[12/13] Use GCC builtins for round functions if desired.
Commit Message
This patch is using the corresponding GCC builtin for roundf, round,
roundl and roundf128 if the USE_FUNCTION_BUILTIN macros are defined to one
in math-use-builtins.h.
This is the case for s390 if build with at least --march=z196 --mzarch.
Otherwise the generic implementation is used.
---
sysdeps/generic/math-use-builtins.h | 5 +++++
sysdeps/ieee754/dbl-64/s_round.c | 6 +++++-
sysdeps/ieee754/float128/float128_private.h | 3 +++
sysdeps/ieee754/flt-32/s_roundf.c | 6 +++++-
sysdeps/ieee754/ldbl-128/s_roundl.c | 6 +++++-
sysdeps/s390/fpu/math-use-builtins.h | 6 ++++++
6 files changed, 29 insertions(+), 3 deletions(-)
Comments
On 02/12/2019 11:46, Stefan Liebler wrote:
> This patch is using the corresponding GCC builtin for roundf, round,
> roundl and roundf128 if the USE_FUNCTION_BUILTIN macros are defined to one
> in math-use-builtins.h.
>
> This is the case for s390 if build with at least --march=z196 --mzarch.
> Otherwise the generic implementation is used.
LGTM, thanks.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
> ---
> sysdeps/generic/math-use-builtins.h | 5 +++++
> sysdeps/ieee754/dbl-64/s_round.c | 6 +++++-
> sysdeps/ieee754/float128/float128_private.h | 3 +++
> sysdeps/ieee754/flt-32/s_roundf.c | 6 +++++-
> sysdeps/ieee754/ldbl-128/s_roundl.c | 6 +++++-
> sysdeps/s390/fpu/math-use-builtins.h | 6 ++++++
> 6 files changed, 29 insertions(+), 3 deletions(-)
>
> diff --git a/sysdeps/generic/math-use-builtins.h b/sysdeps/generic/math-use-builtins.h
> index ab379f45ba..34ca438a8c 100644
> --- a/sysdeps/generic/math-use-builtins.h
> +++ b/sysdeps/generic/math-use-builtins.h
> @@ -46,4 +46,9 @@
> #define USE_TRUNCL_BUILTIN 0
> #define USE_TRUNCF128_BUILTIN 0
>
> +#define USE_ROUND_BUILTIN 0
> +#define USE_ROUNDF_BUILTIN 0
> +#define USE_ROUNDL_BUILTIN 0
> +#define USE_ROUNDF128_BUILTIN 0
> +
> #endif /* math-use-builtins.h */
Ok.
> diff --git a/sysdeps/ieee754/dbl-64/s_round.c b/sysdeps/ieee754/dbl-64/s_round.c
> index 8f4b7cff86..00c8485efb 100644
> --- a/sysdeps/ieee754/dbl-64/s_round.c
> +++ b/sysdeps/ieee754/dbl-64/s_round.c
> @@ -23,11 +23,14 @@
> #include <math_private.h>
> #include <libm-alias-double.h>
> #include <stdint.h>
> -
> +#include <math-use-builtins.h>
>
> double
> __round (double x)
> {
> +#if USE_ROUND_BUILTIN
> + return __builtin_round (x);
> +#else
> int64_t i0, j0;
>
> EXTRACT_WORDS64 (i0, x);
> @@ -62,5 +65,6 @@ __round (double x)
>
> INSERT_WORDS64 (x, i0);
> return x;
> +#endif /* USE_ROUND_BUILTIN */
> }
> libm_alias_double (__round, round)
Ok.
