Author: Robert Hyatt
Date: 07:51:36 05/20/04
Go up one level in this thread
On May 20, 2004 at 10:19:39, Pablo Rodriguez wrote: > >>generate all ply-3 moves. >> >>That is a _huge_ difference from depth-first. >> >>You _really_ look like an idiot here... >> > > > First, he was looking good. Then he was lookin' better. Finally he was lookin' >_real_ good. No it has come to be that he is looking like an idiot here... >What's next? Lots of choices.. jackass dumbass moron fool idiot nut-case I get tired of typing before I run out of options... but... Anybody that tries to re-define well-known AI terms to suit his own agenda is a _real_ nut-case... You do realize that he has been caught in multiple outright lies here? He said he could not get access to my Ph.D. dissertation. Then after several of us explain "university microfilm" to him he claims to have ordered a copy that will take 6 weeks to arrive. The very next day he is quoting from my dissertation. He already had it. The he takes the title "A High-Performance Parallel Algorithm to Search Depth-First Game Trees" and now tries to claim that searching "depth-first" game trees is worthless as that is an algorithm _nobody_ would show up at a tournament using. Some history: 1. my speedup formula speedup = 1 + (NCPUS - 1) * .7 was posted in many articles here and on usenet news. Vincent claimed that he had "proofed" that it was wrong. I posted a bunch of opteron data and Martin did the analysis to show it was pretty accurate with the formula. Vincent said "nobody else can produce data showing the same speedup." He was right. Someone posted dual G5 data that was _better_ than my speedup on the opteron. Then he started "if Bob would take that stupid tree pointer out of the single-cpu version it would be 10% faster. He is not comparing the best single-cpu algorithm to the N-cpu algorithm. I pointed out that when I originally added the pointer it slowed me down by about 3%. Notice the attempts to discredit that which he can't beat. Repeatedly. 2. He claimed I wrote a JICCA article that claimed my speedup formula worked for any number of CPUs. Rolf quoted an old CCC post of mine where I _clearly_ limited it to 4 as that was all I had tested it on at the time. I personally later amended it to include 8 as by then I had 8cpu data. He could not give a citation for that JICCA article as there was no such article. The only parallel search article I have written in the JICCA was about Cray Blitz. I wrote one on "EPVS" years ago, then the DTS article in 1997 or so. Another false claim. 3. He wants to claim that depth-first search is not used in computer chess engines when _everybody_ uses it. He tries to re-define the term, in direct contradiction of every author that has published a book or paper describing alpha/beta minimax search. He tries this to attempt to claim that my dissertation used a search technique that was more favorable for parallel search but useless in real games. Even though the program used in my dissertation was none other than Cray Blitz, the program that has won _exactly_ two more world computer chess championship tournaments than his program has won. 4. He wants to discredit my DTS article because he doesn't like the speedup results when looking at his results. He wants to discredit my dissertation for the same reason. He wants to discredit crafty's parallel search and overall chess playing skill for the same reason. If he can't out-perform me, then he tries to discredit me. 5. Crafty has finished in first or tied for first in 3 of the 6 CCT events we have held. He has played in some of them. But he prefers to try to discredit CCT as an "amateur engine event" since he can't seem to win one himself. There is a pattern here if you can pick it out. :) I show up at events, and win lose or draw, I accept the result and keep working. He tries to discredit the results and keep whining and making excuses as for why he can't do better. And he ends up looking like a complete fool each and every time. Over and over. But like that energizer bunny, he keeps on repeating the same mistake over and over and over and over and ..... > >> >> >> >>>But it definitely is depth limited. >>> >>>>-- >>>>James
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.