Author: Frank Phillips
Date: 02:39:32 09/02/01
Go up one level in this thread
On September 01, 2001 at 22:05:57, Peter McKenzie wrote: >After the recent discussion about the Nolot positions, and spurred on by Bruce >Morland's superb web pages on the subject, I decided to test LambChop 10.98 with >the positions. > >The last time I'd tested these positions (about 8 months ago), Chop had gotten >0/11 at 1/2 hour per position on my old P133. There had been previous versions >that solved 1/11 under the same conditions, but probably changes to the search >(ie. less extensions) made it worse on this type of test set. > >Anyway, the new 1ghz PIII is quite alot faster than the old P133 and the >software has improved significantly so I was optimistic about solving some this >time! I decided on 1 hour per position, quite a convenient time for an >overnight run. > >To cut a long story short, LambChop solved 2/11 at 1hr/position. >The solved positions were numbers 1 (in 196 seconds) and 11 (in 37 minutes). > >Some comments: > >Positon 1: >Deemed the easiest by Chop which finds Nxh6 in just over 3 minutes compared to >about 6 hours for DT-2. After an hour though, Chop's score is still slightly in >Black's favour (-0.6). Chop's pv is: >Nxh6 c3 Nf5 cxb2 Qg4 g6 Re2 Qd7 Nh4 Nf4 Qxf4 Nd5 Qd2 >I haven't played thru. this but apparently Kh2 is better than Re2. > >Position 2: >I would really like Chop to find this one, as I think its kind of obvious. >Unfortunately Chop goes for some drawing variation starting with Bf6. I will >study this one some more, I suspect singular extensions are useful here, which >would be why both DT-2 and Ferret find it fast. With its current search, I >can't see Chop finding it within a few hours. > >Position 3: >Chop likes Bxg5 (+0.4), and after an hour its pv starts with Bxg5 Bxg5 h3, which >is the same line mentioned by Hsu. > >Position 4: >Chop likes Be2 here which seems reasonable, showing about half a pawn advantage >to white. I could see this one being solved with a bit more time, because the >score for Nxe6 isn't too much lower. > >Position 5: >Nf3, +0.7. Not sure what to make of this one, e5 looks quite deep. > >Oops, running out of time here. Might post some more comments later. > >Peter Does position 2 benefit from (Deep Blue style) singular extension? I ran this overnight on mine and was pretty disappointed. It took about five hours and ply 16 before it finally latached onto Rxc5 with a positive score (greater than a minor, less than a rook). Before then if eventually favoured the draw by Bf6. (The test also showed that, for test like these, I need bigger counters for nodes and stuff. According to the log file I was doing 41Knps on a Tbird 1.2GHz. I am slow, but not that slow.) Frank
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.