Author: leonid
Date: 16:05:41 02/11/01
Go up one level in this thread
On February 11, 2001 at 16:20:16, Tim Foden wrote: >On February 11, 2001 at 12:09:52, leonid wrote: > >>On February 11, 2001 at 08:19:40, Tim Foden wrote: >> >>>On February 11, 2001 at 07:41:00, leonid wrote: >>> >>>>Hi! >>>> >>>>If you like to solve a mate, you can try the next one: >>>> >>>>[D]3n1nbk/5qpp/2Q2nq1/1pbQNBB1/p1p1QNn1/2B2Q2/1PP1P3/2K4R w - - >>>> >>>>Please, say your result. >>> >>>Hi Leonid, >> >>Hi, Tim! >> >>>Green Light Chess v2.13-pr1 finds a mate in 9 ater 10.51 seconds, >>>and then finds a shorter mate in 8 after 20.73 seconds. It sticks >>>with this for a full 15 mins. I guess we'll have to wait for other >>>results to see if a shorter mate is found, or proven not to exist by >>>chest. >> >>My selective was not that good here. Speediest for selective found in 11 moves >>at 1.4 sec. The most probable that my selective can't find nothing below this. >>At least, I disconnected more deep selective when I found that it take more that >>1 min. >> >> >>>Cheers, Tim. >>> >>>Analysis, GLC 2.13-pr1, 24MB hash, 920MHz Duron: >>> >>>>anal >>> Game stage: Opening >>> Current eval: 18.38 >>> Ply Time Score Nodes Principal Variation >>> 3 0.01 29.64 5045 Nfxg6+ Qxg6 Nxg6+ Nxg6 Rxh7+ Bxh7 >>> 3 0.02 29.64 5184 Nfxg6+ Qxg6 Nxg6+ Nxg6 Rxh7+ Bxh7 >>> 4 0.05 29.69 15017 Nfxg6+ Qxg6 Nxg6+ Nxg6 Rxh7+ Bxh7 Qxd8+ Bg8 >>> 4 0.09 29.69 28281 Nfxg6+ Qxg6 Nxg6+ Nxg6 Rxh7+ Bxh7 Qxd8+ Bg8 >>> 5 0.53 -- 279496 Nfxg6+ (a=29.19 b=30.19 e=29.19) >>> 5 0.54 ++ 282935 Nfxg6+ (a=-321.00 b=29.20 e=29.20) >>> 5 0.55 29.22 288504 Nfxg6+ Qxg6 Nxg6+ Nxg6 Bgxf6 Be3+ Qfxe3 Bxd5 >>> 5 0.55 29.22 288589 Nfxg6+ Qxg6 Nxg6+ Nxg6 Bgxf6 Be3+ Qfxe3 Bxd5 >>> 6 0.56 ++ 292411 Nfxg6+ (a=28.72 b=29.72 e=29.72) >>> 6 0.76 33.19 417592 Nfxg6+ Qxg6 Nxg6+ Nxg6 Bgxf6 Nxf6 Qxf6 Bf8 Rxh7+ >>> Bxh7 Qfxd8 >>> 6 1.28 33.19 793375 Nfxg6+ Qxg6 Nxg6+ Nxg6 Bgxf6 Nxf6 Qxf6 Bf8 Rxh7+ >>> Bxh7 Qfxd8 >>> 7 1.32 ++ 805216 Nfxg6+ (a=32.69 b=33.69 e=33.69) >>> 7 2.43 36.34 1614187 Nfxg6+ Nxg6 Nxf7+ Nxf7 Bgxf6 Nxf6 Qxf6 Bd4 Qxg6 Nh6 >>> Qge8 >>> 7 5.08 36.34 3523071 Nfxg6+ Nxg6 Nxf7+ Nxf7 Bgxf6 Nxf6 Qxf6 Bd4 Qxg6 Nh6 >>> Qge8 >>> 8 5.21 ++ 3602654 Nfxg6+ (a=35.84 b=36.84 e=36.84) >>> 8 10.51 320.83 7709270 Nfxg6+ Nxg6 Nxg6+ Qxg6 Qxg8+ Kxg8 Bxg6 Bf8 Qcd5+ Kh8 >>> Rxh7+ Nxh7 Qg8+ Kxg8 Bxh7+ Kh8 Qxf8# <ht> >>> 8 16.66 320.83 13745k Nfxg6+ Nxg6 Nxg6+ Qxg6 Qxg8+ Kxg8 Bxg6 Bf8 Qcd5+ Kh8 >>> Rxh7+ Nxh7 Qg8+ Kxg8 Bxh7+ Kh8 Qxf8# <ht> >>> 9 20.73 320.85 18003k Nfxg6+ Nxg6 Nxg6+ Qxg6 Qxg8+ Kxg8 Bxg6 Bf8 Qee8 Nxc6 >>> <ht> >>> 9 33.64 320.85 31079k Nfxg6+ Nxg6 Nxg6+ Qxg6 Qxg8+ Kxg8 Bxg6 Bf8 Qee8 Nxc6 >>> <ht> >>> 10 46.11 320.85 44792k Nfxg6+ Nxg6 Nxg6+ Qxg6 Qxg8+ Kxg8 Bxg6 Bf8 Qee8 Nxc6 >>> Bxh7+ Kh8 Qxf8+ Ng8 Qxg8# <ht> >>> 10 1:33 320.85 95317k Nfxg6+ Nxg6 Nxg6+ Qxg6 Qxg8+ Kxg8 Bxg6 Bf8 Qee8 Nxc6 >>> Bxh7+ Kh8 Qxf8+ Ng8 Qxg8# <ht> >>> 11 2:07 320.85 131947k Nfxg6+ Nxg6 Nxg6+ Qxg6 Qxg8+ Kxg8 Bxg6 Bf8 Qee8 Nxc6 >>> Bxh7+ Kh8 Qxf8+ Ng8 Qxg8# <ht> >>> 11 5:39 320.85 362972k Nfxg6+ Nxg6 Nxg6+ Qxg6 Qxg8+ Kxg8 Bxg6 Bf8 Qee8 Nxc6 >>> Bxh7+ Kh8 Qxf8+ Ng8 Qxg8# <ht> >>> 12 7:12 320.85 465267k Nfxg6+ Nxg6 Nxg6+ Qxg6 Qxg8+ Kxg8 Bxg6 Bf8 Qee8 Nxc6 >>> Bxh7+ Kh8 Qxf8+ Ng8 Qxg8# <ht> >>> 12 15:05 320.85 979669k Nfxg6+ Nxg6 Nxg6+ Qxg6 Qxg8+ Kxg8 Bxg6 Bf8 Qee8 Nxc6 >>> Bxh7+ Kh8 Qxf8+ Ng8 Qxg8# <ht> >>> 13 15:35 0/77 Nfxg6+ >exit >>> local: t=15:41 nps=1083291.8 n=1020411074 (f=240107661 q=780303413) >>> total: t=15:41 nps=1083291.8 n=1020411074 draws=91764 >> >>This is where I am very curious to ask you one question. Is this just like I >>read it? Your program see 1083291 nodes/second? > >Yes it is :-) But my program is not anywhere near a fast searcher really. It's >just that it seems to like this position, and with my 920MHz Duron this adds up >to a lot of nodes/sec. > >In normal games it gets anywhere between about 450kNPS and 800kNPS usually. It is still very quick. I see seldom 300k on my program in the middle of the game. Actually, my look into position (not searching for forced mate) mainly by brute force. It make search more impressive in NPS counter. Only it have no hash. Expect to do this later when will return to my programming. Today I tried Raffaela (Windows version) chess program done in 1993. Was surprised by good NPS there. Even if this program is buggy and not finished, after its NPS it have very good potential. Program on some final positions went as far as 600k on my computer. Mine will do just opposit. NPS goes down as soon as position is less populated and have less legal moves. Is your program free to see what is really its NPS on my computer? Feel me curious. Cheers, Leonid. >If I run the WAC tests at 5 seconds per position it avaerages about 700kNPS (and >solves 288/300). > >Cheers, Tim. > >>It is too far from what I can expect from my NPS. I don't know exactly speed for >>this position for my mate solver, since I don't install it but I can guess what >>it can be after second part of my program. There I have counter. >> >>Cheers, >>Leonid. >> >>> trans: probes=112029910 hits=9201810 (8.21%) draft=7150131 (6.38%) >>> tcuts: exact=502 (0.00%) upper=3102575 (2.77%) lower=3993515 (3.56%) >>> tstor: exact=1151 (0.00%) upper=23363721 (59.79%) lower=15711435 (40.21%) >>> ext: check=81987653 recap=1071090 ppush=966 1rep=2664045 thrt=0 >>> q-moves: gen=4644993 tested=830649 made/un=164207 max-dep=10 >>> max eval diff: part-1=2.36 part-2=1.71
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.