Author: Tom Kerrigan
Date: 18:11:17 09/22/01
Go up one level in this thread
On September 22, 2001 at 09:58:55, Robert Hyatt wrote: >On September 22, 2001 at 04:10:46, Tom Kerrigan wrote: > >>On September 21, 2001 at 23:03:36, Robert Hyatt wrote: >> >>>On September 21, 2001 at 20:42:34, Eugene Nalimov wrote: >>> >>>>GCC uses CMOVs. MSVC does not. So you can see yourself that CMOVs only hurts :-) >>>> >>>>Eugene >>> >>>:) I assume you mean "hurts portability among processors" and not "hurts >>>speed?" :) >> >>Probably both, if you consider the P5 a processor that you might want to "port" >>to, and the P6's brain-dead implementation of CMOV. >> >>-Tom > > >I'm not sure what you mean by "brain dead". It does eliminate the branch >misprediction problem nicely. The idea came from the alpha. Yes, that's the idea for the instruction, not the implementation. (Notice that I did say implementation.) I can't remember the exact details at the moment, but CMOVs on the P6 take several cycles more than they "should" and it's usually faster to avoid them altogether. -Tom
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.