These intrinsics provide operations including
multiplication:
Vector multiply: vmul
-> Vr[i] := Va[i] * Vb[i]
int8x8_t vmul_s8(int8x8_t a, int8x8_t b); // VMUL.I8 d0,d0,d0
int16x4_t vmul_s16(int16x4_t a, int16x4_t b); // VMUL.I16 d0,d0,d0
int32x2_t vmul_s32(int32x2_t a, int32x2_t b); // VMUL.I32 d0,d0,d0
float32x2_t vmul_f32(float32x2_t a, float32x2_t b); // VMUL.F32 d0,d0,d0
uint8x8_t vmul_u8(uint8x8_t a, uint8x8_t b); // VMUL.I8 d0,d0,d0
uint16x4_t vmul_u16(uint16x4_t a, uint16x4_t b); // VMUL.I16 d0,d0,d0
uint32x2_t vmul_u32(uint32x2_t a, uint32x2_t b); // VMUL.I32 d0,d0,d0
poly8x8_t vmul_p8(poly8x8_t a, poly8x8_t b); // VMUL.P8 d0,d0,d0
int8x16_t vmulq_s8(int8x16_t a, int8x16_t b); // VMUL.I8 q0,q0,q0
int16x8_t vmulq_s16(int16x8_t a, int16x8_t b); // VMUL.I16 q0,q0,q0
int32x4_t vmulq_s32(int32x4_t a, int32x4_t b); // VMUL.I32 q0,q0,q0
float32x4_t vmulq_f32(float32x4_t a, float32x4_t b); // VMUL.F32 q0,q0,q0
uint8x16_t vmulq_u8(uint8x16_t a, uint8x16_t b); // VMUL.I8 q0,q0,q0
uint16x8_t vmulq_u16(uint16x8_t a, uint16x8_t b); // VMUL.I16 q0,q0,q0
uint32x4_t vmulq_u32(uint32x4_t a, uint32x4_t b); // VMUL.I32 q0,q0,q0
poly8x16_t vmulq_p8(poly8x16_t a, poly8x16_t b); // VMUL.P8 q0,q0,q0
Vector multiply accumulate:
vmla -> Vr[i] := Va[i] + Vb[i] * Vc[i]
int8x8_t vmla_s8(int8x8_t a, int8x8_t b, int8x8_t c); // VMLA.I8 d0,d0,d0
int16x4_t vmla_s16(int16x4_t a, int16x4_t b, int16x4_t c); // VMLA.I16 d0,d0,d0
int32x2_t vmla_s32(int32x2_t a, int32x2_t b, int32x2_t c); // VMLA.I32 d0,d0,d0
float32x2_t vmla_f32(float32x2_t a, float32x2_t b, float32x2_t c); // VMLA.F32 d0,d0,d0
uint8x8_t vmla_u8(uint8x8_t a, uint8x8_t b, uint8x8_t c); // VMLA.I8 d0,d0,d0
uint16x4_t vmla_u16(uint16x4_t a, uint16x4_t b, uint16x4_t c); // VMLA.I16 d0,d0,d0
uint32x2_t vmla_u32(uint32x2_t a, uint32x2_t b, uint32x2_t c); // VMLA.I32 d0,d0,d0
int8x16_t vmlaq_s8(int8x16_t a, int8x16_t b, int8x16_t c); // VMLA.I8 q0,q0,q0
int16x8_t vmlaq_s16(int16x8_t a, int16x8_t b, int16x8_t c); // VMLA.I16 q0,q0,q0
int32x4_t vmlaq_s32(int32x4_t a, int32x4_t b, int32x4_t c); // VMLA.I32 q0,q0,q0
float32x4_t vmlaq_f32(float32x4_t a, float32x4_t b, float32x4_t c); // VMLA.F32 q0,q0,q0
uint8x16_t vmlaq_u8(uint8x16_t a, uint8x16_t b, uint8x16_t c); // VMLA.I8 q0,q0,q0
uint16x8_t vmlaq_u16(uint16x8_t a, uint16x8_t b, uint16x8_t c); // VMLA.I16 q0,q0,q0
uint32x4_t vmlaq_u32(uint32x4_t a, uint32x4_t b, uint32x4_t c); // VMLA.I32 q0,q0,q0
Vector multiply accumulate
long: vmla -> Vr[i] := Va[i] + Vb[i] * Vc[i]
int16x8_t vmlal_s8(int16x8_t a, int8x8_t b, int8x8_t c); // VMLAL.S8 q0,d0,d0
int32x4_t vmlal_s16(int32x4_t a, int16x4_t b, int16x4_t c); // VMLAL.S16 q0,d0,d0
int64x2_t vmlal_s32(int64x2_t a, int32x2_t b, int32x2_t c); // VMLAL.S32 q0,d0,d0
uint16x8_t vmlal_u8(uint16x8_t a, uint8x8_t b, uint8x8_t c); // VMLAL.U8 q0,d0,d0
uint32x4_t vmlal_u16(uint32x4_t a, uint16x4_t b, uint16x4_t c); // VMLAL.U16 q0,d0,d0
uint64x2_t vmlal_u32(uint64x2_t a, uint32x2_t b, uint32x2_t c); // VMLAL.U32 q0,d0,d0
Vector multiply subtract:
vmls -> Vr[i] := Va[i] - Vb[i] * Vc[i]
int8x8_t vmls_s8(int8x8_t a, int8x8_t b, int8x8_t c); // VMLS.I8 d0,d0,d0
int16x4_t vmls_s16(int16x4_t a, int16x4_t b, int16x4_t c); // VMLS.I16 d0,d0,d0
int32x2_t vmls_s32(int32x2_t a, int32x2_t b, int32x2_t c); // VMLS.I32 d0,d0,d0
