Author: Jonas Bylund
Date: 04:38:14 04/30/03
Go up one level in this thread
On April 30, 2003 at 07:08:25, Peter Berger wrote: >On April 30, 2003 at 05:29:44, Uri Blass wrote: > >>On April 29, 2003 at 17:52:27, Peter Berger wrote: >> >>>On April 29, 2003 at 17:31:46, Jarkko Pesonen wrote: >>> >>>>You can extend search time in analysis by command: st time_in_milliseconds >>>>e.g. st 8640000 >>>>and search will continue for 24 hours >>>> >>>>i.e if I remenber correctly >>>> >>>> >>>>Jarkko >>>> >>>>P.S. >>>>Ruffian is incredible >>> >>>Yes, this should definitely work if you use Ruffian in text modus (and it is >>>pretty clever ;) ). According to the author current public version of Ruffian >>>interprets analysis as "Game in 833 hours", next one won't. >>>I have only used Ruffian as an UCI engine so far or in engine matches with >>>WinBoard but I will give your tip a try when I feel like letting it analyze >>>forever next time. >>> >>>And yes, I agree that Ruffian is _really_ incredible, not only when compaired to >>>other free programs! Especially with slow timecontrols. >> >>Does Ruffian do better at long time control relative to the best commercials >>programs (Shredder7.04 and chessmaster9000). >> >>I doubt it. >> > >IMHO it is possible to play weaker than the best commercials and still be >incredible, especially for a program that is free. > >On my computers Ruffian played more successful in engine matches against other >WinBoard programs at slower time controls compaired to Blitz. > >But I am most impressed by its analysis abilities with deep searches,especially >its evaluation looks very accurate, but that can't really be tested IMHO, nor am >I interested to do such a test. Some time ago i ran these two tournaments: ------------------------------Specs:------------------------------------------ Hardware: One Celeron 1.1 Ghz 256 Mb ram (laptop). OS : WinXP PRO. GUI : Arena 0.90, Oct 02nd, 2002. Ponder=off. All engines with own book No tablebases 64 Mb hash set as default for UCI engines. 64 Mb or closest match for each Wb engine. Time control 2 min + 6 seconds for each move. ------------------------------Results:----------------------------------------- Program Elo + - Games Score Av.Op. Draws 1 Ruffian 1.0.0 : 2694 28 33 400 64.8 % 2588 26.5 % 2 Yace 0.99.56 : 2632 32 29 400 55.2 % 2596 26.5 % 3 Crafty-19.0 : 2615 33 27 400 52.5 % 2598 29.0 % 4 Aristarch 4.4 : 2610 34 27 400 51.6 % 2599 28.2 % 5 SoS_Arena : 2594 27 34 400 49.1 % 2601 25.8 % 6 Pharaon 2.62 : 2588 28 34 400 48.1 % 2601 23.8 % 7 Comet B.48 : 2575 30 33 400 46.0 % 2603 21.5 % 8 AnMon5.21 : 2565 32 32 400 44.4 % 2604 18.2 % 9 Tao 5.4 : 2526 35 29 400 38.2 % 2609 19.5 % ------------------------------Specs:------------------------------------------ Hardware: One Celeron 1.1 Ghz 256 Mb ram (laptop). OS : WinXP PRO. GUI : Arena 0.90, Oct 02nd, 2002. Ponder=off. All engines with own book No tablebases 64 Mb hash set as default for UCI engines. 64 Mb or closest match for each Wb engine. Time control 40 moves in 15 min + 40 moves in 15 + 15 min for the rest of the game. ------------------------------Results:----------------------------------------- Program Elo + - Games Score Av.Op. Draws 1 Ruffian 1.0.0 : 2706 43 49 160 66.6 % 2586 33.1 % 2 Yace 0.99.56 : 2644 49 47 160 57.2 % 2594 25.6 % 3 Aristarch 4.4 : 2629 51 42 160 54.7 % 2596 31.9 % 4 Crafty-19.0 : 2596 40 55 160 49.4 % 2600 32.5 % 5 SOS.2 for Arena : 2594 43 55 160 49.1 % 2600 25.6 % 6 Pharaon 2.62 : 2578 42 52 160 46.6 % 2602 30.6 % 7 AnMon5.21 : 2578 46 52 160 46.6 % 2602 23.1 % 8 Comet B.50 : 2539 45 48 160 40.3 % 2607 31.9 % 9 Tao 5.4 : 2535 51 47 160 39.7 % 2608 23.1 % As you can see Ruffian scored better at longer timecontrol, i know it is not inredibly long and more like rapid, but i was going for enough games to make it credible. Regards Jonas
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.