[Bug target/51340] New: SH Target: Make -mfused-madd enabled by default

[Bug target/51340] New: SH Target: Make -mfused-madd enabled by default

[oleg.endo@t-online.de]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51340

             Bug #: 51340
           Summary: SH Target: Make -mfused-madd enabled by default
    Classification: Unclassified
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: enhancement
          Priority: P3
         Component: target
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: oleg.endo@t-online.de
                CC: kkojima@gcc.gnu.org
            Target: sh*-*-*


Currently the FMAC insn has to be enabled by specifying the -mfused-madd
option.  Is there any particular reason why this should not be enabled by
default for SH targets that support the FMAC insn?

[Bug target/51340] SH Target: Make -mfused-madd enabled by default

[kkojima at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51340

--- Comment #1 from Kazumoto Kojima <kkojima at gcc dot gnu.org> 2011-11-28 23:09:32 UTC ---
(In reply to comment #0)
> Is there any particular reason why this should not be enabled by
> default for SH targets that support the FMAC insn?

PR29100?

BTW, if SH fmac satisfies the semantics for fused multiplication and
add operation, the fmaf4 instruction pattern would be better now.

[Bug target/51340] SH Target: Make -mfused-madd enabled by default

[oleg.endo@t-online.de]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51340

--- Comment #2 from Oleg Endo <oleg.endo@t-online.de> 2011-12-19 01:29:12 UTC ---
(In reply to comment #1)
> (In reply to comment #0)
> > Is there any particular reason why this should not be enabled by
> > default for SH targets that support the FMAC insn?
> 
> PR29100?

Uhm, yes...
The title should have been "Enable -mfused-madd by -ffast-math"

> 
> BTW, if SH fmac satisfies the semantics for fused multiplication and
> add operation, the fmaf4 instruction pattern would be better now.

I don't know the exact semantics for the new patterns.  All I know is that
rounding is supposed to be done only once after the two operations.  This is
the case for the SH fmac insn.  Not sure whether this is enough though.

[Bug target/51340] SH Target: Make -mfused-madd enabled by default

[kkojima at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51340

--- Comment #3 from Kazumoto Kojima <kkojima at gcc dot gnu.org> 2011-12-28 22:31:27 UTC ---
(In reply to comment #2)
> Uhm, yes...
> The title should have been "Enable -mfused-madd by -ffast-math"

Do you mean something like this?

--- ORIG/trunk/gcc/config/sh/sh.c    2011-12-03 10:03:41.000000000 +0900
+++ trunk/gcc/config/sh/sh.c    2011-12-27 08:33:23.000000000 +0900
@@ -838,6 +838,11 @@ sh_option_override (void)
     align_functions = min_align;
     }

+  /* Default to use fmac insn when -ffast-math.  See PR target/29100.  */
+  if (global_options_set.x_TARGET_FMAC == 0
+      && fast_math_flags_set_p (&global_options)
+    TARGET_FMAC = 1;
+
   if (sh_fixed_range_str)
     sh_fix_range (sh_fixed_range_str);

> I don't know the exact semantics for the new patterns.  All I know is that
> rounding is supposed to be done only once after the two operations.  This is
> the case for the SH fmac insn.  Not sure whether this is enough though.

It seems that we can use the fma pattern, though it would be an another issue.

[Bug target/51340] SH Target: Make -mfused-madd enabled by default

[oleg.endo@t-online.de]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51340

--- Comment #4 from Oleg Endo <oleg.endo@t-online.de> 2011-12-29 00:02:40 UTC ---
(In reply to comment #3)
> (In reply to comment #2)
> > Uhm, yes...
> > The title should have been "Enable -mfused-madd by -ffast-math"
> 
> Do you mean something like this?
> 
> --- ORIG/trunk/gcc/config/sh/sh.c    2011-12-03 10:03:41.000000000 +0900
> +++ trunk/gcc/config/sh/sh.c    2011-12-27 08:33:23.000000000 +0900
> @@ -838,6 +838,11 @@ sh_option_override (void)
>      align_functions = min_align;
>      }
> 
> +  /* Default to use fmac insn when -ffast-math.  See PR target/29100.  */
> +  if (global_options_set.x_TARGET_FMAC == 0
> +      && fast_math_flags_set_p (&global_options)
> +    TARGET_FMAC = 1;
> +
>    if (sh_fixed_range_str)
>      sh_fix_range (sh_fixed_range_str);
> 

Yes, something like that.  Or maybe check flag_unsafe_math_optimizations, as it
is done for FSCA and FSRRA insns in sh.md.

> > I don't know the exact semantics for the new patterns.  All I know is that
> > rounding is supposed to be done only once after the two operations.  This is
> > the case for the SH fmac insn.  Not sure whether this is enough though.
> 
> It seems that we can use the fma pattern, though it would be an another issue.

Maybe when trunk is back to stage 1.

[Bug target/51340] SH Target: Make -mfused-madd enabled by default

[olegendo at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51340

Oleg Endo <olegendo at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |olegendo at gcc dot gnu.org

--- Comment #5 from Oleg Endo <olegendo at gcc dot gnu.org> 2012-05-28 22:11:31 UTC ---
(In reply to comment #4)
> > 
> > It seems that we can use the fma pattern, though it would be an another issue.
> 
> Maybe when trunk is back to stage 1.

Created PR 53511 for this.

[Bug target/51340] SH Target: Make -mfused-madd enabled by default

[olegendo at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51340

--- Comment #6 from Oleg Endo <olegendo at gcc dot gnu.org> 2012-05-29 20:56:02 UTC ---
Author: olegendo
Date: Tue May 29 20:55:58 2012
New Revision: 187988

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=187988
Log:
    PR target/51340
    * config/sh/sh.c (sh_option_override): Set TARGET_FMAC if
    flag_unsafe_math_optimizations is set.
    * doc/invoke.texi (SH Options): Add -mno-fused-madd description.
    Update description of -mfused-madd.

    PR target/51340
    * gcc.target/sh/pr51340-1.c: New.
    * gcc.target/sh/pr51340-2.c: New.
    * gcc.target/sh/pr51340-3.c: New.


Added:
    trunk/gcc/testsuite/gcc.target/sh/pr51340-1.c
    trunk/gcc/testsuite/gcc.target/sh/pr51340-2.c
    trunk/gcc/testsuite/gcc.target/sh/pr51340-3.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/sh/sh.c
    trunk/gcc/doc/invoke.texi
    trunk/gcc/testsuite/ChangeLog

[Bug target/51340] SH Target: Make -mfused-madd enabled by default

[olegendo at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51340

Oleg Endo <olegendo at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |FIXED

--- Comment #7 from Oleg Endo <olegendo at gcc dot gnu.org> 2012-05-29 21:01:30 UTC ---
-mfused-madd option is now automatically enabled when
-funsafe-math-optimizations is enabled.