This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] Add C99 FP_FAST_FMA{,F,L} macros to math.h
- From: Jakub Jelinek <jakub at redhat dot com>
- To: Michael Meissner <meissner at linux dot vnet dot ibm dot com>
- Cc: libc-alpha at sourceware dot org
- Date: Mon, 18 Oct 2010 17:42:24 +0200
- Subject: Re: [PATCH] Add C99 FP_FAST_FMA{,F,L} macros to math.h
- References: <20101015203651.GA14249@hungry-tiger.westford.ibm.com>
- Reply-to: Jakub Jelinek <jakub at redhat dot com>
On Fri, Oct 15, 2010 at 04:36:51PM -0400, Michael Meissner wrote:
> * sysdeps/powerpc/bits/mathdef.h (FP_FAST_FMA): Define, ppc as
> multiply/add.
> (FP_FAST_FMAF): Ditto.
s390{,x} should do the same and likewise ia64 (in that case even
#define FP_FAST_FMAL 1
). For other targets that have their own bits/mathdef.h, the bits/mathdef.h
hunks should be duplicated there too -
sysdeps/{i386,x86_64,sparc,sh/sh4}/bits/mathdef.h
(at least for i386 and x86_64, that is, sh probably don't
currently have an instruction for fma, for sparc I think SPARCVI/SPARCVII
have it, but whether Linux supports those chips or not is unclear to me,
DaveM might know).
Jakub