Author: Bruce Moreland
Date: 20:29:18 07/09/02
Go up one level in this thread
On July 09, 2002 at 22:55:02, Robert Hyatt wrote: >On July 09, 2002 at 20:31:51, Russell Reagan wrote: > >>Very interesting. I have a few questions. What happens when you continue to >>increase the "false match" rate? Surely it will start to collapse at some point. >>If it doesn't, then something must be broken. >> >>The other question that came to mind was about if the search was able to >>withstand many false matches, would it increase performance at all to use a >>32-bit hash key instead of a 64-bit hash key. I recall you speaking many times >>of your (and others) tests that demonstrated that 32-bit hash keys generated far >>more collisions, but now the question is, is that as bad of a thing as we >>thought? >> >>Russell > > >I will report more as I test. The obvious question is "how many false matches >can be tolerated?" My gut feeling is "zero" because I can print out a small >tree, and change _one_ score of my choosing, and change the entire PV and >score. But in reality, the search appears to be much more robust than I would >have guessed... The search has redundancy. You can break a node and it may not matter. The PV is a line of nodes, but the tree that generated it is more two-dimensional, for lack of a better term. An error has to be pretty bad to show up. The reason I brought this up is that I saw that Bob had developed a nice scheme for avoiding collisions, but he never actually showed that collisions were bad. I felt that this was a missing step. bruce
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.