> diff --git a/sysdeps/ieee754/float128/float128_private.h b/sysdeps/ieee754/float128/float128_private.h
> index 9cd3a63f11..a6c76ce364 100644
> --- a/sysdeps/ieee754/float128/float128_private.h
> +++ b/sysdeps/ieee754/float128/float128_private.h
> @@ -150,6 +150,8 @@
> #define USE_CEILL_BUILTIN USE_CEILF128_BUILTIN
> #undef USE_TRUNCL_BUILTIN
> #define USE_TRUNCL_BUILTIN USE_TRUNCF128_BUILTIN
> +#undef USE_ROUNDL_BUILTIN
> +#define USE_ROUNDL_BUILTIN USE_ROUNDF128_BUILTIN
>
> /* IEEE function renames. */
> #define __ieee754_acoshl __ieee754_acoshf128
> @@ -358,6 +360,7 @@
> #define __builtin_floorl __builtin_floorf128
> #define __builtin_ceill __builtin_ceilf128
> #define __builtin_truncl __builtin_truncf128
> +#define __builtin_roundl __builtin_roundf128
>
> /* Get the constant suffix from bits/floatn-compat.h. */
> #define L(x) __f128 (x)
Ok.
> diff --git a/sysdeps/ieee754/flt-32/s_roundf.c b/sysdeps/ieee754/flt-32/s_roundf.c
> index cddab7b505..33304b21f1 100644
> --- a/sysdeps/ieee754/flt-32/s_roundf.c
> +++ b/sysdeps/ieee754/flt-32/s_roundf.c
> @@ -22,11 +22,14 @@
>
> #include <math_private.h>
> #include <libm-alias-float.h>
> -
> +#include <math-use-builtins.h>
>
> float
> __roundf (float x)
> {
> +#if USE_ROUNDF_BUILTIN
> + return __builtin_roundf (x);
> +#else
> int32_t i0, j0;
>
> GET_FLOAT_WORD (i0, x);
> @@ -61,5 +64,6 @@ __roundf (float x)
>
> SET_FLOAT_WORD (x, i0);
> return x;
> +#endif /* USE_ROUNDF_BUILTIN */
> }
> libm_alias_float (__round, round)
Ok.
> diff --git a/sysdeps/ieee754/ldbl-128/s_roundl.c b/sysdeps/ieee754/ldbl-128/s_roundl.c
> index d41c9a04dc..8d3f42cd42 100644
> --- a/sysdeps/ieee754/ldbl-128/s_roundl.c
> +++ b/sysdeps/ieee754/ldbl-128/s_roundl.c
> @@ -23,11 +23,14 @@
>
> #include <math_private.h>
> #include <libm-alias-ldouble.h>
> -
> +#include <math-use-builtins.h>
>
> _Float128
> __roundl (_Float128 x)
> {
> +#if USE_ROUNDL_BUILTIN
> + return __builtin_roundl (x);
> +#else
> int32_t j0;
> uint64_t i1, i0;
>
> @@ -78,5 +81,6 @@ __roundl (_Float128 x)
>
> SET_LDOUBLE_WORDS64 (x, i0, i1);
> return x;
> +#endif /* USE_ROUNDL_BUILTIN */
> }
> libm_alias_ldouble (__round, round)
Ok.
> diff --git a/sysdeps/s390/fpu/math-use-builtins.h b/sysdeps/s390/fpu/math-use-builtins.h
> index 1d88a2c40f..5838a31c50 100644
> --- a/sysdeps/s390/fpu/math-use-builtins.h
> +++ b/sysdeps/s390/fpu/math-use-builtins.h
> @@ -46,18 +46,24 @@
> # define USE_TRUNCF_BUILTIN 1
> # define USE_TRUNCL_BUILTIN 1
>
> +# define USE_ROUND_BUILTIN 1
> +# define USE_ROUNDF_BUILTIN 1
> +# define USE_ROUNDL_BUILTIN 1
> +
> # if __GNUC_PREREQ (8, 1)
> # define USE_NEARBYINTF128_BUILTIN 1
> # define USE_RINTF128_BUILTIN 1
> # define USE_FLOORF128_BUILTIN 1
> # define USE_CEILF128_BUILTIN 1
> # define USE_TRUNCF128_BUILTIN 1
> +# define USE_ROUNDF128_BUILTIN 1
> # else
> # define USE_NEARBYINTF128_BUILTIN 0
> # define USE_RINTF128_BUILTIN 0
> # define USE_FLOORF128_BUILTIN 0
> # define USE_CEILF128_BUILTIN 0
> # define USE_TRUNCF128_BUILTIN 0
> +# define USE_ROUNDF128_BUILTIN 0
> # endif
>
> #else
>
Ok.
