| Message ID | 1482156263-22267-4-git-send-email-adhemerval.zanella@linaro.org |
|---|---|
| State | New |
| Headers | show |
PING. On 19/12/2016 12:04, Adhemerval Zanella wrote: > This patch removes the powerpc assembly implementation of fmax/fmin. > Based on benchtests, the assembly ones shows: > > $ ./testrun.sh benchtests/bench-fmax > "fmax": { > "": { > "duration": 5.07586e+09, > "iterations": 2.01676e+09, > "max": 1350.39, > "min": 2.073, > "mean": 2.51684 > }, > "qNaN": { > "duration": 5.09315e+09, > "iterations": 8.4568e+08, > "max": 2788, > "min": 5.806, > "mean": 6.02255 > }, > "sNaN": { > "duration": 5.09073e+09, > "iterations": 8.42316e+08, > "max": 4215.84, > "min": 5.737, > "mean": 6.04373 > } > > And > > $ ./testrun.sh benchtests/bench-fmin > "fmin": { > "": { > "duration": 5.07711e+09, > "iterations": 2.02982e+09, > "max": 497.094, > "min": 2.073, > "mean": 2.50126 > }, > "qNaN": { > "duration": 5.09134e+09, > "iterations": 8.46968e+08, > "max": 2255.14, > "min": 5.807, > "mean": 6.01125 > }, > "sNaN": { > "duration": 5.09122e+09, > "iterations": 8.4746e+08, > "max": 1969.38, > "min": 5.729, > "mean": 6.00763 > } > } > > The default implementation (math/s_f{max.min}_template.c) shows slight better > latency for all cases: > > $ ./testrun.sh benchtests/bench-fmax > "fmax": { > "": { > "duration": 5.07044e+09, > "iterations": 2.38695e+09, > "max": 2048.58, > "min": 2.073, > "mean": 2.12423 > }, > "qNaN": { > "duration": 5.09004e+09, > "iterations": 9.45428e+08, > "max": 3306.93, > "min": 5.138, > "mean": 5.38385 > }, > "sNaN": { > "duration": 5.08458e+09, > "iterations": 1.15959e+09, > "max": 972.008, > "min": 3.321, > "mean": 4.3848 > } > } > > And: > > $ ./testrun.sh benchtests/bench-fmin > "fmin": { > "": { > "duration": 5.06817e+09, > "iterations": 2.3913e+09, > "max": 1177.9, > "min": 2.073, > "mean": 2.11942 > }, > "qNaN": { > "duration": 5.08857e+09, > "iterations": 9.45656e+08, > "max": 2658.83, > "min": 5.09, > "mean": 5.38099 > }, > "sNaN": { > "duration": 5.08093e+09, > "iterations": 1.16725e+09, > "max": 1030.74, > "min": 3.323, > "mean": 4.3529 > } > } > > Both were run with GCC 5.4 (ubuntu 16 default installation) using default > compiler flags on POWER8E 3.4GHz (powerpc64le-linux-gnu). > --- > ChangeLog | 10 +++++ > sysdeps/powerpc/fpu/s_fmax.S | 77 ---------------------------------- > sysdeps/powerpc/fpu/s_fmaxf.S | 1 - > sysdeps/powerpc/fpu/s_fmin.S | 77 ---------------------------------- > sysdeps/powerpc/fpu/s_fminf.S | 1 - > sysdeps/powerpc/powerpc32/fpu/s_fmax.S | 5 --- > sysdeps/powerpc/powerpc32/fpu/s_fmin.S | 5 --- > sysdeps/powerpc/powerpc64/fpu/s_fmax.S | 5 --- > sysdeps/powerpc/powerpc64/fpu/s_fmin.S | 5 --- > 9 files changed, 10 insertions(+), 176 deletions(-) > delete mode 100644 sysdeps/powerpc/fpu/s_fmax.S > delete mode 100644 sysdeps/powerpc/fpu/s_fmaxf.S > delete mode 100644 sysdeps/powerpc/fpu/s_fmin.S > delete mode 100644 sysdeps/powerpc/fpu/s_fminf.S > delete