Author: leonid
Date: 17:31:30 10/05/99
Go up one level in this thread
On October 03, 1999 at 09:07:08, blass uri wrote: >On October 03, 1999 at 08:26:51, leonid wrote: > >>On October 03, 1999 at 04:40:59, Ed Schröder wrote: >> >>>>Posted by leonid on October 02, 1999 at 08:23:35: >>>> >>>>Is really Rebel 10 raw speed (search by brute force)is higher that of >>>>Hiarcs 7.32? >>>> >>>>For the last few months I am trying to find exactly where the raw speed of >>>>my logic for positional search stays. I tryed few best games. My response >>>>was that my "branching factor" was not good. It went between bad, very bad and >>>>just normal. Problem is that you can't say for sure that the game that you >>>>comparer with do the "brute force search". During the last surprise I >>>>found that in Hiacrs to reach the "brute force" you must put its "selectivity" >>>>to zero. Now it is likely that Rebel raw speed is even higher that Hiacrs. >>>>Is this last finding only mirage or "final truth"? >>>> >>>>Thanks for response, >>>>Leonid. >>> >>>Unfortunately there is no "final truth", so you can stop searching :) >>> >>>Thinking (for example) 6 plies deep with Program_X can't be compared >>>with 6 plies deep for Program_Y. Main reason: different type of >>>extensions chess programs use. >>> >>>Best example is Deep Blue. DB's iteration depth is in the same league as >>>the micro's. It is said DB uses massive extensions which explains it all. >>> >>>It is impossible to compare chess programs on ply-depth. >>> >>>Ed Schroder >> >> >>I am not sure what you want to say with "extention" but my trying was very >>simple. > > The identical positions were tryed on Rebel and Hiacrs using same >>computer. Number of plys demanded were the same and from boths games it was >>asked to make search by "brute force". > >If you choose 6 plies brute force >It means that both programs were asked to see at least 6 plies forward but it >does not mean nothing about how many plies they can see practically in some >lines. > >For example if Hiarcs does extensions and sees some lines with 8 plies then of >course it has more work to do and may be slower. > >The point is that both programs do extensions and do not ignore everything after >more than 6 plies. > >Uri Desgracefully what you said sound to me sound like a sad reality. But how really raw speed of game engin could be compared with the next one? Maybe you have some idea that my mind miss completely. Leonid.
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.