diff options
| -rw-r--r-- | lib/Headers/arm_neon.h | 54 |
1 files changed, 54 insertions, 0 deletions
diff --git a/lib/Headers/arm_neon.h b/lib/Headers/arm_neon.h index 00b0d9e30c..ba571a2f84 100644 --- a/lib/Headers/arm_neon.h +++ b/lib/Headers/arm_neon.h @@ -173,6 +173,14 @@ typedef struct __poly16x8_t { __ai uint32x4_t op##_u16(uint16x4_t a, uint16x4_t b) { return (uint32x4_t){ builtin(a.val, b.val) }; } \ __ai uint64x2_t op##_u32(uint32x2_t a, uint32x2_t b) { return (uint64x2_t){ builtin(a.val, b.val) }; } +#define INTTYPES_WIDENING_MUL(op, builtin) \ + __ai int16x8_t op##_s8(int16x8_t a, int8x8_t b, int8x8_t c) { return (int16x8_t){ builtin(a.val, b.val, c.val) }; } \ + __ai int32x4_t op##_s16(int32x4_t a, int16x4_t b, int16x4_t c) { return (int32x4_t){ builtin(a.val, b.val, c.val) }; } \ + __ai int64x2_t op##_s32(int64x2_t a, int32x2_t b, int32x2_t c) { return (int64x2_t){ builtin(a.val, b.val, c.val) }; } \ + __ai uint16x8_t op##_u8(uint16x8_t a, uint8x8_t b, uint8x8_t c) { return (uint16x8_t){ builtin(a.val, b.val, c.val) }; } \ + __ai uint32x4_t op##_u16(uint32x4_t a, uint16x4_t b, uint16x4_t c) { return (uint32x4_t){ builtin(a.val, b.val, c.val) }; } \ + __ai uint64x2_t op##_u32(uint64x2_t a, uint32x2_t b, uint32x2_t c) { return (uint64x2_t){ builtin(a.val, b.val, c.val) }; } + #define INTTYPES_NARROWING(op, builtin) \ __ai int8x8_t op##_s16(int16x8_t a, int16x8_t b) { return (int8x8_t){ builtin(a.val, b.val) }; } \ __ai int16x4_t op##_s32(int32x4_t a, int32x4_t b) { return (int16x4_t){ builtin(a.val, b.val) }; } \ @@ -201,6 +209,26 @@ typedef struct __poly16x8_t { __ai int64x2_t op##q_s64(int64x2_t a, int64x2_t b) { return (int64x2_t){ builtin(a.val, b.val) }; } \ __ai uint64x2_t op##q_u64(uint64x2_t a, uint64x2_t b) { return (uint64x2_t){ builtin(a.val, b.val) }; } +#define FLOATTYPES_CMP(op, builtin) \ + __ai uint32x2_t op##_f32(float32x2_t a, float32x2_t b) { return (uint32x2_t){ builtin(a.val, b.val) }; } \ + __ai uint32x4_t op##q_f32(float32x4_t a, float32x4_t b) { return (uint32x4_t){ builtin(a.val, b.val) }; } + +#define INT_FLOAT_CMP_OP(op, builtin, cc) \ + __ai uint8x8_t op##_s8(int8x8_t a, int8x8_t b) { return (uint8x8_t){(__neon_uint8x8_t)(a.val cc b.val)}; } \ + __ai uint16x4_t op##_s16(int16x4_t a, int16x4_t b) { return (uint16x4_t){(__neon_uint16x4_t)(a.val cc b.val)}; } \ + __ai uint32x2_t op##_s32(int32x2_t a, int32x2_t b) { return (uint32x2_t){(__neon_uint32x2_t)(a.val cc b.val)}; } \ + __ai uint32x2_t op##_f32(float32x2_t a, float32x2_t b) { return (uint32x2_t){(__neon_uint32x2_t)(a.val cc b.val)}; } \ + __ai uint8x8_t op##_u8(uint8x8_t a, uint8x8_t b) { return (uint8x8_t){a.val