mode 100644 sysdeps/powerpc/powerpc32/fpu/s_fmax.S > delete mode 100644 sysdeps/powerpc/powerpc32/fpu/s_fmin.S > delete mode 100644 sysdeps/powerpc/powerpc64/fpu/s_fmax.S > delete mode 100644 sysdeps/powerpc/powerpc64/fpu/s_fmin.S > > diff --git a/sysdeps/powerpc/fpu/s_fmax.S b/sysdeps/powerpc/fpu/s_fmax.S > deleted file mode 100644 > index e6405c0..0000000 > --- a/sysdeps/powerpc/fpu/s_fmax.S > +++ /dev/null > @@ -1,77 +0,0 @@ > -/* Floating-point maximum. PowerPC version. > - Copyright (C) 1997-2016 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 Lesser General Public > - License as published by the Free Software Foundation; either > - version 2.1 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 > - Lesser General Public License for more details. > - > - You should have received a copy of the GNU Lesser General Public > - License along with the GNU C Library; if not, see > - <http://www.gnu.org/licenses/>. */ > - > -#include <sysdep.h> > - > -ENTRY(__fmax) > -/* double [f1] fmax (double [f1] x, double [f2] y); */ > - fcmpu cr0,fp1,fp2 > - blt cr0,0f /* if x < y, neither x nor y can be NaN... */ > - bnulr+ cr0 > -/* x and y are unordered, so one of x or y must be a NaN... */ > - fcmpu cr1,fp2,fp2 > - bun cr1,1f > -/* x is a NaN; y is not. Test if x is signaling. */ > -#ifdef __powerpc64__ > - stfd fp1,-8(r1) > - lwz r3,-8+HIWORD(r1) > -#else > - stwu r1,-16(r1) > - cfi_adjust_cfa_offset (16) > - stfd fp1,8(r1) > - lwz r3,8+HIWORD(r1) > - addi r1,r1,16 > - cfi_adjust_cfa_offset (-16) > -#endif > - andis. r3,r3,8 > - bne cr0,0f > - b 2f > -1: /* y is a NaN; x may or may not be. */ > - fcmpu cr1,fp1,fp1 > - bun cr1,2f > -/* y is a NaN; x is not. Test if y is signaling. */ > -#ifdef __powerpc64__ > - stfd fp2,-8(r1) > - lwz r3,-8+HIWORD(r1) > -#else > - stwu r1,-16(r1) > - cfi_adjust_cfa_offset (16) > - stfd fp2,8(r1) > - lwz r3,8+HIWORD(r1) > - addi r1,r1,16 > - cfi_adjust_cfa_offset (-16) > -#endif > - andis. r3,r3,8 > - bnelr cr0 > -2: /* x and y are NaNs, or one is a signaling NaN. */ > - fadd fp1,fp1,fp2 > - blr > -0: fmr fp1,fp2 > - blr > -END(__fmax) > - > -weak_alias (__fmax,fmax) > - > -/* It turns out that it's safe to use this code even for single-precision. */ > -strong_alias(__fmax,__fmaxf) > -weak_alias (__fmax,fmaxf) > - > -#ifdef NO_LONG_DOUBLE > -weak_alias (__fmax,__fmaxl) > -weak_alias (__fmax,fmaxl) > -#endif > diff --git a/sysdeps/powerpc/fpu/s_fmaxf.S b/sysdeps/powerpc/fpu/s_fmaxf.S > deleted file mode 100644 > index 3c2d62b..0000000 > --- a/sysdeps/powerpc/fpu/s_fmaxf.S > +++ /dev/null > @@ -1 +0,0 @@ > -/* __fmaxf is in s_fmax.c */ > diff --git a/sysdeps/powerpc/fpu/s_fmin.S b/sysdeps/powerpc/fpu/s_fmin.S > deleted file mode 100644 > index 9ae77fe..0000000 > --- a/sysdeps/powerpc/fpu/s_fmin.S > +++ /dev/null > @@ -1,77 +0,0 @@ > -/* Floating-point minimum. PowerPC version. > - Copyright (C) 1997-2016 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 Lesser General Public > - License as published by the Free Software Foundation; either > - version 2.1 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 > - Lesser General Public License for more details. > - > - You should have received a copy of the GNU Lesser General Public > - License along with the GNU C Library; if not, see > - <http://www.gnu.org/licenses/>. */ > - > -#include <sysdep.h> > - > -ENTRY(__fmin) > -/* double [f1] fmin (double [f1] x, double [f2] y); */ > - fcmpu cr0,fp1,fp2 > - bgt cr0,0f /* if x > y, neither x nor y can be NaN... */ > - bnulr+ cr0 > -/* x and y are unordered, so one of x or y must be a NaN... */ > - fcmpu cr1,fp2,fp2 > - bun cr1,1f > -/* x is a NaN; y is not. Test if x is signaling. */ > -#ifdef __powerpc64__ > - stfd fp1,-8(r1) > - lwz r3,-8+HIWORD(r1) > -#else > - stwu r1,-16(r1) > - cfi_adjust_cfa_offset (16) > - stfd fp1,8(r1) > - lwz r3,8+HIWORD(r1) > - addi r1,r1,16 > - cfi_adjust_cfa_offset (-16) > -#endif > - andis. r3,r3,8 > - bne cr0,0f > - b 2f > -1: /* y is a NaN; x may or may not be. */ > - fcmpu cr1,fp1,fp1 > - bun cr1,2f > -/* y is a NaN; x is not. Test if y is signaling. */ > -#ifdef __powerpc64__ > - stfd fp2,-8(r1) > - lwz r3,-8+HIWORD(r1) > -#else > - stwu r1,-16(r1) > - cfi_adjust_cfa_offset (16) > - stfd fp2,8(r1) > - lwz r3,8+HIWORD(r1) > - addi r1,r1,16 > - cfi_adjust_cfa_offset (-16) > -#endif > - andis. r3,r3,8 > - bnelr cr0 > -2: /* x and y are NaNs, or one is a signaling NaN. */ > - fadd fp1,fp1,fp2 > - blr > -0: fmr fp1,fp2 > - blr > -END(__fmin) > - > -weak_alias (__fmin,fmin) > - > -/* It turns out that it's safe to use this code even for single-precision. */ > -strong_alias(__fmin,__fminf) > -weak_alias (__fmin,fminf) > - > -#ifdef NO_LONG_DOUBLE > -weak_alias (__fmin,__fminl) > -weak_alias (__fmin,fminl) > -#endif > diff --git a/sysdeps/powerpc/fpu/s_fminf.S b/sysdeps/powerpc/fpu/s_fminf.S > deleted file mode 100644 > index 10ab7fe..0000000 > --- a/sysdeps/powerpc/fpu/s_fminf.S > +++ /dev/null > @@ -1 +0,0 @@ > -/* __fminf is in s_fmin.c */ > diff --git a/sysdeps/powerpc/powerpc32/fpu/s_fmax.S b/sysdeps/powerpc/powerpc32/fpu/s_fmax.S > deleted file mode 100644 > index 6973576..0000000 > --- a/sysdeps/powerpc/powerpc32/fpu/s_fmax.S > +++ /dev/null > @@ -1,5 +0,0 @@ > -#include <math_ldbl_opt.h> > -#include <sysdeps/powerpc/fpu/s_fmax.S> > -#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1) > -compat_symbol (libm, __fmax, fmaxl, GLIBC_2_1) > -#endif > diff --git a/sysdeps/powerpc/powerpc32/fpu/s_fmin.S b/sysdeps/powerpc/powerpc32/fpu/s_fmin.S > deleted file mode 100644 > index 6d4a0a9..0000000 > --- a/sysdeps/powerpc/powerpc32/fpu/s_fmin.S > +++ /dev/null > @@ -1,5 +0,0 @@ > -#include <math_ldbl_opt.h> > -#include <sysdeps/powerpc/fpu/s_fmin.S> > -#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1) > -compat_symbol (libm, __fmin, fminl, GLIBC_2_1) > -#endif > diff --git a/sysdeps/powerpc/powerpc64/fpu/s_fmax.S b/sysdeps/powerpc/powerpc64/fpu/s_fmax.S > deleted file mode 100644 > index 6973576..0000000 > --- a/sysdeps/powerpc/powerpc64/fpu/s_fmax.S > +++ /dev/null > @@ -1,5 +0,0 @@ > -#include <math_ldbl_opt.h> > -#include <sysdeps/powerpc/fpu/s_fmax.S> > -#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1) > -compat_symbol (libm, __fmax, fmaxl, GLIBC_2_1) > -#endif > diff --git a/sysdeps/powerpc/powerpc64/fpu/s_fmin.S b/sysdeps/powerpc/powerpc64/fpu/s_fmin.S > deleted file mode 100644 > index 6d4a0a9..0000000 > --- a/sysdeps/powerpc/powerpc64/fpu/s_fmin.S > +++ /dev/null > @@ -1,5 +0,0 @@ > -#include <math_ldbl_opt.h> > -#include <sysdeps/powerpc/fpu/s_fmin.S> > -#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1) > -compat_symbol (libm, __fmin, fminl, GLIBC_2_1) > -#endif >
Adhemerval Zanella <adhemerval.zanella@linaro.org> writes: > This patch removes the powerpc assembly implementation of fmax/fmin. > Based on benchtests, the assembly ones shows: > > $ ./testrun.sh benchtests/bench-fmax > "fmax": { > "": { > "duration": 5.07586e+09, > "iterations": 2.01676e+09, > "max": 1350.39, > "min": 2.073, > "mean": 2.51684 > }, > "qNaN": { > "duration": 5.09315e+09, > "iterations": 8.4568e+08, > "max": 2788, > "min": 5.806, > "mean": 6.02255 > }, > "sNaN": { > "duration": 5.09073e+09, > "iterations": 8.42316e+08, > "max": 4215.84, > "min": 5.737, > "mean": 6.04373 > } > > And > > $ ./testrun.sh benchtests/bench-fmin > "fmin": { > "": { > "duration": 5.07711e+09, > "iterations": 2.02982e+09, > "max": 497.094, > "min": 2.073, > "mean": 2.50126 > }, > "qNaN": { > "duration": 5.09134e+09, > "iterations": 8.46968e+08, > "max": 2255.14, > "min": 5.807, > "mean": 6.01125 > }, > "sNaN": { > "duration": 5.09122e+09, > "iterations": 8.4746e+08, > "max": 1969.38, > "min": 5.729, > "mean": 6.00763 > } > } > > The default implementation (math/s_f{max.min}_template.c) shows slight better > latency for all cases: > > $ ./testrun.sh benchtests/bench-fmax > "fmax": { > "": { > "duration": 5.07044e+09, > "iterations": 2.38695e+09, > "max": 2048.58, > "min": 2.073, > "mean": 2.12423 > }, > "qNaN": { > "duration": 5.09004e+09, > "iterations": 9.45428e+08, > "max": 3306.93, > "min": 5.138, > "mean": 5.38385 > }, > "sNaN": { > "duration": 5.08458e+09, > "iterations": 1.15959e+09, > "max": 972.008, > "min": 3.321, > "mean": 4.3848 > } > } > > And: > > $ ./testrun.sh benchtests/bench-fmin > "fmin": { > "": { > "duration": 5.06817e+09, > "iterations": 2.3913e+09, > "max": 1177.9, > "min": 2.073, > "mean": 2.11942 > }, > "qNaN": { > "duration": 5.08857e+09, > "iterations": 9.45656e+08, > "max": 2658.83, > "min": 5.09, > "mean": 5.38099 > }, > "sNaN": { > "duration": 5.08093e+09, > "iterations": 1.16725e+09, > "max": 1030.74, > "min": 3.323, > "mean": 4.3529 > } > } > > Both were run with GCC 5.4 (ubuntu 16 default installation) using default > compiler flags on POWER8E 3.4GHz (powerpc64le-linux-gnu). LGTM. For the record: the results with older GCC versions (e.g. 4.8) are getting worse, but I don't think this regression in an old GCC version should block this patch. -- Tulio Magno
