[PATCH v2 3/5] powerpc: Remove modff optimization

[Sachin Monga]

Looks fine.

Reviewed-by: Sachin Monga <smonga@linux.ibm.com>


Regards:
Sachin.

On 16/06/25 6:47 pm, Adhemerval Zanella wrote:
> The generic implementation is slight more optimized than the powerpc
> one, where it has a more optimized inf/nan check (by not using FP
> unit checks, along with branch prediction hints), and removed one
> branch by issuing trunc instead of a combination of floor/ceil (which
> also generated less code).
>
> On power10 with gcc 14.2.1:
>
> reciprocal-throughput        master        patch        difference
> workload-0_1                 1.5210       1.3942             8.34%
> workload-1_maxint            2.0926       1.3940            33.38%
> workload-maxint_maxfloat     1.7851       1.3940            21.91%
> workload-integral            1.5216       1.3941             8.37%
>
> latency                      master        patch        difference
> workload-0_1                 1.5928       2.6337           -65.35%
> workload-1_maxint            3.2929       2.6337            20.02%
> workload-maxint_maxfloat     1.9697       2.6341           -33.73%
> workload-integral            2.0597       2.6337           -27.87%
>
> Checked on powerpc64le-linux-gnu.
> ---
>   sysdeps/ieee754/flt-32/s_modff.c              |  2 +
>   sysdeps/powerpc/fpu/math-use-builtins-trunc.h |  8 +++
>   sysdeps/powerpc/fpu/s_modff.c                 | 55 -------------------
>   .../power4/fpu/multiarch/s_modff-power5+.c    |  2 +-
>   .../be/fpu/multiarch/s_modff-power5+.c        |  2 +-
>   5 files changed, 12 insertions(+), 57 deletions(-)
>   create mode 100644 sysdeps/powerpc/fpu/math-use-builtins-trunc.h
>   delete mode 100644 sysdeps/powerpc/fpu/s_modff.c
>
> diff --git a/sysdeps/ieee754/flt-32/s_modff.c b/sysdeps/ieee754/flt-32/s_modff.c
> index e6d2e4ab49..965136bac9 100644
> --- a/sysdeps/ieee754/flt-32/s_modff.c
> +++ b/sysdeps/ieee754/flt-32/s_modff.c
> @@ -64,4 +64,6 @@ __modff (float x, float *iptr)
>     return asfloat (t & SIGN_MASK);
>   #endif
>   }
> +#ifndef __modff
>   libm_alias_float (__modf, modf)
> +#endif
> diff --git a/sysdeps/powerpc/fpu/math-use-builtins-trunc.h b/sysdeps/powerpc/fpu/math-use-builtins-trunc.h
> new file mode 100644
> index 0000000000..fd186d3510
> --- /dev/null
> +++ b/sysdeps/powerpc/fpu/math-use-builtins-trunc.h
> @@ -0,0 +1,8 @@
> +#define USE_TRUNC_BUILTIN 0
> +#ifdef _ARCH_PWR5X
> +# define USE_TRUNCF_BUILTIN 1
> +#else
> +# define USE_TRUNCF_BUILTIN 0
> +#endif
> +#define USE_TRUNCL_BUILTIN 0
> +#define USE_TRUNCF128_BUILTIN 0
> diff --git a/sysdeps/powerpc/fpu/s_modff.c b/sysdeps/powerpc/fpu/s_modff.c
> deleted file mode 100644
> index 79eeb7b810..0000000000
> --- a/sysdeps/powerpc/fpu/s_modff.c
> +++ /dev/null
> @@ -1,55 +0,0 @@
> -/* Copyright (C) 2013-2025 Free Software Foundation, Inc.
> -   This file is part of the GNU C Library
> -
> -   The GNU C Library is free software; you can redistribute it and/or
> -   modify it under the terms of the GNU Library General Public License as
> -   published by the Free Software Foundation; either version 2 of the
> -   License, or (at your option) any later version.
> -
> -   The GNU C Library is distributed in the hope that it will be useful,
> -   but WITHOUT ANY WARRANTY; without even the implied warranty of
> -   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> -   Library General Public License for more details.
> -
> -   You should have received a copy of the GNU Library General Public
> -   License along with the GNU C Library; see the file COPYING.LIB.  If
> -   not, see <https://www.gnu.org/licenses/>.  */
> -
> -/* ISA 2.07 provides fast GPR to FP instruction (mfvsr{d,wz}) which make
> -   generic implementation faster.  Also disables for old ISAs that do not
> -   have ceil/floor instructions.  */
> -#if defined(_ARCH_PWR8) || !defined(_ARCH_PWR5X)
> -# include <sysdeps/ieee754/flt-32/s_modff.c>
> -#else
> -# include <math.h>
> -# include <libm-alias-float.h>
> -
> -float
> -__modff (float x, float *iptr)
> -{
> -  if (__builtin_isinff (x))
> -    {
> -      *iptr = x;
> -      return copysignf (0.0, x);
> -    }
> -  else if (__builtin_isnanf (x))
> -    {
> -      *iptr = NAN;
> -      return NAN;
> -    }
> -
> -  if (x >= 0.0)
> -    {
> -      *iptr = floorf (x);
> -      return copysignf (x - *iptr, x);
> -    }
> -  else
> -    {
> -      *iptr = ceilf (x);
> -      return copysignf (x - *iptr, x);
> -    }
> -}
> -# ifndef __modff
> -libm_alias_float (__modf, modf)
> -# endif
> -#endif
> diff --git a/sysdeps/powerpc/powerpc32/power4/fpu/multiarch/s_modff-power5+.c b/sysdeps/powerpc/powerpc32/power4/fpu/multiarch/s_modff-power5+.c
> index 69591da43f..15bfa0bba9 100644
> --- a/sysdeps/powerpc/powerpc32/power4/fpu/multiarch/s_modff-power5+.c
> +++ b/sysdeps/powerpc/powerpc32/power4/fpu/multiarch/s_modff-power5+.c
> @@ -17,4 +17,4 @@
>      <https://www.gnu.org/licenses/>.  */
>   
>   #define __modff __modff_power5plus
> -#include <sysdeps/powerpc/fpu/s_modff.c>
> +#include <sysdeps/ieee754/flt-32/s_modff.c>
> diff --git a/sysdeps/powerpc/powerpc64/be/fpu/multiarch/s_modff-power5+.c b/sysdeps/powerpc/powerpc64/be/fpu/multiarch/s_modff-power5+.c
> index 69591da43f..15bfa0bba9 100644
> --- a/sysdeps/powerpc/powerpc64/be/fpu/multiarch/s_modff-power5+.c
> +++ b/sysdeps/powerpc/powerpc64/be/fpu/multiarch/s_modff-power5+.c
> @@ -17,4 +17,4 @@
>      <https://www.gnu.org/licenses/>.  */
>   
>   #define __modff __modff_power5plus
> -#include <sysdeps/powerpc/fpu/s_modff.c>
> +#include <sysdeps/ieee754/flt-32/s_modff.c>