Author: Ed Schröder
Date: 23:18:52 09/30/99
Go up one level in this thread
>Posted by Harald Faber on September 30, 1999 at 14:29:01: > >Hi all, >this is what Ed wrote: > >I know now there is not a (big) problem other than that I have noticed the >below list of weird things which are good to know for everybody who use >auto232. > >#1. In rare cases I have seen Rebel's NPS drop with a factor of 3 during >an auto232 match. Thus playing a match on a PII-450 the machine at some >moment in the auto232 match the PC Rebel was running suddenly behaved >as a PII-150 machine. The factor 3 is remarkable because it is an indication >it smells to hardware and not to software. > >Of course the outcome of these matches are worthless and I wouldn't exclude >the possibility such things also happen on the machines of Rebel opponents. >The BIG problem is: how do you recognize it? Perhaps it happens a lot more >because it is difficult to notice it. I think that for the future I will add >some code to test the reliability of the program after each game. >----------------------------------------------------------------------------- > >After Ed having published how to activate the autoplayer within Rebel10, I >started a testmatch against the new Nimzo7.32 early in the morning. >When I came back I saw Nimzo lead 2-0 and the 3rd game Nimzo was a piece up >for 2 pawns. I watched this game and couldn't believe my eyes. Rebel was >incredibly slow, only doing 24-30knps on my MMX166. Usually it is between 45 and >60knps so I stopped this game and played "normal" to check if this was a position in >which Rebel slows down because of deep threats and deep tactics like it happens >sometimes. But nothing. I saw the normal 50knps. >This of course explains why Rebel played some incredible bad moves in the 2 >games it lost as it didn't reach sufficient search depth. >(BTW this is a way to recognize Rebel having played much slower) I don't know if I should laugh or cry reading this. Are you sure there wasn't any swapping taking place because of a too big hash table? >My conclusion is that I will stop testing at least Rebel10 with auto232, As far as I know it doesn't occur much. Maybe you should do more testing? >and you guys who can't resist should carefully take a look at the games and the nps. And I wonder about the "other" programs. Maybe they are caught too? Ed
This page took 0.01 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.