Author: Robert Hyatt
Date: 14:47:38 11/02/02
Go up one level in this thread
On November 02, 2002 at 14:30:09, Gian-Carlo Pascutto wrote: >On November 02, 2002 at 14:21:48, Eugene Nalimov wrote: > >>On November 02, 2002 at 13:37:41, Gian-Carlo Pascutto wrote: >> >>>On November 01, 2002 at 17:26:39, Eugene Nalimov wrote: >>> >>>>Net result: you can look at the numbers I posted, and you will definitely see >>>>that hyperthreading gives current Crafty, without any hyperthread-related >>>>modifications, double-digit improvement. >>> >>>Only in nodes per second, not in real speedup. >> >>Dual P4/2.4GHz, hyperthreading is on. >> >>mt=2: times are 42, 43, 39, 40, 38, average is 40. >>mt=4: times are 36, 31, 37, 29, 38, average is 34. >> >>No real speedup, are you sure? > >You reported 8% earlier on for the real speedup, which is not >'double-digit improvement' > >I don't really understand how mt=4 can give an additional speedup. >Is this the effect of 'pause' and slow memory? > >Is hyperthreading able to 'emulate' more than 2 cpus? > >-- >GCP He's running it on a dual. With SMT it looks like 4 cpus, which is what linux reports, for example. So mt=4 makes sense there.
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.