[Bug fortran/32049] New: Support on x86_64 also kind=16

[Bug fortran/32049] New: Support on x86_64 also kind=16

[burnus at gcc dot gnu dot org]

It would be great if kind=16 could also be supported on x86_64 besides the
kinds 4, 8 and 10. With this some legacy code could be run and we would catch
up with ifort, which support kind 4, 8 and 16. (16 with software emulation.)

Problems:
- software fp is still a bit incomplete
- Supporting kind = 10 and kind = 16 at the same time is not supported by
libgfortran.

For the current software floating point support, see:
http://gcc.gnu.org/ml/gcc-patches/2007-05/msg01490.html
http://gcc.gnu.org/ml/gcc-patches/2007-04/msg00728.html

Support for __float128 (TFmode) IEEE quad type is
available via soft-fp library on x86_64 targets.  The
support includes basic arithmetic operations (addition, subtraction,
negation, multiplication and division of __float128</code>
values), full set of IEEE comparisons between <code>__float128</code>
values, conversions to and from <code>float</code>, <code>double</code>
and <code>long double</code> floating point types, as well as
conversions to and from <code>signed</code> or
unsigned integer, signed or unsigned long integer and signed or
unsigned quad integer (TImode) types.  Additionally, all
operations generate full set of IEEE exceptions and support full set
of IEEE rounding modes.


-- 
           Summary: Support on x86_64 also kind=16
           Product: gcc
           Version: 4.3.0
            Status: UNCONFIRMED
          Severity: enhancement
          Priority: P3
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: burnus at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32049

[Bug fortran/32049] Support on x86_64 also kind=16

[fxcoudert at gcc dot gnu dot org]

------- Comment #1 from fxcoudert at gcc dot gnu dot org  2007-05-31 11:56 -------
To have decent support for __float128 in gfortran, we need to provide a
__float128 math library. Do you happen to know any?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32049

[Bug fortran/32049] Support on x86_64 also kind=16

[ubizjak at gmail dot com]

------- Comment #2 from ubizjak at gmail dot com  2007-05-31 21:44 -------
(In reply to comment #1)
> To have decent support for __float128 in gfortran, we need to provide a
> __float128 math library. Do you happen to know any?

http://sourceware.org/cgi-bin/cvsweb.cgi/libc/sysdeps/ieee754/?cvsroot=glibc

Look into ldbl-128 directory...

BTW: What exactly is incomplete with software fp?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32049

[Bug fortran/32049] Support on x86_64 also kind=16

[jvdelisle at gcc dot gnu dot org]

------- Comment #3 from jvdelisle at gcc dot gnu dot org  2007-06-01 00:53 -------
FX, this library might do the trick.  I have x86-64 machine here to help on
this.

What about rolling our own with mpfr?  Or is that too difficult.  :)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32049

[Bug fortran/32049] Support on x86_64 also kind=16

[kargl at gcc dot gnu dot org]

------- Comment #4 from kargl at gcc dot gnu dot org  2007-06-01 02:52 -------
(In reply to comment #3)
> 
> What about rolling our own with mpfr?  Or is that too difficult.  :)
> 

It could be done with mpfr, but note that at the moment gfortran
will provide either REAL(10) or REAL(16).  This has to do with
mapping Fortran types to equivalent C long double.

If you really want to get interesting, GCC has libdecnumber.
We could offer a base 10 REAL(X) where X is 42.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32049

[Bug fortran/32049] Support on x86_64 also kind=16

[jvdelisle at gcc dot gnu dot org]

------- Comment #5 from jvdelisle at gcc dot gnu dot org  2007-06-01 04:06 -------
It seems then the first step is to support both real 10 and real 16
simultaneously if available. Then implement the soft library for real 16 


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32049

[Bug fortran/32049] Support on x86_64 also kind=16

[fxcoudert at gcc dot gnu dot org]

------- Comment #6 from fxcoudert at gcc dot gnu dot org  2007-06-02 07:37 -------
(In reply to comment #2)
> http://sourceware.org/cgi-bin/cvsweb.cgi/libc/sysdeps/ieee754/?cvsroot=glibc
> 
> Look into ldbl-128 directory...

