Author: Robert Hyatt
Date: 07:59:42 05/12/04
Go up one level in this thread
On May 12, 2004 at 09:34:14, Anthony Cozzie wrote: >On May 12, 2004 at 04:57:35, Aaron Gordon wrote: > >>1.856 speedup here in iteration time. May want to check over my numbers, falling >>asleep at the keyboard trying to get all this done at this hour :P Whats weird >>is I got a pretty good speedup on the last two that had slowdown problems for >>you. Crafty was v19.6 compiled w/ the Intel C 5.0.1 compiler. >> >>I'll do the second 2cpu run tomorrow and post it along w/ this, for now, sleep! >>:) >> >>1cpu log: ftp://newageoc.com/pub/crafty/logs/1cpu.log >>2cpu log: ftp://newageoc.com/pub/crafty/logs/2cpu-a.log >> >> pos T T2 >> 1 55 83/1.51 >> 2 189 100/1.89 >> 3 235 171/1.37 >> 4 253 157/1.61 >> 5 123 65/1.79 >> 6 70 39/1.68 >> 7 128 67/1.91 >> 8 211 123/1.72 >> 9 64 31/2.06 >> 10 146 95/1.54 >> 11 226 117/1.93 >> 12 248 135/1.84 >> 13 99 54/1.83 >> 14 293 179/1.64 >> 15 147 83/1.77 >> 16 227 120/1.89 >> 17 76 46/1.65 >> 18 224 98/2.28 >> 19 183 77/2.38 >> 20 214 109/1.96 >> 21 90 40/2.25 >> 22 209 98/2.13 >> 23 250 156/1.60 >> 24 72 31/2.32 >>average SU 1.856 > > >1 out of 4 positions had a superlinear speedup? > >anthony It varies wildly in positions that are not "native" to the program it seems. My xeon times didn't do that much. The opteron did. Obviously this is _not_ a hardware issue. Which just shows how wildly the times can vary in a parallel search.
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.