Author: Vincent Diepeveen
Date: 05:30:09 08/08/00
Go up one level in this thread
On August 07, 2000 at 12:41:08, Dan Andersson wrote: >That's not proof, that's bulverism. So you find the restults as published in the ICCA real good proofs? Like the publication of Zugzwangs speedup at 256 processors, where the speedup gets compared to another person at 256 processors, however zugzwang using bigger hashtables (which gives a better parallel speedup already), and using normal alphabeta, and having a program that is dead slow on a single cpu. Is that the kind of acceptible proofs of getting a good speedup for parallellism? Or proving a new search algorithm at 24 test positions which all every ply will give a fail high (because there is a trick to find) and where about a third of the positions are real children tricks, like a mate in 2 or 3. That's proof of a search algorithm? How about using something with a good worst case behaviour? No one ever talks about worst case of MTD, which by accident happens in tough games.
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.