I don't think that's built into glibc on x86_64, is it? And, as far as I know,
politics prevents us from integrating parts of glibc into gcc (the libgcc-math
issue). What do you think is the best route for us? Can we convince the glibc
maintainers to build and expose that library?


-- 

fxcoudert at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |fxcoudert at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2007-06-02 07:37:11
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32049

[Bug fortran/32049] Support on x86_64 also kind=16

[burnus at gcc dot gnu dot org]

------- Comment #7 from burnus at gcc dot gnu dot org  2008-08-26 06:11 -------
Some initial patch: http://gcc.gnu.org/ml/fortran/2008-08/msg00238.html


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32049

[Bug fortran/32049] Support on x86_64 also kind=16

[fxcoudert at gcc dot gnu dot org]

-- 

fxcoudert at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |fxcoudert at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2007-06-02 07:37:11         |2010-06-08 20:42:25
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32049

[Bug fortran/32049] Support on x86_64 also kind=16

[fxcoudert at gcc dot gnu dot org]

------- Comment #8 from fxcoudert at gcc dot gnu dot org  2010-08-27 21:24 -------
Subject: Bug 32049

Author: fxcoudert
Date: Fri Aug 27 21:24:13 2010
New Revision: 163597

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=163597
Log:
        PR fortran/32049

        * gfortran.h (gfc_real_info): Add c_float128 field.
        * mathbuiltins.def: Indicate which builtins are const.
        * trans-types.h (float128_type_node, complex_float128_type_node,
        gfc_real16_is_float128): New variables.
        * trans-types.c (float128_type_node, complex_float128_type_node,
        gfc_real16_is_float128): New variables.
        (gfc_init_kinds): Allow TFmode.
        (gfc_build_real_type): Mark __float128 types as such.
        (gfc_init_types): Initialize float128_type_node and
        complex_float128_type_node
        * f95-lang.c (gfc_init_builtin_functions): Adjust for new
        argument of OTHER_BUILTIN macro.
        * trans-intrinsic.c (gfc_intrinsic_map_t): Likewise.
        (builtin_decl_for_precision): Special case for __float128.
        (builtin_decl_for_float_kind): Likewise.
        (define_quad_builtin): New function.
        (gfc_build_intrinsic_lib_fndecls): Create all __float128
        library decls if necessary. Store them in the real16_decl and
        complex16_decl builtin map fields.
        (gfc_get_intrinsic_lib_fndecl): Handle q-suffixed __float128
        library function names.

        * gfortran.dg/random_seed_1.f90: Adjust test.
        * gfortran.dg/float128_1.f90: New test.

Added:
    trunk/gcc/testsuite/gfortran.dg/float128_1.f90
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/f95-lang.c
    trunk/gcc/fortran/gfortran.h
    trunk/gcc/fortran/mathbuiltins.def
    trunk/gcc/fortran/trans-intrinsic.c
    trunk/gcc/fortran/trans-types.c
    trunk/gcc/fortran/trans-types.h
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/gfortran.dg/random_seed_1.f90


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32049

[Bug fortran/32049] Support on x86_64 also kind=16

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32049

Tobias Burnus <burnus at gcc dot gnu.org> changed:

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

--- Comment #12 from Tobias Burnus <burnus at gcc dot gnu.org> 2010-11-16 21:31:25 UTC ---
Hooray! REAL(16) is now available with GCC 4.6!

Close as fixed!

