Author: Robert Hyatt
Date: 14:41:29 10/04/01
Go up one level in this thread
On October 04, 2001 at 17:08:28, Uri Blass wrote: >On October 04, 2001 at 14:49:37, Robert Hyatt wrote: > >>On October 04, 2001 at 07:09:30, Uri Blass wrote: >> >>>On October 04, 2001 at 06:46:48, Sune Fischer wrote: >>> >>>>On October 04, 2001 at 04:56:08, Uri Blass wrote: >>>> >>>>>The difference is that the other side(I and Bruce) do not say that they know >>>>>that there is a super linear improvement but that we cannot say that it is >>>>>impossible with the known algorithm of today. >>>>> >>>>>I said that the only way is to investigate the problem by test positions in >>>>>order to see if programs can get a super linear improvement from 2 processors. >>>> >>>>But that is not the only way, you can also use logic. >>>>This is like explaning to an inventor why he can't make a perpetuum mobile >>>>machine. If he doesn't understand the laws of physics, does not know of energy >>>>conservation, then he will keep arguing till the day he die, that "we can not >>>>know for certain until we have tried everything". >>>> >>>>We _do_ know, there is proof and Bob has outlined it several times, but if you >>>>won't listen or understand, then we have a communication problem. >>>> >>>>-S. >>> >>>There is no proof when we do not know the source code of Deep fritz. >> >> >> >>Please go find _any_ book that discusses "the analysis of parallel algorithms" >>somewhere in it. You will find a very detailed discussion of this stuff with >>mathematical analysis of why it is impossible. It has nothing to do with a >>specific algorithm. >> >>It has _everything_ to do with the simple proof of the statement "A two-tape >>turing machine has no more computational power than a one-tape turing machine." >> >>Once you read and understand that, there is no need to discuss the concept >>further. IE I _know_ there is no way to do a sort in less than O(N) >>complexity, because It takes N cycles just to read the input. And in reality, >>I know that there is no way to do a sort in < o(N log N) cycles. This has >>_also_ been proven beyond a shadow of a doubt. To say "that isn't true because >>we have not tried all possible sort algorithms" won't fly. >> >> >> >>> >>>There is an agreement that a poor sequential algorithm can be more than 2 times >>>slower than a good parallel algorithm with 2 procesors. >> >>Only in special cases... >> >>> >>>We do not know that the algorithm of Deep Fritz for one processor is not poor >>> >> >>Who cares. It is the _same_ algorithm used on 2 processors. _exactly_ the >>same. Same alpha/beta algorithm. Same evaluation. Same move ordering. Etc.. > >Same move ordering does not seems to me to be truth. > >Uri Same move ordering != same search ordering. I use the _exact_ same move ordering algorithms in the serial or parallel versions of Crafty. The moves may well be searched in a different order due to an occasional hash hit from another thread. But the basic ordering strategy is unchanged. Which means that any such hash hit is based on serendipity. And when you depend on random chance to help you out, you can also depend on it to hurt you just as frequently. And that super-linear speedup goes right out the window on other cases...
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.