Author: Robert Hyatt
Date: 10:33:41 11/12/00
Go up one level in this thread
On November 12, 2000 at 11:12:33, Thorsten Czub wrote: >>I used programs for anlysis at long time control and I found the problem of big >>branching factor mainly with hiarcs. >>I usually do not see it with other programs like Junior or Gandalf or Tiger. >> >>I did not do statistics about but I believe that hiarcs has bigger branching >>factor at long time control when I found no problem with the branching factor of >>hiarcs in the first minutes and the problem was only later. >> >>conclusion:I guess that fast processors are not going to help hiarcs even >>without the hash tables bug. > >hiarcs always had a search that seems to need much time to come deeper. >guess he is not that much forward pruning than other programs. or whatever. >but the effect you describe was in all versions. especially the older ones. >on my fast machine hiarcs makes a good look IMO, he seems to profit from >this machine. >whatever. it would be better for mark to speed up the search in a new >version. if he is still working on chess programming. >i guess he has less time... >anyway. i like his program. it is - like mchess - full of nice knowledge. > > >>Uri There is likely a simple explanation. If you expect to search to 10 plies, you can tune your extensions to solve those tactical issues you want to solve within 10 plies. But going to 11 or 12 may well cause the extensions to over-trigger and bog the search down badly. This has always been a problem, which is why varying the hardware speeds makes no real sense once you get away from the speed of the hardware the engine is actually developed on. Some might not tune this way. Others do. It is certainly a good idea to maximize performance on the machine you are going to use...
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.