Author: Sune Fischer
Date: 14:24:04 10/02/01
Go up one level in this thread
On October 02, 2001 at 17:11:46, Sune Fischer wrote: >>But there is a way to do that same thing on one >>processor. Search a few nodes on move 1, then a few nodes on move 2, and >>bounce back and forth. You quickly find (again) that move 2 leads to a quick >>mate and use this to prune the other move. And that super-linear speedup goes >>away. Okay, I see you point, but what you have done here _is_ to paralize it !? In a way I think this is what they are talking about when they say a parallel can be faster than a serial. However, as I said before and as you have also written several times, this only works when move ordering is reversed. If you want to make the serial faster, reverse the move ordering and wupti, your serial is 2x as fast! -S.
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.