diff --git a/sysdeps/powerpc/fpu/s_fmax.S b/sysdeps/powerpc/fpu/s_fmax.S deleted file mode 100644 index e6405c0..0000000 --- a/sysdeps/powerpc/fpu/s_fmax.S +++ /dev/null @@ -1,77 +0,0 @@ -/* Floating-point maximum. PowerPC version. - Copyright (C) 1997-2016 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 Lesser General Public - License as published by the Free Software Foundation; either - version 2.1 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 - Lesser General Public License for more details. - - You should have received a copy of the GNU Lesser General Public - License along with the GNU C Library; if not, see - <http://www.gnu.org/licenses/>. */ - -#include <sysdep.h> - -ENTRY(__fmax) -/* double [f1] fmax (double [f1] x, double [f2] y); */ - fcmpu cr0,fp1,fp2 - blt cr0,0f /* if x < y, neither x nor y can be NaN... */ - bnulr+ cr0 -/* x and y are unordered, so one of x or y must be a NaN... */ - fcmpu cr1,fp2,fp2 - bun cr1,1f -/* x is a NaN; y is not. Test if x is signaling. */ -#ifdef __powerpc64__ - stfd fp1,-8(r1) - lwz r3,-8+HIWORD(r1) -#else - stwu r1,-16(r1) - cfi_adjust_cfa_offset (16) - stfd fp1,8(r1) - lwz r3,8+HIWORD(r1) - addi r1,r1,16 - cfi_adjust_cfa_offset (-16) -#endif - andis. r3,r3,8 - bne cr0,0f - b 2f -1: /* y is a NaN; x may or may not be. */ - fcmpu cr1,fp1,fp1 - bun cr1,2f -/* y is a NaN; x is not. Test if y is signaling. */ -#ifdef __powerpc64__ - stfd fp2,-8(r1) - lwz r3,-8+HIWORD(r1) -#else - stwu r1,-16(r1) - cfi_adjust_cfa_offset (16) - stfd fp2,8(r1) - lwz r3,8+HIWORD(r1) - addi r1,r1,16 - cfi_adjust_cfa_offset (-16) -#endif - andis. r3,r3,8 - bnelr cr0 -2: /* x and y are NaNs, or one is a signaling NaN. */ - fadd fp1,fp1,fp2 - blr -0: fmr fp1,fp2 - blr -END(__fmax) - -weak_alias (__fmax,fmax) - -/* It turns out that it's safe to use this code even for single-precision. */ -strong_alias(__fmax,__fmaxf) -weak_alias (__fmax,fmaxf) - -#ifdef NO_LONG_DOUBLE -weak_alias (__fmax,__fmaxl) -weak_alias (__fmax,fmaxl) -#endif diff --git a/sysdeps/powerpc/fpu/s_fmaxf.S b/sysdeps/powerpc/fpu/s_fmaxf.S deleted file mode 100644 index 3c2d62b..0000000 --- a/sysdeps/powerpc/fpu/s_fmaxf.S +++ /dev/null @@ -1 +0,0 @@ -/* __fmaxf is in s_fmax.c */ diff --git a/sysdeps/powerpc/fpu/s_fmin.S b/sysdeps/powerpc/fpu/s_fmin.S deleted file mode 100644 index 9ae77fe..0000000 --- a/sysdeps/powerpc/fpu/s_fmin.S +++ /dev/null @@ -1,77 +0,0 @@ -/* Floating-point minimum. PowerPC version. - Copyright (C) 1997-2016 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 Lesser General Public - License as published by the Free Software Foundation; either - version 2.1 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 - Lesser General Public License for more details. - - You should have received a copy of the GNU Lesser General Public - License along with the GNU C Library; if not, see - <http://www.gnu.org/licenses/>. */ - -#include <sysdep.h> - -ENTRY(__fmin) -/* double [f1] fmin (double [f1] x, double [f2] y); */ - fcmpu cr0,fp1,fp2 - bgt cr0,0f /* if x > y, neither x nor y can be NaN... */ - bnulr+ cr0 -/* x and y are unordered, so one of x or y must be a NaN... */ - fcmpu cr1,fp2,fp2 - bun cr1,1f -/* x is a NaN; y is not. Test if x is signaling. */ -#ifdef __powerpc64__ - stfd fp1,-8(r1) - lwz r3,-8+HIWORD(r1) -#else - stwu r1,-16(r1) - cfi_adjust_cfa_offset (16) - stfd fp1,8(r1) - lwz r3,8+HIWORD(r1) - addi r1,r1,16 - cfi_adjust_cfa_offset (-16) -#endif - andis. r3,r3,8 - bne cr0,0f - b 2f -1: /* y is a NaN; x may or may not be. */ - fcmpu cr1,fp1,fp1 - bun cr1,2f -/* y is a NaN; x is not. Test if y is signaling. */ -#ifdef __powerpc64__ - stfd fp2,-8(r1) - lwz r3,-8+HIWORD(r1) -#else - stwu r1,-16(r1) - cfi_adjust_cfa_offset (16) - stfd fp2,8(r1) - lwz r3,8+HIWORD(r1) - addi r1,r1,16 - cfi_adjust_cfa_offset (-16) -#endif - andis. r3,r3,8 - bnelr cr0 -2: /* x and y are NaNs, or one is a signaling NaN. */ - fadd fp1,fp1,fp2 - blr -0: fmr fp1,fp2 - blr -END(__fmin) - -weak_alias (__fmin,fmin) - -/* It turns out that it's safe to use this code even for single-precision. */ -strong_alias(__fmin,__fminf) -weak_alias (__fmin,fminf) - -#ifdef NO_LONG_DOUBLE -weak_alias (__fmin,__fminl) -weak_alias (__fmin,fminl) -#endif diff --git a/sysdeps/powerpc/fpu/s_fminf.S b/sysdeps/powerpc/fpu/s_fminf.S deleted file mode 100644 index 10ab7fe..0000000 --- a/sysdeps/powerpc/fpu/s_fminf.S +++ /dev/null @@ -1 +0,0 @@ -/* __fminf is in s_fmin.c */ diff --git a/sysdeps/powerpc/powerpc32/fpu/s_fmax.S b/sysdeps/powerpc/powerpc32/fpu/s_fmax.S deleted file mode 100644 index 6973576..0000000 --- a/sysdeps/powerpc/powerpc32/fpu/s_fmax.S +++ /dev/null @@ -1,5 +0,0 @@ -#include <math_ldbl_opt.h> -#include <sysdeps/powerpc/fpu/s_fmax.S> -#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1) -compat_symbol (libm, __fmax, fmaxl, GLIBC_2_1) -#endif diff --git a/sysdeps/powerpc/powerpc32/fpu/s_fmin.S b/sysdeps/powerpc/powerpc32/fpu/s_fmin.S deleted file mode 100644 index 6d4a0a9..0000000 --- a/sysdeps/powerpc/powerpc32/fpu/s_fmin.S +++ /dev/null @@ -1,5 +0,0 @@ -#include <math_ldbl_opt.h> -#include <sysdeps/powerpc/fpu/s_fmin.S> -#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1) -compat_symbol (libm, __fmin, fminl, GLIBC_2_1) -#endif diff --git a/sysdeps/powerpc/powerpc64/fpu/s_fmax.S b/sysdeps/powerpc/powerpc64/fpu/s_fmax.S deleted file mode 100644 index 6973576..0000000 --- a/sysdeps/powerpc/powerpc64/fpu/s_fmax.S +++ /dev/null @@ -1,5 +0,0 @@ -#include <math_ldbl_opt.h> -#include <sysdeps/powerpc/fpu/s_fmax.S> -#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1) -compat_symbol (libm, __fmax, fmaxl, GLIBC_2_1) -#endif diff --git a/sysdeps/powerpc/powerpc64/fpu/s_fmin.S b/sysdeps/powerpc/powerpc64/fpu/s_fmin.S deleted file mode 100644 index 6d4a0a9..0000000 --- a/sysdeps/powerpc/powerpc64/fpu/s_fmin.S +++ /dev/null @@ -1,5 +0,0 @@ -#include <math_ldbl_opt.h> -#include <sysdeps/powerpc/fpu/s_fmin.S> -#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1) -compat_symbol (libm, __fmin, fminl, GLIBC_2_1) -#endif
