Author: Paul Richards
Date: 09:57:48 07/02/99
Go up one level in this thread
On July 02, 1999 at 12:33:07, Peter Kappler wrote: > ><silly teenager comment snipped...> :-) That's wasn't directed at you, I just think that must be the target audience when I see a post that essentially says: HIARCS ROOOOOOOLLLZZZ!!!!. :) I won't mention any names though (tania). :) >I have read some of Bob Hyatt's comments on the matter, and I don't agree with >him. I suspect that too many people here take Bob's word as the gospel on all >matters. (note: I have the utmost respect for Dr. Hyatt, and this isn't >intended as an insult against him.) Well when it comes to Crafty internals his word is gospel, but you don't have to be a programmer to realize that there could be number of resource allocation issues, and that one program may have a different issue than another. Crafty makes assumptions about pondering, Fritz needs huge hash tables, etc. It's also clear that we just don't know what the issues are, which means there are an unknown number of variables. It's also common sense that programmers have little reason to design or test for this situation. They're too busy debugging the parallel code for the serious competition. All of this is avoided with two-machine testing, though even there one program may do better than another if you use Xeon processors instead of regular pentiums, etc. But when someone raves about how Hiarcs kicks the living crap out of Fritz on a single machine test in a dozen games, when the two machine testing with a larger number of games disagrees, it does more to show the problems of single machine testing than really say anything about the relative strength of the programs.
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.