[Bug fortran/36821] New: Flush denormals to Zero Flag

[Bug fortran/36821] New: Flush denormals to Zero Flag

[J dot Hogg at rl dot ac dot uk]

On current x86 and x86_64 processors denormal numbers are handled as a special
case and very slowly. This can have a detrimental impact on the performance of
numerical programs. A common fix for this problem, at the cost of some
numerical stability, is to set the processor flag to flush denormals to zero.

This can be achieved in C with the following on SSE machines:
   _MM_SET_FLUSH_ZERO_MODE (_MM_FLUSH_ZERO_ON);

However it is not desirable to include C in a fortran program or library. Other
compiler tackle this by adding flags (ifort, NAG f95) or environment variables
(g95) which set this flush mode during program initialization. It would be good
if gfortran offered a similar facility (I have not been able to locate any
reference to one at the moment).


-- 
           Summary: Flush denormals to Zero Flag
           Product: gcc
           Version: 4.3.1
            Status: UNCONFIRMED
          Severity: enhancement
          Priority: P3
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: J dot Hogg at rl dot ac dot uk


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36821

[Bug fortran/36821] Flush denormals to Zero Flag

[rguenth at gcc dot gnu dot org]

------- Comment #1 from rguenth at gcc dot gnu dot org  2008-07-14 11:21 -------
Compiling with -ffast-math will automatically do that at program startup time
via linking against crtfastmath.o.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36821

[Bug fortran/36821] Flush denormals to Zero Flag

[J dot Hogg at rl dot ac dot uk]

------- Comment #2 from J dot Hogg at rl dot ac dot uk  2008-07-14 11:34 -------
-ffast-math also enables other (even more unsafe?) optimizations which may not
be desirable.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36821

[Bug target/36821] Flush denormals to Zero Flag

[nick2893 at hotmail dot com]

------- Comment #3 from nick2893 at hotmail dot com  2010-04-09 03:42 -------
I am looking for both Optimized like O2 and FTZ.  I have a large Fortran
program.  Without FTZ it gets weird crashes that should not be happening (no
problems on any other compilers!)  Without optimization it runs way too slow.

Any optimization overrides any attempts at FTZ, such as

_MM_SET_FLUSH_ZERO_MODE (_MM_FLUSH_ZERO_ON); 
-ffast-math

The tradeoff of "either FTZ or Optimized" I think will not work for me.


-- 

nick2893 at hotmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |nick2893 at hotmail dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36821

[Bug target/36821] Flush denormals to Zero Flag

[ubizjak at gmail dot com]

------- Comment #4 from ubizjak at gmail dot com  2010-04-09 06:26 -------
(In reply to comment #3)

> The tradeoff of "either FTZ or Optimized" I think will not work for me.

Find crtfastmath.o in gcc build directory and link it to your final executable.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36821

[Bug target/36821] Flush denormals to Zero Flag

[nick2893 at hotmail dot com]

------- Comment #5 from nick2893 at hotmail dot com  2010-04-09 15:30 -------
I'm still not getting FTZ ... unless I misunderstood your suggestion, here's
what I did after compiling with -O2:

root@nick-desktop:/home/test408# gfortran -o test4f
/usr/lib/gcc/i486-linux-gnu/4.4.0/crtfastmath.o TEST4SRC1.o TEST4SRC2.o
TEST4SRCU.o


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36821

[Bug target/36821] Flush denormals to Zero Flag

[crazylht at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=36821

Hongtao.liu <crazylht at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |crazylht at gmail dot com

--- Comment #8 from Hongtao.liu <crazylht at gmail dot com> ---
> Does -mdaz-ftz actually activate all of -ffast-math and/or
> -funsafe-math-optimizations?

No, -mdaz-ftz only set DAZ and FTZ bits.

[Bug target/36821] Flush denormals to Zero Flag

[kargl at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=36821

kargl at gcc dot gnu.org changed:

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

--- Comment #7 from kargl at gcc dot gnu.org ---
(In reply to CVS Commits from comment #6)
> The master branch has been updated by hongtao Liu <liuhongt@gcc.gnu.org>:
> 
> https://gcc.gnu.org/g:e54375d85d4aa5889869c2672158083b2106b623
> 
> commit r13-4891-ge54375d85d4aa5889869c2672158083b2106b623
> Author: liuhongt <hongtao.liu@intel.com>
> Date:   Mon Dec 12 15:43:58 2022 +0800
> 
>     x86: Add a new option -mdaz-ftz to enable FTZ and DAZ flags in MXCSR.
>     
>     if (mdaz-ftz)
>       link crtfastmath.o
>     else if ((Ofast || ffast-math || funsafe-math-optimizations)
>              && !shared && !mno-daz-ftz)
>       link crtfastmath.o
>     else
>       Don't link crtfastmath.o
>     
>     gcc/ChangeLog:
>     
>             PR target/55522
>             PR target/36821
>             * config/i386/gnu-user-common.h (GNU_USER_TARGET_MATHFILE_SPEC):
>             Link crtfastmath.o whenever -mdaz-ftz is specified. Don't link
>             crtfastmath.o when -share or -mno-daz-ftz is specified.
>             * config/i386/i386.opt (mdaz-ftz): New option.
>             * doc/invoke.texi (x86 options): Document mftz-daz.

Does -mdaz-ftz actually activate all of -ffast-math and/or
-funsafe-math-optimizations?

[Bug target/36821] Flush denormals to Zero Flag

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=36821

--- Comment #6 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by hongtao Liu <liuhongt@gcc.gnu.org>:

https://gcc.gnu.org/g:e54375d85d4aa5889869c2672158083b2106b623

commit r13-4891-ge54375d85d4aa5889869c2672158083b2106b623
Author: liuhongt <hongtao.liu@intel.com>
Date:   Mon Dec 12 15:43:58 2022 +0800

    x86: Add a new option -mdaz-ftz to enable FTZ and DAZ flags in MXCSR.

    if (mdaz-ftz)
      link crtfastmath.o
    else if ((Ofast || ffast-math || funsafe-math-optimizations)
             && !shared && !mno-daz-ftz)
      link crtfastmath.o
    else
      Don't link crtfastmath.o

    gcc/ChangeLog:

            PR target/55522
            PR target/36821
            * config/i386/gnu-user-common.h (GNU_USER_TARGET_MATHFILE_SPEC):
            Link crtfastmath.o whenever -mdaz-ftz is specified. Don't link
            crtfastmath.o when -share or -mno-daz-ftz is specified.
            * config/i386/i386.opt (mdaz-ftz): New option.
            * doc/invoke.texi (x86 options): Document mftz-daz.