diff options
author | Adhemerval Zanella <azanella@linux.vnet.ibm.com> | 2011-11-11 13:33:38 -0500 |
---|---|---|
committer | Ulrich Drepper <drepper@gmail.com> | 2011-11-11 13:33:38 -0500 |
commit | 8a6d525522b2fb04cec86b48c5b25818d395bbbd (patch) | |
tree | 34e9e7e8b6734ad4d57fa7d3e28d90722cdb329e /sysdeps/powerpc/fpu | |
parent | 1d8f7ddc04a4782c1bf2d6f48ce40e7d57caa920 (diff) | |
download | glibc-8a6d525522b2fb04cec86b48c5b25818d395bbbd.zip glibc-8a6d525522b2fb04cec86b48c5b25818d395bbbd.tar.gz glibc-8a6d525522b2fb04cec86b48c5b25818d395bbbd.tar.bz2 |
PowerPC: Arithmetic function optimizations for POWER
Diffstat (limited to 'sysdeps/powerpc/fpu')
-rw-r--r-- | sysdeps/powerpc/fpu/e_sqrt.c | 1 | ||||
-rw-r--r-- | sysdeps/powerpc/fpu/e_sqrtf.c | 2 | ||||
-rw-r--r-- | sysdeps/powerpc/fpu/math_private.h | 137 |
3 files changed, 136 insertions, 4 deletions
diff --git a/sysdeps/powerpc/fpu/e_sqrt.c b/sysdeps/powerpc/fpu/e_sqrt.c index f9ded25..d59bd08 100644 --- a/sysdeps/powerpc/fpu/e_sqrt.c +++ b/sysdeps/powerpc/fpu/e_sqrt.c @@ -154,6 +154,7 @@ __slow_ieee754_sqrt (double x) return f_wash (x); } +#undef __ieee754_sqrt double __ieee754_sqrt (double x) { diff --git a/sysdeps/powerpc/fpu/e_sqrtf.c b/sysdeps/powerpc/fpu/e_sqrtf.c index 965faee..9c6b860 100644 --- a/sysdeps/powerpc/fpu/e_sqrtf.c +++ b/sysdeps/powerpc/fpu/e_sqrtf.c @@ -130,7 +130,7 @@ __slow_ieee754_sqrtf (float x) return f_washf (x); } - +#undef __ieee754_sqrtf float __ieee754_sqrtf (float x) { diff --git a/sysdeps/powerpc/fpu/math_private.h b/sysdeps/powerpc/fpu/math_private.h index 90021c6..c4dd217 100644 --- a/sysdeps/powerpc/fpu/math_private.h +++ b/sysdeps/powerpc/fpu/math_private.h @@ -1,5 +1,5 @@ /* Private inline math functions for powerpc. - Copyright (C) 2006 + Copyright (C) 2006, 2011 Free Software Foundation, Inc. This file is part of the GNU C Library. @@ -25,12 +25,145 @@ #include <ldsodefs.h> #include <dl-procinfo.h> +#include <math/math_private.h> + # if __WORDSIZE == 64 || defined _ARCH_PWR4 # define __CPU_HAS_FSQRT 1 + +#ifndef __ieee754_sqrt +# define __ieee754_sqrt(x) \ + ({ double __z; \ + __asm __volatile ( \ + " fsqrt %0,%1\n" \ + : "=f" (__z) \ + : "f"(x)); \ + __z; }) +#endif +#ifndef __ieee754_sqrtf +# define __ieee754_sqrtf(x) \ + ({ float __z; \ + __asm __volatile ( \ + " fsqrts %0,%1\n" \ + : "=f" (__z) \ + : "f"(x)); \ + __z; }) +#endif + # else # define __CPU_HAS_FSQRT ((GLRO(dl_hwcap) & PPC_FEATURE_64) != 0) +# endif // __WORDSIZE == 64 || defined _ARCH_PWR4 + + +#if defined _ARCH_PWR5X + +# ifndef __round +# define __round(x) \ + ({ double __z; \ + __asm __volatile ( \ + " frin %0,%1\n" \ + : "=f" (__z) \ + : "f" (x)); \ + __z; }) +# endif +# ifndef __roundf +# define __roundf(x) \ + ({ float __z; \ + __asm __volatile ( \ + " frin %0,%1\n" \ + " frsp %0,%0\n" \ + : "=f" (__z) \ + : "f" (x)); \ + __z; }) +# endif + +# ifndef __trunc +# define __trunc(x) \ + ({ double __z; \ + __asm __volatile ( \ + " friz %0,%1\n" \ + : "=f" (__z) \ + : "f" (x)); \ + __z; }) +# endif +# ifndef __truncf +# define __truncf(x) \ + ({ float __z; \ + __asm __volatile ( \ + " friz %0,%1\n" \ + " frsp %0,%0\n" \ + : "=f" (__z) \ + : "f" (x)); \ + __z; }) +# endif + +# ifndef __ceil +# define __ceil(x) \ + ({ double __z; \ + __asm __volatile ( \ + " frip %0,%1\n" \ + : "=f" (__z) \ + : "f" (x)); \ + __z; }) +# endif +# ifndef __ceilf +# define __ceilf(x) \ + ({ float __z; \ + __asm __volatile ( \ + " frip %0,%1\n" \ + " frsp %0,%0\n" \ + : "=f" (__z) \ + : "f" (x)); \ + __z; }) # endif +# ifndef __floor +# define __floor(x) \ + ({ double __z; \ + __asm __volatile ( \ + " frim %0,%1\n" \ + : "=f" (__z) \ + : "f" (x)); \ + __z; }) +# endif +# ifndef __floorf +# define __floorf(x) \ + ({ float __z; \ + __asm __volatile ( \ + " frim %0,%1\n" \ + " frsp %0,%0\n" \ + : "=f" (__z) \ + : "f" (x)); \ + __z; }) +# endif + +#endif /* defined _ARCH_PWR5X */ + + +#if defined _ARCH_PWR6 + +# ifndef __copysign +# define __copysign(x, y) \ + ({ double __z; \ + __asm __volatile ( \ + " fcpsgn %0,%1,%2\n" \ + : "=f" (__z) \ + : "f" (y), "f" (x)); \ + __z; }) +# endif +# ifndef __copysignf +# define __copysignf(x, y) \ + ({ float __z; \ + __asm __volatile ( \ + " fcpsgn %0,%1,%2\n" \ + " frsp %0,%0\n" \ + : "=f" (__z) \ + : "f" (y), "f" (x)); \ + __z; }) +# endif + +#endif /* defined _ARCH_PWR6 */ + + # ifndef __LIBC_INTERNAL_MATH_INLINES extern double __slow_ieee754_sqrt (double); __inline double @@ -78,6 +211,4 @@ __ieee754_sqrtf (float __x) } #endif /* __LIBC_INTERNAL_MATH_INLINES */ -#include <math/math_private.h> - #endif /* _PPC_MATH_PRIVATE_H_ */ |