Author: Uri Blass
Date: 08:44:49 03/18/01
Go up one level in this thread
On March 18, 2001 at 10:39:53, Enrique Irazoqui wrote: >On March 18, 2001 at 10:16:04, Robert Hyatt wrote: > >>On March 18, 2001 at 07:34:42, Sune Larsson wrote: >> >>>On March 17, 2001 at 21:08:37, Robert Hyatt wrote: >>> >>>>On March 17, 2001 at 12:21:45, Sune Larsson wrote: >>>> >>>>> >>>>> 8/2R5/k7/P7/2PK4/8/1P2r3/8 w - - 0 1 >>>>> >>>>> >>>>> This position was sent to me just two hours ago (thanks!) >>>>> White has a won position but must avoid the last trap, >>>>> which is 1.b4?? and a draw. The programs I have tried >>>>> sees this but in different speed. (PIII 800) >>>>> >>>>> Deep Fritz 0 sec >>>>> Nimzo 8 2 " >>>>> Hiarcs 7.32 2 " >>>>> Crafty 18.03 20 " >>>>> Fritz 5.32 48 " >>>>> >>>>> >>>>> Test: 1) Does your program realize that 1.b4?? is a draw? >>>>> 2) How much time needed? >>>>> >>>>> Sune >>>> >>>> >>>>I am trying to understand your question. >>>> >>>>On my PIII/750 notebook, crafty changes from b4 to Kc3 after >>>>4 seconds... and never considers b4 again... >>>> >>>>If I play b4 myself, it takes .8 seconds to get a score >>>>of 0.00... >>>> >>>>Not sure why yours took 20 seconds to see that... >>> >>> >>> Here is the output from Crafty 18.03, CB native, PIII 800, 128 MB Hash. >>> Infinite analyse mode: >>> >>> >>>Ny ställning >>>8/2R5/k7/P7/2PK4/8/1P2r3/8 w - - 0 1 >>> >>>Analysis by Crafty 18.03: >>> >>>1.b4 >>> +- (4.88) Djup: 1/2 00:00:01 >>>1.b4 Te4+ 2.Kd5 Td4+ 3.Ke5 Te4+ 4.Kd6 >>> +- (4.78) Djup: 5/9 00:00:01 >>>1.b4 Te4+ 2.Kd5 Td4+ 3.Ke5 Te4+ 4.Kd6 >>> +- (4.78) Djup: 5/9 00:00:01 >>>1.b4 Te4+ 2.Kd5 Td4+ 3.Ke5 Te4+ 4.Kd6 >>> +- (4.78) Djup: 5/9 00:00:01 >>>1.b4 Te4+ 2.Kd5 Td4+ 3.Ke5 Te4+ 4.Kd6 >>> +- (4.78) Djup: 5/9 00:00:01 >>>1.b4 Te4+ 2.Kd5 Td4+ 3.Ke5 Te4+ 4.Kd6 Td4+ 5.Ke6 >>> +- (4.58) Djup: 6/13 00:00:01 8kN, tb=18 >>>1.b4 Te4+ 2.Kd5 Td4+ 3.Ke5 Te4+ 4.Kd6 Td4+ 5.Ke6 Td6+ >>> +- (4.58) Djup: 7/14 00:00:01 15kN, tb=21 >>>1.b4 Te4+ 2.Kd5 Td4+ 3.Ke5 Te4+ 4.Kd6 Td4+ 5.Ke6 Te4+ 6.Kd7 Td4+ 7.Ke7 Td1 >>> +- (4.48) Djup: 8/16 00:00:01 31kN, tb=41 >>>1.b4 Te4+ 2.Kd5 Td4+ 3.Ke5 Te4+ 4.Kd6 Td4+ 5.Ke6 Te4+ 6.Kd7 Td4+ 7.Ke8 Te4+ >>>8.Kd8 >>> +- (4.38) Djup: 9/17 00:00:01 66kN, tb=129 >>>1.b4 Te4+ 2.Kd5 Td4+ 3.Ke5 Te4+ 4.Kd6 Td4+ 5.Ke6 Te4+ 6.Kf6 Te6+ 7.Kg5 Te5+ >>>8.Kf4 Te4+ 9.Kf5 Td4 >>> +- (4.38) Djup: 10/19 00:00:02 137kN, tb=270 >>>1.b4 Te4+ 2.Kd5 Td4+ 3.Ke5 Te4+ 4.Kd6 Td4+ 5.Ke6 Te4+ 6.Kf6 Tf4+ 7.Kg6 Tf6+ >>>8.Kh5 Tf5+ 9.Kh6 Tf6+ 10.Kg5 Te6 >>> +- (4.18) Djup: 11/21 00:00:03 347kN, tb=693 >>>1.b4 Te4+ 2.Kd5 Td4+ 3.Ke5 Te4+ 4.Kd6 Td4+ 5.Ke6 Te4+ 6.Kf6 Tf4+ 7.Kg6 Tg4+ >>>8.Kh6 Tg6+ 9.Kh5 Tg5+ 10.Kh4 Tg4+ 11.Kh3 Tf4 >>> +- (4.00) Djup: 12/25 00:00:05 765kN, tb=1686 >>>1.b4 Te4+ 2.Kd5 Td4+ 3.Ke5 Te4+ 4.Kd6 Td4+ 5.Ke6 Te4+ 6.Kf6 Tf4+ 7.Kg6 Tg4+ >>>8.Kh6 Tg6+ 9.Kh5 Tg5+ 10.Kh4 Tg4+ 11.Kh3 Tg3+ 12.Kh2 Tf3 >>> +- (4.00) Djup: 13/25 00:00:12 2668kN, tb=5042 >>>1.b4 Te4+ 2.Kc5 Txc4+ 3.Kd6 Txb4 >>> = (0.01) Djup: 14/29 00:00:26 6649kN, tb=13601 >>>1.Kc3 >>> ² (0.41) Djup: 14/29 00:00:28 7605kN, tb=14685 >>> >>>(Larsson, KIL 18.03.2001) >> >> >>Something is odd there. ie above, 3 seconds, 347K nodes. That doesn't >>look like a PIII/800 speed. My PIII/750 searches at about 500K nodes per >>second, rather than your 115K. >> >>Is something else running at the same time? > >On a dual P933, Crafty 18.6 drops b4 after 1 second. Maybe Sune's Crafty is >slowed down by tablebase access or memory swapping? The explanation for the difference cannot be only the fact that crafty is slowed down. Sune's Crafty finds b4 after 28 seconds and is searching 7605 knodes in these 28 seconds. It is still slow for pIII800 but it is not 115 knodes per second but more than 270 knodes per second. I see that sune's Crafty needs 7605 knodes when your Crafty18.06 needs only 1277 knodes to avoid b4. I wonder what is the reason for this difference. Uri
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.