This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Loop unrolling
- To: amylaar at cygnus dot co dot uk, law at cygnus dot com
- Subject: Re: Loop unrolling
- From: mrs at wrs dot com (Mike Stump)
- Date: Fri, 12 Jun 1998 18:48:40 -0700
- Cc: branko dot cibej at hermes dot si, egcs at cygnus dot com, pfeifer at dbai dot tuwien dot ac dot at
> Date: Fri, 12 Jun 1998 16:09:22 -0600
> From: Jeffrey A Law <law@hurl.cygnus.com>
> An empty loop would be a loop with a body that consists of a counter
> increment and branch to the top of the loop (at the RTL level).
> Nothing more, nothing less.
> I don't think trying to describe what an empty loop is with C
> source is all that interesting since in the end it's going to be
> detected by the loop optimizer.
Well, this wasn't the result of the consersation when last we agreed
on what to do and what the doc meant and what the compiler should do.
The decision whether or not a loop could be removed was to be decided
by the source input tokens to the lexer. Not by the existing rtl
codes.
Oh course, the optimizer still has to examine the loop to figure out
if it can remove it but this is separate if you will...