@@ -46,4 +46,9 @@
#define USE_TRUNCL_BUILTIN 0
#define USE_TRUNCF128_BUILTIN 0
+#define USE_ROUND_BUILTIN 0
+#define USE_ROUNDF_BUILTIN 0
+#define USE_ROUNDL_BUILTIN 0
+#define USE_ROUNDF128_BUILTIN 0
+
#endif /* math-use-builtins.h */
@@ -23,11 +23,14 @@
#include <math_private.h>
#include <libm-alias-double.h>
#include <stdint.h>
-
+#include <math-use-builtins.h>
double
__round (double x)
{
+#if USE_ROUND_BUILTIN
+ return __builtin_round (x);
+#else
int64_t i0, j0;
EXTRACT_WORDS64 (i0, x);
@@ -62,5 +65,6 @@ __round (double x)
INSERT_WORDS64 (x, i0);
return x;
+#endif /* USE_ROUND_BUILTIN */
}
libm_alias_double (__round, round)
@@ -150,6 +150,8 @@
#define USE_CEILL_BUILTIN USE_CEILF128_BUILTIN
#undef USE_TRUNCL_BUILTIN
#define USE_TRUNCL_BUILTIN USE_TRUNCF128_BUILTIN
+#undef USE_ROUNDL_BUILTIN
+#define USE_ROUNDL_BUILTIN USE_ROUNDF128_BUILTIN
/* IEEE function renames. */
#define __ieee754_acoshl __ieee754_acoshf128
@@ -358,6 +360,7 @@
#define __builtin_floorl __builtin_floorf128
#define __builtin_ceill __builtin_ceilf128
#define __builtin_truncl __builtin_truncf128
+#define __builtin_roundl __builtin_roundf128
/* Get the constant suffix from bits/floatn-compat.h. */
#define L(x) __f128 (x)
@@ -22,11 +22,14 @@
#include <math_private.h>
#include <libm-alias-float.h>
-
+#include <math-use-builtins.h>
float
__roundf (float x)
{
+#if USE_ROUNDF_BUILTIN
+ return __builtin_roundf (x);
+#else
int32_t i0, j0;
GET_FLOAT_WORD (i0, x);
@@ -61,5 +64,6 @@ __roundf (float x)
SET_FLOAT_WORD (x, i0);
return x;
+#endif /* USE_ROUNDF_BUILTIN */
}
libm_alias_float (__round, round)
@@ -23,11 +23,14 @@
#include <math_private.h>
#include <libm-alias-ldouble.h>
-
+#include <math-use-builtins.h>
_Float128
__roundl (_Float128 x)
{
+#if USE_ROUNDL_BUILTIN
+ return __builtin_roundl (x);
+#else
int32_t j0;
uint64_t i1, i0;
@@ -78,5 +81,6 @@ __roundl (_Float128 x)
SET_LDOUBLE_WORDS64 (x, i0, i1);
return x;
+#endif /* USE_ROUNDL_BUILTIN */
}
libm_alias_ldouble (__round, round)
@@ -46,18 +46,24 @@
# define USE_TRUNCF_BUILTIN 1
# define USE_TRUNCL_BUILTIN 1
+# define USE_ROUND_BUILTIN 1
+# define USE_ROUNDF_BUILTIN 1
+# define USE_ROUNDL_BUILTIN 1
+
# if __GNUC_PREREQ (8, 1)
# define USE_NEARBYINTF128_BUILTIN 1
# define USE_RINTF128_BUILTIN 1
# define USE_FLOORF128_BUILTIN 1
# define USE_CEILF128_BUILTIN 1
# define USE_TRUNCF128_BUILTIN 1
+# define USE_ROUNDF128_BUILTIN 1
# else
# define USE_NEARBYINTF128_BUILTIN 0
# define USE_RINTF128_BUILTIN 0
# define USE_FLOORF128_BUILTIN 0
# define USE_CEILF128_BUILTIN 0
# define USE_TRUNCF128_BUILTIN 0
+# define USE_ROUNDF128_BUILTIN 0
# endif
#else