This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Scheduling of asm()
On Sun, Jan 06, 2002 at 11:49:10PM +0100, Falk Hueffner wrote:
> I'd end up with one builtin per opcode, which seems pretty silly,
> since unlike for example prefetches, those don't exist on other
> platforms.
We do support machine-specific builtins, and they are in use on
a number of targets.
> Rather than doing that I'd teach gcc to look at the opcode in the
> asm and find its scheduling class in a table.
I'd rather go with the builtins, really.
The looking up scheduling classes in a table doesn't mesh particulaly
well with what is on the dfa-scheduler-branch.
How do you handle cases in which different cpu implementations have
very different pipelines, and so one scheduling class name doesn't
map well between the implementations?
r~