Author: Odd Gunnar Malin
Date: 04:34:40 09/22/03
Go up one level in this thread
On September 22, 2003 at 07:10:28, Mike Byrne wrote: >On September 22, 2003 at 02:50:59, Odd Gunnar Malin wrote: > >>There exist some data for how much a ply is worth. Eg. the Belle test was (from >>memory) around 200 elo/ply and a newer from Heinz nearer 100 elo/ply. > >But they also there are diminishing returns ...e.g. going from 4 ply to 5 ply is >huge ..going from 14 to 15 ply is not as beneficial > > Yes, you are right this is some of the task for the selfplay. But I wanted to see your numbers before starting to interpret those. I have no ideas if this can be done by PFII but a comparable list of search with time, ply, score and pv from null and upto plies within your patience whould have been nice. Your indication of 16 times slower could be 2 ply with bf of 4. I think the latest Heinz-test went prety high in ply (with Fritz), it was in one of the latest ICGA I think, but I'm not at home now to check. Else you could start some hundred games but that would be a torture with a handheld. Odd Gunnar
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.