[Bug fortran/32049] Support on x86_64 also kind=16

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32049

--- Comment #11 from Tobias Burnus <burnus at gcc dot gnu.org> 2010-11-16 21:23:28 UTC ---
Author: burnus
Date: Tue Nov 16 21:23:19 2010
New Revision: 166825

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=166825
Log:
/
2010-11-13  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>
            Tobias Burnus  <burnus@net-b.de>

        PR fortran/32049
        * Makefile.def: Add libquadmath; build it with language=fortran.
        * configure.ac: Add libquadmath.
        * Makefile.tpl: Handle multiple libs in check-[+language+].
        * Makefile.in: Regenerate.
        * configure: Regenerate.

libquadmath/
2010-11-13  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>
            Tobias Burnus  <burnus@net-b.de>

        PR fortran/32049
        Initial implementation and checkin.

gcc/fortran/
2010-11-13  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>
            Tobias Burnus  <burnus@net-b.de>

        PR fortran/32049
        * gfortranspec.c (find_spec_file): New function.
        (lang_specific_driver): Try to find .spec file and use it.
        * trans-io.c (iocall): Define
        * IOCALL_X_REAL128/COMPLEX128(,write).
        (gfc_build_io_library_fndecls): Build decl for __float128 I/O.
        (transfer_expr): Call __float128 I/O functions.
        * trans-types.c (gfc_init_kinds): Allow kind-16 belonging
        to __float128.

gcc/testsuite/
2010-11-13  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>
            Tobias Burnus  <burnus@net-b.de>

        PR fortran/32049
        * gfortran.dg/quad_1.f90: New.
        * lib/gcc-defs.exp (gcc-set-multilib-library-path): Use also
        compiler arguments.
        * lib/gfortran.exp (gfortran_link_flags): Add libquadmath to
        library search path; call gcc-set-multilib-library-path with
        arguments such that libgfortran.spec is found.
        (gfortran_init): Add path for libgfortran.spec to
GFORTRAN_UNDER_TEST.

libgomp/
2010-11-13  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>
            Tobias Burnus  <burnus@net-b.de>

        PR fortran/32049
        * configure.ac: 
        * configure: Regenerate.

libgfortran/
2010-11-13  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>
            Tobias Burnus  <burnus@net-b.de>

        PR fortran/32049
        * Makefile.am: Add missing pow_r16_i4.c, add transfer128.c,
        link libquadmath, if used.
        * acinclude.m4 (LIBGFOR_CHECK_FLOAT128): Add.
        * configure.ac: Use it, touch spec file.
        * gfortran.map: Add pow_r16_i4 and
        transfer_(real,complex)128(,write) functions.
        * intrinsics/cshift0.c (cshift0): Handle __float128 type.
        * intrinsics/erfc_scaled_inc.c: Ditto.
        * intrinsics/pack_generic.c (pack): Ditto
        * intrinsics/spread_generic.c (spread): Ditto.
        * intrinsics/unpack_generic.c (unpack1): Ditto.
        * io/read.c (convert_real): Ditto.
        * io/transfer.c: Update comments.
        * io/transfer128.c: New file.
        * io/write_float.def (write_float): Handle __float128 type.
        * libgfortran.h: #include quadmath_weak.h, define __builtin_infq
        and nanq.
        * m4/mtype.m4: Handle __float128 type.
        * runtime/in_pack_generic.c (internal_pack): Ditto.
        * runtime/in_unpack_generic.c (internal_unpack): Ditto.
        * kinds-override.h: New file.
        * libgfortran.spec.in: Ditto.
        * generated/pow_r16_i4.c: Generated.
        * Makefile.in: Regenerate.
        * configure: Regenerate.
        * config.h: Regenerate.
        * bessel_r10.c: Regenerate.
        * bessel_r16.c: Regenerate.
        * bessel_r4.c: Regenerate.
        * bessel_r8.c: Regenerate.
        * exponent_r16.c: Regenerate.
        * fraction_r16.c: Regenerate.
        * nearest_r16.c: Regenerate.
        * norm2_r10.c: Regenerate.
        * norm2_r16.c: Regenerate.
        * norm2_r4.c: Regenerate.
        * norm2_r8.c: Regenerate.
        * rrspacing_r16.c: Regenerate.
        * set_exponent_r16.c: Regenerate.
        * spacing_r16.c: Regenerate.


