This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH, middle-end, i386]: Use signbit insn in copysing insn and implement signbit using fxam x87 insn
- From: Wolfgang Gellerich <gellerich at de dot ibm dot com>
- To: Uros Bizjak <ubizjak at gmail dot com>
- Cc: GCC Patches <gcc-patches at gcc dot gnu dot org>
- Date: Thu, 26 Jul 2007 17:22:24 +0200
- Subject: Re: [PATCH, middle-end, i386]: Use signbit insn in copysing insn and implement signbit using fxam x87 insn
Uros Bizjak <ubizjak@gmail.com> wrote on 21.07.2007 12:08:06:
> Hello!
>
> This patch implements signbit built-in using x87 fxam instruction for
> -mfpmath=387. Using this approach, partial memory stalls when using bit
> operations on FP value are avoided (see chapters 14.7 and 15.8 of
> "Pentium optimization guide" for further information on this topic).
>
> I also took this opportunity and optimize copysign expander when using
> abs/neg approach. Sign bit of the input operand can be determined using
> signbit insn if available, further reducing number of partial memory
> stalls. The patch also changes signbit handling a bit (to be more
> consistent with other builtins). s390 target (which is currently the
> only user of signbit optab) _could_ also be affected by or benefit from
> this patch.
This is fine for s390!
Regards, Wolfgang
---
Dr. Wolfgang Gellerich
IBM Deutschland Entwicklung GmbH
Schönaicher Strasse 220
71032 Böblingen, Germany
Tel. +49 / 7031 / 162598
gellerich@de.ibm.com
=======================
IBM Deutschland Entwicklung GmbH
Vorsitzender des Aufsichtsrats: Martin Jetter
Geschäftsführung: Herbert Kircher
Sitz der Gesellschaft: Böblingen
Registergericht: Amtsgericht Stuttgart, HRB 243294