cc b.val}; } \ + __ai uint16x4_t op##_u16(uint16x4_t a, uint16x4_t b) { return (uint16x4_t){a.val cc b.val}; } \ + __ai uint32x2_t op##_u32(uint32x2_t a, uint32x2_t b) { return (uint32x2_t){a.val cc b.val}; } \ + __ai uint8x16_t op##q_s8(int8x16_t a, int8x16_t b) { return (uint8x16_t){(__neon_uint8x16_t)(a.val cc b.val)}; } \ + __ai uint16x8_t op##q_s16(int16x8_t a, int16x8_t b) { return (uint16x8_t){(__neon_uint16x8_t)(a.val cc b.val)}; } \ + __ai uint32x4_t op##q_s32(int32x4_t a, int32x4_t b) { return (uint32x4_t){(__neon_uint32x4_t)(a.val cc b.val)}; } \ + __ai uint32x4_t op##q_f32(float32x4_t a, float32x4_t b) { return (uint32x4_t){(__neon_uint32x4_t)(a.val cc b.val)}; } \ + __ai uint8x16_t op##q_u8(uint8x16_t a, uint8x16_t b) { return (uint8x16_t){a.val cc b.val}; } \ + __ai uint16x8_t op##q_u16(uint16x8_t a, uint16x8_t b) { return (uint16x8_t){a.val cc b.val}; } \ + __ai uint32x4_t op##q_u32(uint32x4_t a, uint32x4_t b) { return (uint32x4_t){a.val cc b.val}; } + // vector add __ai int8x8_t vadd_s8(int8x8_t a, int8x8_t b) { return (int8x8_t){a.val + b.val}; } __ai int16x4_t vadd_s16(int16x4_t a, int16x4_t b) { return (int16x4_t){a.val + b.val}; } @@ -254,6 +282,8 @@ INTTYPES_NARROWING(vraddhn, __builtin_neon_vraddhn) // multiple accumulate long // multiple subtract long +INTTYPES_WIDENING_MUL(vmlal, __builtin_neon_vmlal) +INTTYPES_WIDENING_MUL(vmlsl, __builtin_neon_vmlsl) // saturating doubling multiply high // saturating rounding doubling multiply high @@ -269,22 +299,46 @@ __ai poly16x8_t vmull_p8(poly8x8_t a, poly8x8_t b) { return (poly16x8_t){ __buil // saturating doubling long multiply // subtract + // long subtract +INTTYPES_WIDENING(vsubl, __builtin_neon_vsubl) + // wide subtract + // saturating subtract +INTTYPES_ADD_32(vqsub, __builtin_neon_vqsub) +INTTYPES_ADD_64(vqsub, __builtin_neon_vqsub) + // halving subtract +INTTYPES_ADD_32(vhsub, __builtin_neon_vhsub) + // subtract high half // rounding subtract high half +INTTYPES_NARROWING(vsubhn, __builtin_neon_vsubhn) +INTTYPES_NARROWING(vrsubhn, __builtin_neon_vrsubhn) // compare eq // compare ge // compare le // compare gt // compare lt +INT_FLOAT_CMP_OP(vceq, __builtin_neon, ==) +INT_FLOAT_CMP_OP(vcge, __builtin_neon, >=) +INT_FLOAT_CMP_OP(vcle, __builtin_neon, <=) +INT_FLOAT_CMP_OP(vcgt, __builtin_neon, >) +INT_FLOAT_CMP_OP(vclt, __builtin_neon, <) + +// compare eq-poly + // compare abs ge // compare abs le // compare abs gt // compare abs lt +FLOATTYPES_CMP(vcage, __builtin_neon_vcage) +FLOATTYPES_CMP(vcale, __builtin_neon_vcale) +FLOATTYPES_CMP(vcagt, __builtin_neon_vcagt) +FLOATTYPES_CMP(vcalt, __builtin_neon_vcalt) + // test bits // abs diff |