Author: leonid
Date: 16:17:43 03/03/00
Go up one level in this thread
On March 03, 2000 at 16:31:45, blass uri wrote: >On March 03, 2000 at 16:11:34, Tom Kerrigan wrote: > >>On March 03, 2000 at 15:53:31, blass uri wrote: >> >>>1)Can hash tables help to reduce the branching factor when they are full? >> >>Of course. >> >>They obviously don't reduce the branching factor when they're empty. >> >>-Tom > >They are never empty if the program uses them but there is a time when they are >also not full. > >My question was about comparing the branching factor after they are full >relative to the branching factor at the same ply depths assuming not using hash >tables. > >I know that Fritz tells me when the hash tables are full and I think that the >branching factor is bigger after they are full. > >I think based on my experience that the branching factor of Hiarcs is also >bigger when I give it to analyze for a long time but I did not do statistics >about it. > >Uri I tried the same positions during the Summer on Hiarcs, Rebel and Genius in order to see what is usual branching factor. All positions were searched by "brute force". Hiarcs and Rebel were the programs that I just bought in the Summer at 1999 and Genius 4 in 1995. Worst branching factor I found on Hiarcs. Later I found that extensions can mix-up everything. Also saying of "fixed depth" can be real joke once gain because of the extensions. With extensions logic can go at least one ply beyond what was said. The other factor, something that probably must be called "bounds", can also make you think that logic goes real brute force when it work otherwise. Bound can be done in such a way that it can "speed" the logic and make branching factor look like more attractive. In the same time you can find that this logic lie somehow when you will try to solve some mate containing positions. Leonid.
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.