Added:
    trunk/gcc/testsuite/gfortran.dg/quad_1.f90
    trunk/libgfortran/generated/pow_r16_i4.c
    trunk/libgfortran/io/transfer128.c
    trunk/libgfortran/kinds-override.h
    trunk/libgfortran/libgfortran.spec.in
    trunk/libquadmath/
    trunk/libquadmath/COPYING.LIB
    trunk/libquadmath/ChangeLog
    trunk/libquadmath/Makefile.am
    trunk/libquadmath/Makefile.in
    trunk/libquadmath/acinclude.m4
    trunk/libquadmath/aclocal.m4
    trunk/libquadmath/config.h.in
    trunk/libquadmath/configure
    trunk/libquadmath/configure.ac
    trunk/libquadmath/gdtoa/
    trunk/libquadmath/gdtoa/README.gdtoa
    trunk/libquadmath/gdtoa/arith.h
    trunk/libquadmath/gdtoa/dmisc.c
    trunk/libquadmath/gdtoa/g_Qfmt.c
    trunk/libquadmath/gdtoa/g__fmt.c
    trunk/libquadmath/gdtoa/gd_qnan.h
    trunk/libquadmath/gdtoa/gdtoa.c
    trunk/libquadmath/gdtoa/gdtoa.h
    trunk/libquadmath/gdtoa/gdtoa_fltrnds.h
    trunk/libquadmath/gdtoa/gdtoaimp.h
    trunk/libquadmath/gdtoa/gethex.c
    trunk/libquadmath/gdtoa/gmisc.c
    trunk/libquadmath/gdtoa/hd_init.c
    trunk/libquadmath/gdtoa/hexnan.c
    trunk/libquadmath/gdtoa/makefile
    trunk/libquadmath/gdtoa/misc.c
    trunk/libquadmath/gdtoa/smisc.c
    trunk/libquadmath/gdtoa/strtodg.c
    trunk/libquadmath/gdtoa/strtopQ.c
    trunk/libquadmath/gdtoa/sum.c
    trunk/libquadmath/gdtoa/ulp.c
    trunk/libquadmath/libtool-version
    trunk/libquadmath/math/
    trunk/libquadmath/math/acoshq.c
    trunk/libquadmath/math/acosq.c
    trunk/libquadmath/math/asinhq.c
    trunk/libquadmath/math/asinq.c
    trunk/libquadmath/math/atan2q.c
    trunk/libquadmath/math/atanhq.c
    trunk/libquadmath/math/atanq.c
    trunk/libquadmath/math/cbrtq.c
    trunk/libquadmath/math/ceilq.c
    trunk/libquadmath/math/complex.c
    trunk/libquadmath/math/copysignq.c
    trunk/libquadmath/math/coshq.c
    trunk/libquadmath/math/cosq.c
    trunk/libquadmath/math/cosq_kernel.c
    trunk/libquadmath/math/erfq.c
    trunk/libquadmath/math/expm1q.c
    trunk/libquadmath/math/expq.c
    trunk/libquadmath/math/fabsq.c
    trunk/libquadmath/math/finiteq.c
    trunk/libquadmath/math/floorq.c
    trunk/libquadmath/math/fmodq.c
    trunk/libquadmath/math/frexpq.c
    trunk/libquadmath/math/hypotq.c
    trunk/libquadmath/math/isinfq.c
    trunk/libquadmath/math/isnanq.c
    trunk/libquadmath/math/j0q.c
    trunk/libquadmath/math/j1q.c
    trunk/libquadmath/math/jnq.c
    trunk/libquadmath/math/ldexpq.c
    trunk/libquadmath/math/lgammaq.c
    trunk/libquadmath/math/llroundq.c
    trunk/libquadmath/math/log10q.c
    trunk/libquadmath/math/log1pq.c
    trunk/libquadmath/math/logq.c
    trunk/libquadmath/math/lroundq.c
    trunk/libquadmath/math/modfq.c
    trunk/libquadmath/math/nanq.c
    trunk/libquadmath/math/nextafterq.c
    trunk/libquadmath/math/powq.c
    trunk/libquadmath/math/rem_pio2q.c
    trunk/libquadmath/math/remainderq.c
    trunk/libquadmath/math/rintq.c
    trunk/libquadmath/math/roundq.c
    trunk/libquadmath/math/scalblnq.c
    trunk/libquadmath/math/scalbnq.c
    trunk/libquadmath/math/signbitq.c
    trunk/libquadmath/math/sincos_table.c