This patch removes the powerpc assembly implementation of fmax/fmin. Based on benchtests, the assembly ones shows: $ ./testrun.sh benchtests/bench-fmax "fmax": { "": { "duration": 5.07586e+09, "iterations": 2.01676e+09, "max": 1350.39, "min": 2.073, "mean": 2.51684 }, "qNaN": { "duration": 5.09315e+09, "iterations": 8.4568e+08, "max": 2788, "min": 5.806, "mean": 6.02255 }, "sNaN": { "duration": 5.09073e+09, "iterations": 8.42316e+08, "max": 4215.84, "min": 5.737, "mean": 6.04373 } And $ ./testrun.sh benchtests/bench-fmin "fmin": { "": { "duration": 5.07711e+09, "iterations": 2.02982e+09, "max": 497.094, "min": 2.073, "mean": 2.50126 }, "qNaN": { "duration": 5.09134e+09, "iterations": 8.46968e+08, "max": 2255.14, "min": 5.807, "mean": 6.01125 }, "sNaN": { "duration": 5.09122e+09, "iterations": 8.4746e+08, "max": 1969.38, "min": 5.729, "mean": 6.00763 } } The default implementation (math/s_f{max.min}_template.c) shows slight better latency for all cases: $ ./testrun.sh benchtests/bench-fmax "fmax": { "": { "duration": 5.07044e+09, "iterations": 2.38695e+09, "max": 2048.58, "min": 2.073, "mean": 2.12423 }, "qNaN": { "duration": 5.09004e+09, "iterations": 9.45428e+08, "max": 3306.93, "min": 5.138, "mean": 5.38385 }, "sNaN": { "duration": 5.08458e+09, "iterations": 1.15959e+09, "max": 972.008, "min": 3.321, "mean": 4.3848 } } And: $ ./testrun.sh benchtests/bench-fmin "fmin": { "": { "duration": 5.06817e+09, "iterations": 2.3913e+09, "max": 1177.9, "min": 2.073, "mean": 2.11942 }, "qNaN": { "duration": 5.08857e+09, "iterations": 9.45656e+08, "max": 2658.83, "min": 5.09, "mean": 5.38099 }, "sNaN": { "duration": 5.08093e+09, "iterations": 1.16725e+09, "max": 1030.74, "min": 3.323, "mean": 4.3529 } } Both were run with GCC 5.4 (ubuntu 16 default installation) using default compiler flags on POWER8E 3.4GHz (powerpc64le-linux-gnu). --- ChangeLog | 10 +++++ sysdeps/powerpc/fpu/s_fmax.S | 77 ---------------------------------- sysdeps/powerpc/fpu/s_fmaxf.S | 1 - sysdeps/powerpc/fpu/s_fmin.S | 77 ---------------------------------- sysdeps/powerpc/fpu/s_fminf.S | 1 - sysdeps/powerpc/powerpc32/fpu/s_fmax.S | 5 --- sysdeps/powerpc/powerpc32/fpu/s_fmin.S | 5 --- sysdeps/powerpc/powerpc64/fpu/s_fmax.S | 5 --- sysdeps/powerpc/powerpc64/fpu/s_fmin.S | 5 --- 9 files changed, 10 insertions(+), 176 deletions(-) delete mode 100644 sysdeps/powerpc/fpu/s_fmax.S delete mode 100644 sysdeps/powerpc/fpu/s_fmaxf.S delete mode 100644 sysdeps/powerpc/fpu/s_fmin.S delete mode 100644 sysdeps/powerpc/fpu/s_fminf.S delete mode 100644 sysdeps/powerpc/powerpc32/fpu/s_fmax.S delete mode 100644 sysdeps/powerpc/powerpc32/fpu/s_fmin.S delete mode 100644 sysdeps/powerpc/powerpc64/fpu/s_fmax.S delete mode 100644 sysdeps/powerpc/powerpc64/fpu/s_fmin.S -- 2.7.4