Author: Vincent Diepeveen
Date: 06:58:06 12/10/99
On December 07, 1999 at 20:23:41, Robert Hyatt wrote: >On December 07, 1999 at 19:44:35, Vincent Diepeveen wrote: > >>Hello to All studiers of the game of chess with a program, >> >>Here a cool study which has next problems >> - some zugzwangs (black getting forced to move the queen) >> - the line(s) that win for white are forced till the end (singular) >> >>7k/2q3pP/6P1/8/8/8/6Pp/2RR3K/w - - > >I get this on my tiny sony notebook: > > > > > > 20 27:29 4.21 1. Rc4 Qd7 2. Rd5 Qe8 3. Re4 Qb8 4. > Re1 Qa8 5. Rf1 Qe8 6. Rd4 Qa8 7. Re4 > Qc8 8. Ra4 Qe8 9. Ra3 Qd8 10. Kxh2 > Qh4+ 11. Kg1 Qe4 > interesting, at a dual xeon with 800mb hash crafty gets at 20 ply a much lower score at 21 ply even less than half a pawn up for white. machess gets also a very small score, whereas the double nullmove should see this obviously. without nullmove, so fullwidth diep says +20.0 at 15 ply already (but needs a full hour). Diep's now running at the same hardware as crafty ran on (dual xeon with 800mb hash), so let's wait and see when diep gets an absolute winscore here with nullmove turned on. >It has liked Rc4 since about 40 seconds, and the score has gone up every >iteration. Quad xeon would be 3-4-5 times faster... I'm personally amazed that EGTB and big hashtables don't find the win for crafty but give a 0.xx score ! personally i think this is because of nullmove in crafty not seeing zugzwangs. Does macchess also not see zugzwangs, +4.0 is a bit low. >> >>black timeleft=666:40.00.00 >> - = - = - = - k ... 1 ... >> = - q - = - o O ... 2 ... >> - = - = - = O = ... 3 ... >> = - = - = - = - ... 4 ... >> - = - = - = - = ... 5 ... >> = - = - = - = - ... 6 ... >> - = - = - = O o ... 7 ... >> = - R R = - = K ... 8 ... >>white timeleft=666:40.00.00 >>white to move >> >>I estimated that for a program doing checks in qsearch >>and NOT having any extension and not having hashtable luck >>the solution is 22 ply to see. 23 ply otherwise. >> >>the solution is not found yet by my program (running >>at a dual celeron 550 with 300mb hash now under linux of >>cuz), as it's PVs are crap so far and it gives black the >>chance to sac queen somewhere near end of pv. >> >>I'm not a dreamer so i'm not gonna ask for an output >>of deep blue junior, but i love to see whether other >>progs make a chance here. >> >>It's adviced to use KRPKP and KPKP KPK >>egtb at least for this position, that should produce when >>seeing the good lines a score at HUGE depths of mate in XXX >>or something of over +15.0 >> >>I'm not using them with diep here, so perhaps diep never finds it... >>...i just pray for transposition luck right now... ...right now >>score is +7.7 which is by accident also the score for KRP KP here >>but i guess it didn't see it yet. >> >>Vincent
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.