Author: Robert Hyatt
Date: 10:45:06 08/16/05
Go up one level in this thread
On August 16, 2005 at 11:57:25, Andreas Guettinger wrote: >On August 16, 2005 at 07:05:54, Robert Hyatt wrote: > >>On August 16, 2005 at 02:50:10, Andreas Guettinger wrote: >> >>>The question remains if other applications using posix threads (like Apache2, >>>mysql etc) have the same problems on dual-core AMD linux systems, which should >>>be the case if the error lies in the library. >>>Or was just the crafty pthreads implementation fishy. >>> >>>regrads >>>Andy >> >> >>I doubt most applications care about accumulated CPU time, for one thing. And >>there may well be a work-around that only works on linux, not sure. I know that >>the old code works fine on redhat 9 and earlier versions... > >Sorry, i didn't read your first post careful enough :) >I thought the scaling was bad. If it's only the timing, then I dont have to >worry. > >regards >Andy The problem was I used the bad timing data to compute the scaling inside Crafty. The main clue should have been the atrocious CPU utilization numbers. On the 8-processor opteron, I occasionally saw 6400% cpu utilization, which is not exactly reasonable. :)
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.