Author: Robert Hyatt
Date: 18:04:35 07/11/00
Go up one level in this thread
On July 11, 2000 at 17:56:16, Landon Rabern wrote: >Can someone give me an idea of how much switching to PVS will cut down my tree >size. > >Also, in my q-search I have been cutting out all captures where >attackerValue>defenderValue. I made an exchange function that works and tried >using it to keep captures where attackerValue>defenderValue, but he exchange >returned a good value. The new version got 2 more positions correct on WAC, is >this what I should expect, or is there something wrong? > >Thanks, > >Landon W. Rabern If you have good move ordering, PVS should save you about 10%. But you have to average that over lots of positions, and you can _not_ use tactical test suites to measure this. PVS is at its best when the program locks on to one move at the root and sticks with it. Tactical suites don't do that and will make PVS look less efficient rather than more efficient.
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.