Author: Robert Hyatt
Date: 19:59:24 01/05/03
Go up one level in this thread
On January 05, 2003 at 22:50:18, Bruce Moreland wrote: >On January 05, 2003 at 22:35:45, Miguel A. Ballicora wrote: > >>Wayne is right. >> >>It was not on target. Vincent mentioned something supporting why he does not >>like bitboards. Big deal, he does not like them, so what? Russell's comment is >>uncalld for, it is "ad hominem" and totally out of the blue. Well, not out of >>the blue, out of "not so old" flame wars. >>And not, the response IMHO is not reasonable. I am really sick and tired of >>perpetual wars with little or no content. It makes it very difficult to follow >>the interesting part of the threads!!. This has been going on for so long that a >>member of CCC would have lots of trouble to distinguish who was first. >>Please, guys, control yourself. There is no need to start all over again. >>_Please_. > >Vincent answered my question, maybe. I was talking about move generation and >supporting stuff. > >I'd like mine to go faster, but realistically, what would happen if I made it >faster? Not a whole heck of a lot. > >Everyone who starts a chess program becomes an instant move generation fanatic, >because that's what you do first. There is a post here now (a fine post) about >making the fastest possible popcount, which is used by move generation. > >I ask how much this matters. > >bruce It is the classic question that comes from discussions about "Amdahl's Law" and parallel speedup. If you assume that the parallel part of the code can be sped up infinitely by infinite processors, what is the max performance increase you can get? With 90% of the code parallel, you can go 10 times faster an no more. The question you mention is exactly the same. If my eval was instant, I could go 2x faster. What would that be worth? Not enough for the work it would take to make my eval run in 0 time. :)
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.