This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Compilation-time suggestion #2
- From: Roger Sayle <roger at eyesopen dot com>
- To: gcc at gcc dot gnu dot org
- Cc: Scott Robert Ladd <coyote at coyotegulch dot com>
- Date: Mon, 19 Jan 2004 10:11:13 -0700 (MST)
- Subject: Compilation-time suggestion #2
Given I was in a brain-storming mood, I've just had a second thought.
[Twice in one day, its a good thing I'm sitting down!]
I was thinking that the compiler itself is always built just the
compiler flags, "-O2 -g", i.e. BOOT_CFLAGS. The thought occurred
to me that if it weren't for debugging issues we could use -O3 or
-fomit-frame-pointer. Then it struck me that this might be an
excellent case-study for Scott Robert Ladd's genetic algorithms.
Are there any optimizations that should be explicitly enabled or
disabled during bootstrap, to produce a faster compiler.
Does "-O2 -funroll-loops -ffast-math -fomit-frame-pointer" produce
a debuggable compiler that is any faster than the current default
build?
My apologies if this has already come up and been rejected before.
It certainly unifies the issues of debuggability and compilation-time.
Roger
--