float32x2_t vmls_f32(float32x2_t a, float32x2_t b, float32x2_t c); // VMLS.F32 d0,d0,d0
uint8x8_t vmls_u8(uint8x8_t a, uint8x8_t b, uint8x8_t c); // VMLS.I8 d0,d0,d0
uint16x4_t vmls_u16(uint16x4_t a, uint16x4_t b, uint16x4_t c); // VMLS.I16 d0,d0,d0
uint32x2_t vmls_u32(uint32x2_t a, uint32x2_t b, uint32x2_t c); // VMLS.I32 d0,d0,d0
int8x16_t vmlsq_s8(int8x16_t a, int8x16_t b, int8x16_t c); // VMLS.I8 q0,q0,q0
int16x8_t vmlsq_s16(int16x8_t a, int16x8_t b, int16x8_t c); // VMLS.I16 q0,q0,q0
int32x4_t vmlsq_s32(int32x4_t a, int32x4_t b, int32x4_t c); // VMLS.I32 q0,q0,q0
float32x4_t vmlsq_f32(float32x4_t a, float32x4_t b, float32x4_t c); // VMLS.F32 q0,q0,q0
uint8x16_t vmlsq_u8(uint8x16_t a, uint8x16_t b, uint8x16_t c); // VMLS.I8 q0,q0,q0
uint16x8_t vmlsq_u16(uint16x8_t a, uint16x8_t b, uint16x8_t c); // VMLS.I16 q0,q0,q0
uint32x4_t vmlsq_u32(uint32x4_t a, uint32x4_t b, uint32x4_t c); // VMLS.I32 q0,q0,q0
Vector multiply subtract
long
int16x8_t vmlsl_s8(int16x8_t a, int8x8_t b, int8x8_t c); // VMLSL.S8 q0,d0,d0
int32x4_t vmlsl_s16(int32x4_t a, int16x4_t b, int16x4_t c); // VMLSL.S16 q0,d0,d0
int64x2_t vmlsl_s32(int64x2_t a, int32x2_t b, int32x2_t c); // VMLSL.S32 q0,d0,d0
uint16x8_t vmlsl_u8(uint16x8_t a, uint8x8_t b, uint8x8_t c); // VMLSL.U8 q0,d0,d0
uint32x4_t vmlsl_u16(uint32x4_t a, uint16x4_t b, uint16x4_t c); // VMLSL.U16 q0,d0,d0
uint64x2_t vmlsl_u32(uint64x2_t a, uint32x2_t b, uint32x2_t c); // VMLSL.U32 q0,d0,d0
Vector saturating doubling
multiply high
int16x4_t vqdmulh_s16(int16x4_t a, int16x4_t b); // VQDMULH.S16 d0,d0,d0
int32x2_t vqdmulh_s32(int32x2_t a, int32x2_t b); // VQDMULH.S32 d0,d0,d0
int16x8_t vqdmulhq_s16(int16x8_t a, int16x8_t b); // VQDMULH.S16 q0,q0,q0
int32x4_t vqdmulhq_s32(int32x4_t a, int32x4_t b); // VQDMULH.S32 q0,q0,q0
Vector saturating rounding
doubling multiply high
int16x4_t vqrdmulh_s16(int16x4_t a, int16x4_t b); // VQRDMULH.S16 d0,d0,d0
int32x2_t vqrdmulh_s32(int32x2_t a, int32x2_t b); // VQRDMULH.S32 d0,d0,d0
int16x8_t vqrdmulhq_s16(int16x8_t a, int16x8_t b); // VQRDMULH.S16 q0,q0,q0
int32x4_t vqrdmulhq_s32(int32x4_t a, int32x4_t b); // VQRDMULH.S32 q0,q0,q0
Vector saturating doubling
multiply accumulate long
int32x4_t vqdmlal_s16(int32x4_t a, int16x4_t b, int16x4_t c); // VQDMLAL.S16 q0,d0,d0
int64x2_t vqdmlal_s32(int64x2_t a, int32x2_t b, int32x2_t c); // VQDMLAL.S32 q0,d0,d0
Vector saturating doubling
multiply subtract long
int32x4_t vqdmlsl_s16(int32x4_t a, int16x4_t b, int16x4_t c); // VQDMLSL.S16 q0,d0,d0
int64x2_t vqdmlsl_s32(int64x2_t a, int32x2_t b, int32x2_t c); // VQDMLSL.S32 q0,d0,d0
int16x8_t vmull_s8(int8x8_t a, int8x8_t b); // VMULL.S8 q0,d0,d0
int32x4_t vmull_s16(int16x4_t a, int16x4_t b); // VMULL.S16 q0,d0,d0
int64x2_t vmull_s32(int32x2_t a, int32x2_t b); // VMULL.S32 q0,d0,d0
uint16x8_t vmull_u8(uint8x8_t a, uint8x8_t b); // VMULL.U8 q0,d0,d0
uint32x4_t vmull_u16(uint16x4_t a, uint16x4_t b); // VMULL.U16 q0,d0,d0
uint64x2_t vmull_u32(uint32x2_t a, uint32x2_t b); // VMULL.U32 q0,d0,d0
poly16x8_t vmull_p8(poly8x8_t a, poly8x8_t b); // VMULL.P8 q0,d0,d0
Vector saturating doubling
long multiply
int32x4_t vqdmull_s16(int16x4_t a, int16x4_t b); // VQDMULL.S16 q0,d0,d0
int64x2_t vqdmull_s32(int32x2_t a, int32x2_t b); // VQDMULL.S32 q0,d0,d0
PDF version