Author: leonid
Date: 06:52:51 06/10/00
Go up one level in this thread
On June 10, 2000 at 06:34:51, Ernst Walet wrote: >On June 09, 2000 at 22:46:40, leonid wrote: > >>On June 09, 2000 at 20:08:37, Jon Dart wrote: >> >>>12 ply is about right to see this, but your node counts are way >>>above what mine are. So, yes, it seems like you are over-extending >>>somehow, or maybe have inefficient move ordering. >>> >>>--Jon >> >>Thank for saying that this position demand 12 plies! I tried it first up to 10 >>ply but could not reach indicated response. Expected that, maybe, I have some >>bug more. Just fixed 5 in the last month. Dreamed to end my cleaning work. >> >>On AMD 400Mhz. >> >>Hiarcs look and find indicated response by brute force 8 ply deep. It take 14 >>minutes 58 sec. Probably its extensions goes as far as 4 next plies. >> >>My logic find indicated response only 12 ply deep. Search done by brute force. >>No extensions or transpositional tables. Time: 1 hour and 42 minutes. Average >>speed: 200200 nps. >> >>Leonid. > > >On my PIII-450 Hiarcs 7.32 switches to Qb7 in 102 seconds. I let the position >run for an hour and a half with the following result: > >New position >4r1k1/p4ppp/1pq1P3/3p1PPP/3B1Q2/P4NK1/8/3n4 b - - 0 1 > >Analysis by Hiarcs 7.32: > >1...fxe6 > -+ (-4.17) Depth: 1 00:00:00 >1...fxe6 2.Ne5 > -+ (-3.66) Depth: 2/4 00:00:00 >1...fxe6 2.f6 > -+ (-3.59) Depth: 2/8 00:00:00 >1...fxe6 2.Qe5 > -+ (-3.32) Depth: 3/11 00:00:00 >1...fxe6 2.Qe5 > -+ (-3.32) Depth: 3/13 00:00:00 >1...fxe6 2.h6 exf5 3.hxg7 Qe6 > -+ (-3.52) Depth: 4/17 00:00:00 >1...fxe6 2.Qe5 > -+ (-3.25) Depth: 5/19 00:00:00 11kN >1...fxe6 2.f6 Rc8 3.f7+ Kf8 4.Bxg7+ Kxg7 5.Qf6+ Kf8 6.Qh8+ Ke7 7.Qf6+ Kd6 8.f8Q+ >Rxf8 9.Qxf8+ Kc7 > -+ (-2.10) Depth: 5/20 00:00:01 38kN >1...fxe6 2.f6 > -+ (-1.85) Depth: 6/21 00:00:02 91kN >1...fxe6 2.f6 > -+ (-1.85) Depth: 6/23 00:00:02 101kN >1...fxe6 2.f6 Qb7 3.Qd2 gxf6 4.Bxf6 Qb8+ 5.Kh3 Qd6 6.Qxd1 Qxa3 > -+ (-1.94) Depth: 7/23 00:00:07 300kN >1...fxe6 2.f6 > -+ (-1.69) Depth: 8/25 00:00:11 482kN >1...fxe6 2.f6 Qb7 3.Qc1 gxf6 4.gxf6 e5 5.Nxe5 Rf8 6.Qxd1 Rxf6 > ³ (-0.55) Depth: 8/27 00:00:20 858kN >1...Qb7 2.Qc1 Qe7 3.Qxd1 fxe6 4.Qd3 exf5 5.Qxf5 Qf7 6.Qxf7+ Kxf7 > µ (-0.72) Depth: 8/27 00:01:42 4339kN >1...Qb7 2.Qc1 > ³ (-0.47) Depth: 9/27 00:02:57 7491kN >1...Qb7 2.Qc1 Qe7 3.a4 Qf8 4.Qxd1 fxe6 5.f6 gxf6 > ³ (-0.43) Depth: 9/29 00:03:26 8697kN >1...Qb7 2.Ne5 > = (-0.18) Depth: 10/29 00:08:13 20399kN >1...Qb7 2.Ne5 fxe6 3.f6 Rf8 4.Qg4 Qc7 5.Qxe6+ > = (-0.16) Depth: 10/30 00:11:19 28667kN >1...Qb7 2.Ne5 > = (0.09) Depth: 11/30 00:23:01 58378kN >1...Qb7 2.Ne5 fxe6 3.f6 Rf8 4.h6 g6 5.Qg4 Rd8 6.Qxd1 Qc7 > ² (0.55) Depth: 11/30 00:27:45 70055kN > >Ernst. Easually I don't look for response through the partial or quick logic. It can give very evasive and completely wrong idea about real speed of the program. Don't use also odd number of ply, since my logic don't provide all the time perfect response for odd numbers. So, when I tried on Hiarcs through the "brute force" search (fixed depth and selectivity put to zero), right response went only on 8 ply search. Ah, yes! Hash table was put to possible minimum. How you found there NPS number? I have the impression that I never have seen this. Must look more carefully. Only Rebel provide them all the time. It is true also that I almost never used Hiarcs for playing. This game is too good and addicitive for me. Few time, in the past, I was able to play chess up to the 3 o'clock in the morning. Leonid.
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.