This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [committed] Encourage use of VR5500 madd/macc instructions
- From: Richard Sandiford <rsandifo at redhat dot com>
- To: gcc-patches at gcc dot gnu dot org
- Date: Sat, 01 May 2004 13:57:31 +0100
- Subject: Re: [committed] Encourage use of VR5500 madd/macc instructions
- References: <87wu44o3xu.fsf@redhat.com>
Richard Sandiford <rsandifo@redhat.com> writes:
> + t1 = a * b
> + t2 = t1 + c * d
> + t3 = c * d
> + t4 = t3 - c * d
> +
> + t1 will have a higher priority and t2 and t3 will have a higher
...and this patch fixes the lame typos above. Tsk.
Committed to trunk.
Richard
* config/mips/mips.h (TUNE_MACC_CHAINS): Fix comment.
Index: config/mips/mips.h
===================================================================
RCS file: /cvs/gcc/gcc/gcc/config/mips/mips.h,v
retrieving revision 1.338
diff -u -p -F^\([(a-zA-Z0-9_]\|#define\) -r1.338 mips.h
--- config/mips/mips.h 30 Apr 2004 16:27:27 -0000 1.338
+++ config/mips/mips.h 1 May 2004 12:53:04 -0000
@@ -346,10 +346,10 @@ #define TUNE_SB1 (mip
t1 = a * b
t2 = t1 + c * d
- t3 = c * d
- t4 = t3 - c * d
+ t3 = e * f
+ t4 = t3 - g * h
- t1 will have a higher priority and t2 and t3 will have a higher
+ t1 will have a higher priority than t2 and t3 will have a higher
priority than t4. However, before reload, there is no dependence
between t1 and t3, and they can often have similar priorities.
The scheduler will then tend to prefer: