Author: Gerd Isenberg
Date: 00:09:16 07/18/03
Go up one level in this thread
On July 18, 2003 at 01:02:14, Walter Faxon wrote: >On July 17, 2003 at 19:27:28, Gerd Isenberg wrote: > >... >>the only reason nowadays to write assembler stuff is IMHO using MMX, SSE2 and >>special processor instructions, the compiler don't supply, like bsf and cmov for >>MSC. >... > > >Hi, Gerd. > >Viz. MS Visual C 6 and generation of the "cmov" (conditional move) >instruction, Matt Taylor informs me (from a couple of emails): > >----- >Pentium Pro, Pentium 2, and Pentium 3 are all the same design, but the >latter processors have been tweaked significantly. Targeting one >should effectively target them all in this case. > >Compiler option is /arch:SSE|SSE2 > >The bsf/bsr instructions are available as of VC 8. >----- > >If instead you're using the IDE this would be: > > Project Settings dialog > C/C++ tab > Category: Code Generation > Processor: Pentium Pro > >I guess this all means that one can reasonably hope that selecting for the >Pentium Pro should allow generation of cmov, though I would prefer an experiment >to verify this. You might try getting an assembly output for the code I posted >in message: http://www.talkchess.com/forums/1/message.html?306363 Hi Walter, I will try it during the next days. With IsiChess i already tried Pentium Pro settings for my K7 - not much difference, if i remember well, maybe a bit slower than Blend*. If the compiler is not able - one reason to use assembler for your routine... > >If that works I'll post it as a macro, additionally providing more >functionality. Tell me whether you want lsb=0 or lsb=63. lsb=0 ;-) Regards, Gerd > >-- Walter
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.