This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: "fix" alpha register allocation
- From: Toon Moene <toon at moene dot indiv dot nluug dot nl>
- To: Richard Henderson <rth at redhat dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Thu, 15 Jan 2004 21:12:22 +0100
- Subject: Re: "fix" alpha register allocation
- Organization: Moene Computational Physics, Maartensdijk, The Netherlands
- References: <20040115015936.A23982@redhat.com>
Richard Henderson wrote:
I can imagine that this reordering might have the opposite effect
for some specific numeric-heavy test cases, but for the nonce,
I wouldn't worry. I've seen some horrible examples from our code - I
just never got around asking why it did this.
Imagine you have a loop which needs quite a few array pointers
(generated by strength reduction) and also quite a few floating point
registers to hold intermediate values. Because of the wrong ordering
you'll see a lot of int->fp->int shuffling ...
--
Toon Moene - mailto:toon@moene.indiv.nluug.nl - phoneto: +31 346 214290
Saturnushof 14, 3738 XG Maartensdijk, The Netherlands
Maintainer, GNU Fortran 77: http://gcc.gnu.org/onlinedocs/g77_news.html
GNU Fortran 95: http://gcc.gnu.org/fortran/ (under construction)