Author: Harald Faber
Date: 05:52:54 04/30/01
Go up one level in this thread
On April 30, 2001 at 03:02:31, Christophe Theron wrote: >On April 29, 2001 at 23:52:52, Uri Blass wrote: > >>3r2k1/5pp1/3qp2p/2p3P1/1rPnB2P/1p1R4/P4P2/2Q1R1K1 w - - 0 1 >> >>Junior played gxh6 with 0.23 pawns against black >>The logfile of Junior gives: >>38.gxh6 -0.23/17 5:25 53 kn/s > > > >Could it be the game was played at midnight? > >I have had problems with the clock at midnight in the past. I had to write >specific code to correct this. > >I was lucky to find the problem. I found it because I often work at night, so I >noticed several times a problem with the clock, and I eventually realized it >only happened when the program was thinking and the clock passed midnight at the >same time. > >I mention this problem because dispite being officially 20 times slower, Junior >managed to reach a ply depth which is consistent with the next moves. So maybe >the problem comes from the clock. > > Christophe > Usually on my machine (K7-500) Junior in the middlegame reaches depth 15-17/18 considering permanent brain profit. That means d=15 without pb hits, sometimes even 17, and d=17 or more in the case of pb hit. So on the machine which Enrique uses, I'd expect Junior to reach at least d=18. Looking at the nps for these 3 moves there must have been s.th. strange going on. 800-1000knps should be normal on an K7-1000 for Junior6a, Junior7beta on dual should get to much more than 302knps. > > >>After 38...f5 the logfile gives >>39.Qg5 -1.16/17 4:14 302 kn/s >>After 39...Rd7 the logfiles give >>40.Rxd4 -1.29/16 22 1365 kn/s >> >>You can see that Junior had no problem to see one ply after 38.gxh6 that the >>position is bad. >> >>You can see also that Junior was probably slowed down by a factor of more than >>20 when it played 38.gxh6 >> >>The question is if Junior could find axb3 at normal conditions and if this move >>could save the game. >> >>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.