This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] PowerPC extendsfdf2 MEM alternative
- From: Andrew Pinski <pinskia at physics dot uc dot edu>
- To: David Edelsohn <dje at watson dot ibm dot com>
- Cc: gcc-patches at gcc dot gnu dot org, Geoff Keating <geoffk at geoffk dot org>
- Date: Fri, 11 Mar 2005 13:43:20 -0500
- Subject: Re: [PATCH] PowerPC extendsfdf2 MEM alternative
- References: <200503111625.j2BGPTD30920@makai.watson.ibm.com>
On Mar 11, 2005, at 11:25 AM, David Edelsohn wrote:
PowerPC always stores floating point values in double precision
format. That means loading a single precision value implicitly
extends it
for use in a double precision instruction. I think that the PowerPC
extendsfdf2 pattern should advertise that it can extend from memory.
Geoff, does this seem reasonable to you? Any hidden gotchas?
I think this is a great idea, in fact I wish we could produce some
fp constants that are representable to be stored in 32bits, This will
speed up things in general and reduce the memory overhead. I know that
XLC does this already.
-- Pinski