    trunk/libquadmath/math/sincosq.c
    trunk/libquadmath/math/sincosq_kernel.c
    trunk/libquadmath/math/sinhq.c
    trunk/libquadmath/math/sinq.c
    trunk/libquadmath/math/sinq_kernel.c
    trunk/libquadmath/math/sqrtq.c
    trunk/libquadmath/math/tanhq.c
    trunk/libquadmath/math/tanq.c
    trunk/libquadmath/math/tgammaq.c
    trunk/libquadmath/math/truncq.c
    trunk/libquadmath/quadmath-imp.h
    trunk/libquadmath/quadmath.h
    trunk/libquadmath/quadmath.map
    trunk/libquadmath/quadmath_io.c
    trunk/libquadmath/quadmath_weak.h
Modified:
    trunk/ChangeLog
    trunk/Makefile.def
    trunk/Makefile.in
    trunk/Makefile.tpl
    trunk/configure
    trunk/configure.ac
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/gfortranspec.c
    trunk/gcc/fortran/trans-io.c
    trunk/gcc/fortran/trans-types.c
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/lib/gcc-defs.exp
    trunk/gcc/testsuite/lib/gfortran.exp
    trunk/libgfortran/ChangeLog
    trunk/libgfortran/Makefile.am
    trunk/libgfortran/Makefile.in
    trunk/libgfortran/acinclude.m4
    trunk/libgfortran/config.h.in
    trunk/libgfortran/configure
    trunk/libgfortran/configure.ac
    trunk/libgfortran/generated/bessel_r10.c
    trunk/libgfortran/generated/bessel_r16.c
    trunk/libgfortran/generated/bessel_r4.c
    trunk/libgfortran/generated/bessel_r8.c
    trunk/libgfortran/generated/exponent_r16.c
    trunk/libgfortran/generated/fraction_r16.c
    trunk/libgfortran/generated/nearest_r16.c
    trunk/libgfortran/generated/norm2_r10.c
    trunk/libgfortran/generated/norm2_r16.c
    trunk/libgfortran/generated/norm2_r4.c
    trunk/libgfortran/generated/norm2_r8.c
    trunk/libgfortran/generated/rrspacing_r16.c
    trunk/libgfortran/generated/set_exponent_r16.c
    trunk/libgfortran/generated/spacing_r16.c
    trunk/libgfortran/gfortran.map
    trunk/libgfortran/intrinsics/cshift0.c
    trunk/libgfortran/intrinsics/erfc_scaled_inc.c
    trunk/libgfortran/intrinsics/pack_generic.c
    trunk/libgfortran/intrinsics/spread_generic.c
    trunk/libgfortran/intrinsics/unpack_generic.c
    trunk/libgfortran/io/read.c
    trunk/libgfortran/io/transfer.c
    trunk/libgfortran/io/write.c
    trunk/libgfortran/io/write_float.def
    trunk/libgfortran/libgfortran.h
    trunk/libgfortran/m4/mtype.m4
    trunk/libgfortran/mk-kinds-h.sh
    trunk/libgfortran/runtime/in_pack_generic.c
    trunk/libgfortran/runtime/in_unpack_generic.c
    trunk/libgomp/ChangeLog
    trunk/libgomp/configure
    trunk/libgomp/configure.ac

[Bug fortran/32049] Support on x86_64 also kind=16

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32049

--- Comment #10 from Tobias Burnus <burnus at gcc dot gnu.org> 2010-11-15 09:12:19 UTC ---
(In reply to comment #9)
> Last patch:

Now: http://gcc.gnu.org/ml/fortran/2010-11/msg00243.html

[Bug fortran/32049] Support on x86_64 also kind=16

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32049

--- Comment #9 from Tobias Burnus <burnus at gcc dot gnu.org> 2010-10-27 21:53:14 UTC ---
Last patch: http://gcc.gnu.org/ml/fortran/2010-10/msg00262.html