Author: Robert Hyatt
Date: 21:38:40 01/21/00
Go up one level in this thread
On January 21, 2000 at 15:47:02, Amir Ban wrote: >On January 20, 2000 at 20:28:54, Bruce Moreland wrote: > >>On January 20, 2000 at 09:15:20, Amir Ban wrote: >> >>>There were indeed some PV's in moves 36 & 37 that looked like typos. The Rxf6 >>>line could be explained in this way. There was another which looked weird in a >>>different way and I think needed a different theory. >> >>Are you really arguing that the logs were manufactured? That sounds really >>extreme. >> >>bruce > >I don't think the logs were manufatured, and didn't say they were. I think the >transposition table theory may explain the Rxf6 line, but there are other PV >lines that are illogical in a different way, like the line at at T=82 for move >36 2nd game, where it needs to be explained how the move Qa6a7 can be best for >white. > >Amir That looks like the kind of thing I would expect with hashing... Don't forget that DB used a non-shared memory machine (the SP). I am not quite sure how you would reconstruct the PV with a hash table distributed over several cpus. What would you do if you probe two different cpus (trying to reconstruct the PV) and get a hit from both, with a different suggested move? When I experimented a lot with mtd(f) a couple of years ago, this was the most annoying thing I found... because you _never_ get a PV since every search fails high or fails low. Sucks.. but sometimes you put up with it... I think Vincent is doing this, as I have seen some ultra-oddball moves in his PVs from time to time, moves that were obviously not in line with the score he returned...
This page took 0.01 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.