This is the mail archive of the
mailing list for the GCC project.
Re: cr logical insn implementation for rs6000
- To: David Edelsohn <dje at watson dot ibm dot com>
- Subject: Re: cr logical insn implementation for rs6000
- From: Michael Meissner <meissner at cygnus dot com>
- Date: Wed, 6 Sep 2000 18:07:43 -0400
- Cc: Geoff Keating <geoffk at cygnus dot com>, gcc-patches at gcc dot gnu dot org
- References: <email@example.com> <200009062122.RAA26028@mal-ach.watson.ibm.com>
On Wed, Sep 06, 2000 at 05:22:15PM -0400, David Edelsohn wrote:
> >>>>> Geoff Keating writes:
> >> What is the new DFA-based pipeline scheduler?
> Geoff> It's a more accurate way of describing instruction timings. It can
> Geoff> cope with things like function unit dependencies, limits on
> Geoff> instruction dispatch and/or completion, and so on. Of course, it
> Geoff> needs much more detailed information than the current scheduling model
> Geoff> to be done right.
> Is this going to augment the Haifa scheduler or replace it? The
> Haifa scheduler originally had provisions for much more detailed
> information but we were told that no additional scheduling information
> could be required. Issue rate itself was a hurdle.
Technically augment, since it uses haifa-sched.c. However, the tables and
information is disjoint (ie, most or none of the information for describing
scheduling in the current haifa scheduler is used with the DFA scheduler and
vicea versa). Before, people ask when it will be donated, I'll have to defer
to Vlad who is working on it.
Michael Meissner, Red Hat, Inc.
PMB 198, 174 Littleton Road #3, Westford, Massachusetts 01886, USA
Work: firstname.lastname@example.org phone: +1 978-486-9304
Non-work: email@example.com fax: +1 978-692-4482