Re: Interaction between optimizer and inline asm in gcc?

loewis@informatik.hu-berlin.de (Martin v. =?iso-8859-1?q?L=F6wis?=)
23 May 2002 01:23:47 -0400

          From comp.compilers

Related articles
Interaction between optimizer and inline asm in gcc? colohan+@cs.cmu.edu (Christopher Brian Colohan) (2002-05-17)
Re: Interaction between optimizer and inline asm in gcc? loewis@informatik.hu-berlin.de (2002-05-23)
Re: Interaction between optimizer and inline asm in gcc? sam@zoy.org (2002-05-23)
Re: Interaction between optimizer and inline asm in gcc? journeyman@compilerguru.com (2002-05-23)
Re: Interaction between optimizer and inline asm in gcc? kg_russell@yahoo.co.uk (2002-05-23)
| List of all articles for this month |
From: loewis@informatik.hu-berlin.de (Martin v. =?iso-8859-1?q?L=F6wis?=)
Newsgroups: comp.compilers
Date: 23 May 2002 01:23:47 -0400
Organization: Humboldt University Berlin, Department of Computer Science
References: 02-05-092
Keywords: GCC, optimize
Posted-Date: 23 May 2002 01:23:47 EDT

Christopher Brian Colohan <colohan+@cs.cmu.edu> writes:


> I was wondering if anyone has looked into the impact of doing this on
> gcc's optimizer: in other words, how conservatively does gcc treat
> inline asm statements? For example, would inserting such a statement
> inhibit loop unrolling?


The processing of asm statements is completely controlled through the
specification of constraints (i.e. inputs, outputs, clobbers). The
compiler will apply loop unrolling. It will even remove asm statements
if they have no effect - this can happen the compiler finds that the
outputs of the asm statement are not used. Likewise, the compiler may
switch execution order to process the asm as soon as all inputs are
available, until just before the first output is needed (use volatile
asm if you want to execute the asm at exactly point where you place
it).


Regards,
Martin


Post a followup to this message

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