This is the mail archive of the
mailing list for the GCC project.
Re: Graph coloring for register allocation?
- To: dberlin at redhat dot com
- Subject: Re: Graph coloring for register allocation?
- From: Nick Ing-Simmons <nik at tiuk dot ti dot com>
- Date: Mon, 22 Jan 2001 10:02:33 GMT
- Cc: Untitled <gcc at gcc dot gnu dot org>, Robert Dewar <dewar at gnat dot com>, Michael Matz <matzmich at cs dot tu-berlin dot de>, Michael Hayes <m dot hayes at elec dot canterbury dot ac dot nz>, Stan Shebs <shebs at apple dot com>
- Organization: via, but not speaking for : Texas Instruments Ltd.
- References: <Pine.LNX.firstname.lastname@example.org>
- Reply-To: Nick Ing-Simmons <nik at tiuk dot ti dot com>
Daniel Berlin <email@example.com> writes:
>> Just so we are all on the same page, everyone realizes that the new
>> register allocator will replace the regmove, the local-alloc, and the
>> global-alloc passes, right?
>Also, it gets rid of reload unless it's comment at the top of the file is
>lying (we can force).
>That means we replace about 25000 lines of code (24,775 according to wc)
>with like 1000 or 2000.
>If you can do something in 25x less code than you do now, it usually means
>your current way is a *bit* complex.
I suspect your new one will grow some ;-)
But I am very interested in the results - the existing register
allocator(s) have been a pain for DSP targets.
Nick Ing-Simmons <firstname.lastname@example.org>
Via, but not speaking for: Texas Instruments Ltd.