Author: chandler yergin
Date: 01:55:06 11/24/05
Go up one level in this thread
On November 23, 2005 at 14:18:38, Uri Blass wrote: >On November 23, 2005 at 14:15:19, chandler yergin wrote: > >>On November 23, 2005 at 06:33:07, Jarkko Pesonen wrote: >> >>>Could it be hashing thing regarding that game history has to be played >>>before this behaviour presents itself. >> >>Hash is cleared every time a move is made on the Board. >>This becomes the Root Position, and the calculation & evaluation begins again. > >This is correct for Fritz5.32 but not correct for most engines(Fritz6 and later >versions do not clear hash). > >Uri Yes they do! Of course they do. "Hash is cleared every time a move is made on the Board." This becomes the new Root Position, and the calculation & evaluation begins again."The Program evaluates e-v-e-r-y Legal move in every position on the Board, Ranks them in order from Best to worst based on the feedback from the Alpa Beta Algorithm. The Engine makes the best move F-o-u-n-d at the time the software tells it to move. We've been over this. Engines don't have intelligence or plan or make blunders! They 'evaluate' that's all they do! It's not the Hardware either. It's the d-e-p-t-h of Search. Whoever gets there the quickest wins the race. The problem with Computers is finding the 'best' move in the alloted time it has. It may only be able to search for example 14/35 ply when it has to move. It's PV eval at the time may be (-2.50) The Human sees one move deeper by knowledge and plays the right move. At 15/40 Ply the PV Eval may change to (+- 3.0) We saw this in Pom's game against Fritz. chan
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.