Author: Robert Hyatt
Date: 08:19:08 11/21/01
Go up one level in this thread
On November 21, 2001 at 11:05:00, Slater Wold wrote: > >As I have found positions where the NPS search is 2.5x faster, but it solves the >solution in 4x faster than a single cpu. > >Dann and I had this "super" linear discussion before. > >Seems like it would even out, eventually. But like I said, I believe you. And >I'll do it to solution now. (But of course, I'll still look at the NPS!) :) > First, two cpus is going to have a _hard_ time searching 2.5x the raw nodes per second. I have no idea how that might happen, unless there is a bug in the node-counting that sometimes counts nodes twice. Second, "super-linear" can happen on occasional positions. But as you said, it will average out over multiple positions so that the speedup simply can not be >2.0 for two processors on average. I was in the middle of the super-linear speedup discussion. I hope it stays "at rest" now. :) I have seen several cases of spectacular speedups, but then I have also seen an equal number of horrible speedups. Bruce once sent me one that produced a particularly ugly result on Crafty, But I can't seem to locate the thing at present... In any case, NPS is kind of like engine RPM. It should increase linearly with the number of processors, assuming the parallel algorithm is good at keeping both cpus busy all the time and doesn't have one (or more) sitting around waiting excessively. But RPM has nothing to do with vehicle speed, because of losses along the drive train. The MPH value (time to solution) is the thing that wins races (or games).
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.