diff options
Diffstat (limited to 'gcc')
| -rw-r--r-- | gcc/config/aarch64/aarch64-simd-builtins.def | 4 | ||||
| -rw-r--r-- | gcc/config/aarch64/arm_neon.h | 66 |
2 files changed, 28 insertions, 42 deletions
diff --git a/gcc/config/aarch64/aarch64-simd-builtins.def b/gcc/config/aarch64/aarch64-simd-builtins.def index 0d5d22e..e57dee9 100644 --- a/gcc/config/aarch64/aarch64-simd-builtins.def +++ b/gcc/config/aarch64/aarch64-simd-builtins.def @@ -220,13 +220,17 @@ /* Implemented by aarch64_<sur><addsub>hn<mode>. */ BUILTIN_VQN (BINOP, addhn, 0, NONE) + BUILTIN_VQN (BINOPU, addhn, 0, NONE) BUILTIN_VQN (BINOP, subhn, 0, NONE) BUILTIN_VQN (BINOP, raddhn, 0, NONE) + BUILTIN_VQN (BINOPU, raddhn, 0, NONE) BUILTIN_VQN (BINOP, rsubhn, 0, NONE) /* Implemented by aarch64_<sur><addsub>hn2<mode>. */ BUILTIN_VQN (TERNOP, addhn2, 0, NONE) + BUILTIN_VQN (TERNOPU, addhn2, 0, NONE) BUILTIN_VQN (TERNOP, subhn2, 0, NONE) BUILTIN_VQN (TERNOP, raddhn2, 0, NONE) + BUILTIN_VQN (TERNOPU, raddhn2, 0, NONE) BUILTIN_VQN (TERNOP, rsubhn2, 0, NONE) /* Implemented by aarch64_<us>xtl<mode>. */ diff --git a/gcc/config/aarch64/arm_neon.h b/gcc/config/aarch64/arm_neon.h index 05af357..fa404e0 100644 --- a/gcc/config/aarch64/arm_neon.h +++ b/gcc/config/aarch64/arm_neon.h @@ -713,186 +713,168 @@ __extension__ extern __inline int8x8_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vaddhn_s16 (int16x8_t __a, int16x8_t __b) { - return (int8x8_t) __builtin_aarch64_addhnv8hi (__a, __b); + return __builtin_aarch64_addhnv8hi (__a, __b); } __extension__ extern __inline int16x4_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vaddhn_s32 (int32x4_t __a, int32x4_t __b) { - return (int16x4_t) __builtin_aarch64_addhnv4si (__a, __b); + return __builtin_aarch64_addhnv4si (__a, __b); } __extension__ extern __inline int32x2_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vaddhn_s64 (int64x2_t __a, int64x2_t __b) { - return (int32x2_t) __builtin_aarch64_addhnv2di (__a, __b); + return __builtin_aarch64_addhnv2di (__a, __b); } __extension__ extern __inline uint8x8_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vaddhn_u16 (uint16x8_t __a, uint16x8_t __b) { - return (uint8x8_t) __builtin_aarch64_addhnv8hi ((int16x8_t) __a, - (int16x8_t) __b); + return __builtin_aarch64_addhnv8hi_uuu (__a, __b); } __extension__ extern __inline uint16x4_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vaddhn_u32 (uint32x4_t __a, uint32x4_t __b) { - return (uint16x4_t) __builtin_aarch64_addhnv4si ((int32x4_t) __a, - (int32x4_t) __b); + return __builtin_aarch64_addhnv4si_uuu (__a, __b); } __extension__ extern __inline uint32x2_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vaddhn_u64 (uint64x2_t __a, uint64x2_t __b) { - return (uint32x2_t) __builtin_aarch64_addhnv2di ((int64x2_t) __a, - (int64x2_t) __b); + return __builtin_aarch64_addhnv2di_uuu (__a, __b); } __extension__ extern __inline int8x8_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vraddhn_s16 (int16x8_t __a, int16x8_t __b) { - return (int8x8_t) __builtin_aarch64_raddhnv8hi (__a, __b); + return __builtin_aarch64_raddhnv8hi (__a, __b); } __extension__ extern __inline int16x4_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vraddhn_s32 (int32x4_t __a, int32x4_t __b) { - return (int16x4_t) __builtin_aarch64_raddhnv4si (__a, __b); + return __builtin_aarch64_raddhnv4si (__a, __b); } __extension__ extern __inline int32x2_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vraddhn_s64 (int64x2_t __a, int64x2_t __b) { - return (int32x2_t) __builtin_aarch64_raddhnv2di (__a, __b); + return __builtin_aarch64_raddhnv2di (__a, __b); } __extension__ extern __inline uint8x8_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vraddhn_u16 (uint16x8_t __a, uint16x8_t __b) { - return (uint8x8_t) __builtin_aarch64_raddhnv8hi ((int16x8_t) __a, - (int16x8_t) __b); + return __builtin_aarch64_raddhnv8hi_uuu (__a, __b); } __extension__ extern __inline uint16x4_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vraddhn_u32 (uint32x4_t __a, uint32x4_t __b) { - return (uint16x4_t) __builtin_aarch64_raddhnv4si ((int32x4_t) __a, - (int32x4_t) __b); + return __builtin_aarch64_raddhnv4si_uuu (__a, __b); } __extension__ extern __inline uint32x2_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vraddhn_u64 (uint64x2_t __a, uint64x2_t __b) { - return (uint32x2_t) __builtin_aarch64_raddhnv2di ((int64x2_t) __a, - (int64x2_t) __b); + return __builtin_aarch64_raddhnv2di_uuu (__a, __b); } __extension__ extern __inline int8x16_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vaddhn_high_s16 (int8x8_t __a, int16x8_t __b, int16x8_t __c) { - return (int8x16_t) __builtin_aarch64_addhn2v8hi (__a, __b, __c); + return __builtin_aarch64_addhn2v8hi (__a, __b, __c); } __extension__ extern __inline int16x8_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vaddhn_high_s32 (int16x4_t __a, int32x4_t __b, int32x4_t __c) { - return (int16x8_t) __builtin_aarch64_addhn2v4si (__a, __b, __c); + return __builtin_aarch64_addhn2v4si (__a, __b, __c); } __extension__ extern __inline int32x4_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vaddhn_high_s64 (int32x2_t __a, int64x2_t __b, int64x2_t __c) { - return (int32x4_t) __builtin_aarch64_addhn2v2di (__a, __b, __c); + return __builtin_aarch64_addhn2v2di (__a, __b, __c); } __extension__ extern __inline uint8x16_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vaddhn_high_u16 (uint8x8_t __a, uint16x8_t __b, uint16x8_t __c) { - return (uint8x16_t) __builtin_aarch64_addhn2v8hi ((int8x8_t) __a, - (int16x8_t) __b, - (int16x8_t) __c); + return __builtin_aarch64_addhn2v8hi_uuuu (__a, __b, __c); } __extension__ extern __inline uint16x8_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vaddhn_high_u32 (uint16x4_t __a, uint32x4_t __b, uint32x4_t __c) { - return (uint16x8_t) __builtin_aarch64_addhn2v4si ((int16x4_t) __a, - (int32x4_t) __b, - (int32x4_t) __c); + return __builtin_aarch64_addhn2v4si_uuuu (__a, __b, __c); } __extension__ extern __inline uint32x4_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vaddhn_high_u64 (uint32x2_t __a, uint64x2_t __b, uint64x2_t __c) { - return (uint32x4_t) __builtin_aarch64_addhn2v2di ((int32x2_t) __a, - (int64x2_t) __b, - (int64x2_t) __c); + return __builtin_aarch64_addhn2v2di_uuuu (__a, __b, __c); } __extension__ extern __inline int8x16_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vraddhn_high_s16 (int8x8_t __a, int16x8_t __b, int16x8_t __c) { - return (int8x16_t) __builtin_aarch64_raddhn2v8hi (__a, __b, __c); + return __builtin_aarch64_raddhn2v8hi (__a, __b, __c); } __extension__ extern __inline int16x8_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vraddhn_high_s32 (int16x4_t __a, int32x4_t __b, int32x4_t __c) { - return (int16x8_t) __builtin_aarch64_raddhn2v4si (__a, __b, __c); + return __builtin_aarch64_raddhn2v4si (__a, __b, __c); } __extension__ extern __inline int32x4_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vraddhn_high_s64 (int32x2_t __a, int64x2_t __b, int64x2_t __c) { - return (int32x4_t) __builtin_aarch64_raddhn2v2di (__a, __b, __c); + return __builtin_aarch64_raddhn2v2di (__a, __b, __c); } __extension__ extern __inline uint8x16_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vraddhn_high_u16 (uint8x8_t __a, uint16x8_t __b, uint16x8_t __c) { - return (uint8x16_t) __builtin_aarch64_raddhn2v8hi ((int8x8_t) __a, - (int16x8_t) __b, - (int16x8_t) __c); + return __builtin_aarch64_raddhn2v8hi_uuuu (__a, __b, __c); } __extension__ extern __inline uint16x8_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vraddhn_high_u32 (uint16x4_t __a, uint32x4_t __b, uint32x4_t __c) { - return (uint16x8_t) __builtin_aarch64_raddhn2v4si ((int16x4_t) __a, - (int32x4_t) __b, - (int32x4_t) __c); + return __builtin_aarch64_raddhn2v4si_uuuu (__a, __b, __c); } __extension__ extern __inline uint32x4_t __attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) vraddhn_high_u64 (uint32x2_t __a, uint64x2_t __b, uint64x2_t __c) { - return (uint32x4_t) __builtin_aarch64_raddhn2v2di ((int32x2_t) __a, - (int64x2_t) __b, - (int64x2_t) __c); + return __builtin_aarch64_raddhn2v2di_uuuu (__a, __b, __c); } __extension__ extern __inline float32x2_t |