19 11} Rg8 {1.79/14 6
(Rf8)} 46. Re1 {2.08/20 9 (Rb5)} Qc8 {1.94/13 8} 47. Qg4 {2.36/20 8} Qc7 {1.99/
14 6} 48. b4 {2.58/21 7 (Rb5)} Nd8 {2.31/13 8} 49. Rc5 {2.99/21 14} Qd6 {2.44/
14 7 (Qf7)} 50. b5 {3.69/19 16 (Rd1)} b6 {2.52/14 11 (h6)} 51. Rc8 {3.91/20 6
(Rh5)} Qd5+ {3.55/12 7} 52. Qe4 {3.91/18 0} Qd6 {3.77/13 8} 53. Qf3 {4.83/18
15 (Re3)} Qd7 {3.64/13 10 (Bg5)} 54. Rb8 {5.01/16 0} Qd2 {3.64/13 9 (h6)} 55.
Rd1 {5.62/15 1 (Re4)} Qc2 {4.87/12 9} 56. Rxb6 {5.70/14 0} Nf7 {5.07/12 5 (Be5)
} 57. Rxf6 {6.50/13 0 (Rc6)} Ne5 {5.15/12 4} 58. Rc6 {5.74/12 0} Qxa4 {7.32/11
2} 59. Qd5 {5.74/11 0} Nxc6 {7.32/13 7} 60. bxc6 {5.71/10 0} Qc2 {8.88/14 5
(h6)} 61. Bc5 {11.80/21 13} Qg6 {9.29/11 9 (h6)} 62. c7 {#25/18 6} Qg4 {9.29/
11 1 (h6)} 63. Rd4 {#12/17 0 (Bg1)} Qh3 {9.55/11 2} 64. Qf7 {#11/16 0 (Qxg8+)}
Qc8 {#9/10 20} 65. Qxg8+ {#10/15 0} Qxg8 {#7/14 5 (Kxg8)} 66. Rd8 {#9/14 0} h6
{#7/14 7 (h5)} 67. c8=Q {#7/13 0 (Rxg8+)} Qxd8 {#6/14 17} 68. Qxd8+ {#6/12 0}
Kh7 {#5/6 0} 69. Qe8 {#5/19 0} h5 {#4/14 10} 70. Qxh5+ {#4/17 0 (h4)} Kg8 {#2/
1 0} 71. Qg6 {#3/15 0 (Kg2)} Kh8 {#2/9 0} 72. Bf8 {#2/13 0} Kg8 {#1/9 0} 73.
Qxg7# {#1/1 0} 1-0

Re: Houdini 1.5 is released

Posted: Wed Dec 15, 2010 9:57 pm
by S.Taylor
Houdini wrote:The new Houdini 1.5 is available.

Main features are improved evaluation and search (by about 35 Elo), and Gaviota EGTB support.
Cheers,
Robert
Thanks Robert,
Very exciting. (but i'm personally still in transit in person and equipment, so i can only enjoy this by following games, like onChessbomb etc.)

For ME, it is more exciting when i am not fretting over it every day, and indeed, i'm sure that you ARE. I'm sure it wouldn't make sense to believe that anyone, not least the programmers, think, every time they sit down to it, "if only it could soon deliver a great knock out to Rybka, or atleast a clear edge".
So i find it better for my nerves, to take breaks from repeating these thougts every day, and then coming back later.

Well? that has already earned me 35elo!
I really can't wait to see display with Martin Thoresen-like match, asap!

Re: Houdini 1.5 is released

Posted: Wed Dec 15, 2010 10:32 pm
by Martin Thoresen
Houdini wrote: Yes, Houdini 1.5 will automatically use large pages if Windows can provide them.
Robert,

First, thanks for Houdini 1.5!

As for not being able to turn off Large Pages permanently; this is
unfortunate, since in TCEC Large Pages is always disabled due to
not enough system RAM.

Let me give you an example:

If I have two engines with 4 GB hash each, say Houdini 1.05 and Rybka
4, there is not enough free RAM for the Large Pages to work on either of
the two engines. So no problem so far.

But, if we replace Rybka with, say, Naum which can only use a max of
1 GB hash, then there is suddenly enough free RAM to let Houdini use
Large Pages. So this would in fact be unfair.

My wish is that you add in the UCI option to turn Large Pages off/on permanently.

Best Regards,
Martin