This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Register allocation problem?
- To: Peter Bergner <bergner at brule dot borg dot umn dot edu>
- Subject: Re: Register allocation problem?
- From: David Edelsohn <dje at watson dot ibm dot com>
- Date: Tue, 21 Aug 2001 13:22:05 -0400
- cc: Daniel Berlin <dan at cgsoftware dot com>, Lars Brinkhoff <lars at nocrew dot org>, gcc at gcc dot gnu dot org
>>>>> Peter Bergner writes:
Peter> Why is it again we run reload after the new allocator (I assume this
Peter> conversation occurred before I rejoined the mailing list)?
Reload primarily is necessary to handle some perverse restrictions
of some architectures. The difficulty is finding the balance between
leaving enough of reload in place for the difficult corner cases while not
having reload interfere with the correct register allocation and
materialization which the new register allocator has performed.
Rice and others are helping with the design and tuning.
Thanks, David