Author: Gerd Isenberg
Date: 10:23:24 10/15/05
Go up one level in this thread
> 00021 83 fb 03 cmp ebx, 3 ;main.c:22.2
> 00024 b8 00 00 00 00 mov eax, 0 ;main.c:22.2
> 00029 0f 92 c0 setb al ;main.c:22.2
sub xxx, eax
hmm - also four instructions and one additional register.
I don't understand why compilers have so much trouble to do simple arithmetic
with booleans, specially compares with carry flag - using sbb or adc
instructions.
Does Intel also perform the div by reciprocal mul?
May you post the complete assembly of the routine?
>
>although generally I have found the VC++ 2003 Platform
>SDK compiler (which is a free download) produces executables
>as good overall as Intel.
Really?
>
>I tried the VC++ 2005 beta 1 compiler and got some very
>good performance numbers, but it was buggy at that stage.
>
>--Jon
Sometimes i am impressed about the optimization tricks of the compilers,
but sometimes simply shocked about their unability ;-)
Gerd
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.