Re: x86 global floating point register allocation

"Tzvetan Mikov" <ceco@jupiter.com>
8 Sep 2002 22:52:47 -0400

          From comp.compilers
Related articles
x86 global floating point register allocation sverker.is@home.se (Sverker Nilsson) (2002-08-24)
Re: x86 global floating point register allocation jacob@jacob.remcomp.fr (jacob navia) (2002-09-03)
Re: x86 global floating point register allocation jgd@cix.co.uk (John Dallman) (2002-09-03)
Re: x86 global floating point register allocation reig@tenerife.ics.uci.edu (Fermin Reig) (2002-09-03)
Re: x86 global floating point register allocation ceco@jupiter.com (Tzvetan Mikov) (2002-09-08)
| List of all articles for this month |
From: "Tzvetan Mikov" <ceco@jupiter.com>
Newsgroups: comp.compilers
Date: 8 Sep 2002 22:52:47 -0400
Organization: Concentric Internet Services
References: 02-08-087 02-09-007
Keywords: arithmetic
Posted-Date: 08 Sep 2002 22:52:47 EDT

"jacob navia" <jacob@jacob.remcomp.fr> wrote
> Another problem is that if your basic block contains any jumps out of
> the block, you will have quite a few possibilities of where the values
> are. [...]


> if (foo) {
> // some code
> }
> else {
> // another fp code
> }


Not to nitpick, but ... By common definition a basic block can't contain
jumps - it is a sequence of operations, terminated by a jump. Your example
actually has three separate basic blocks:


b1: if (!foo) goto b3; [goto b2]
b2: some code; goto b4
b3: another fp code; [goto b4]
b4:


-tzvetan

Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.