18.7 NEON intrinsics for comparison

A range of comparison intrinsics are provided. If the comparison is true for a lane, the result in that lane is all bits set to one. If the comparison is false for a lane, all bits are set to zero.

The return type is an unsigned integer. This means that you can use the result of a comparison as the first argument for the vbsl intrinsics.

Vector compare equal

uint8x8_t  vceq_s8(int8x8_t a, int8x8_t b);         // VCEQ.I8 d0, d0, d0 
uint16x4_t vceq_s16(int16x4_t a, int16x4_t b);      // VCEQ.I16 d0, d0, d0
uint32x2_t vceq_s32(int32x2_t a, int32x2_t b);      // VCEQ.I32 d0, d0, d0
uint32x2_t vceq_f32(float32x2_t a, float32x2_t b);  // VCEQ.F32 d0, d0, d0
uint8x8_t  vceq_u8(uint8x8_t a, uint8x8_t b);       // VCEQ.I8 d0, d0, d0 
uint16x4_t vceq_u16(uint16x4_t a, uint16x4_t b);    // VCEQ.I16 d0, d0, d0
uint32x2_t vceq_u32(uint32x2_t a, uint32x2_t b);    // VCEQ.I32 d0, d0, d0
uint8x8_t  vceq_p8(poly8x8_t a, poly8x8_t b);       // VCEQ.I8 d0, d0, d0 
uint8x16_t vceqq_s8(int8x16_t a, int8x16_t b);      // VCEQ.I8 q0, q0, q0 
uint16x8_t vceqq_s16(int16x8_t a, int16x8_t b);     // VCEQ.I16 q0, q0, q0
uint32x4_t vceqq_s32(int32x4_t a, int32x4_t b);     // VCEQ.I32 q0, q0, q0
uint32x4_t vceqq_f32(float32x4_t a, float32x4_t b); // VCEQ.F32 q0, q0, q0
uint8x16_t vceqq_u8(uint8x16_t a, uint8x16_t b);    // VCEQ.I8 q0, q0, q0 
uint16x8_t vceqq_u16(uint16x8_t a, uint16x8_t b);   // VCEQ.I16 q0, q0, q0
uint32x4_t vceqq_u32(uint32x4_t a, uint32x4_t b);   // VCEQ.I32 q0, q0, q0
uint8x16_t vceqq_p8(poly8x16_t a, poly8x16_t b);    // VCEQ.I8 q0, q0, q0 

Vector compare greater-than or equal

uint8x8_t  vcge_s8(int8x8_t a, int8x8_t b);         // VCGE.S8 d0, d0, d0 
uint16x4_t vcge_s16(int16x4_t a, int16x4_t b);      // VCGE.S16 d0, d0, d0
uint32x2_t vcge_s32(int32x2_t a, int32x2_t b);      // VCGE.S32 d0, d0, d0
uint32x2_t vcge_f32(float32x2_t a, float32x2_t b);  // VCGE.F32 d0, d0, d0
uint8x8_t  vcge_u8(uint8x8_t a, uint8x8_t b);       // VCGE.U8 d0, d0, d0 
uint16x4_t vcge_u16(uint16x4_t a, uint16x4_t b);    // VCGE.U16 d0, d0, d0
uint32x2_t vcge_u32(uint32x2_t a, uint32x2_t b);    // VCGE.U32 d0, d0, d0
uint8x16_t vcgeq_s8(int8x16_t a, int8x16_t b);      // VCGE.S8 q0, q0, q0 
uint16x8_t vcgeq_s16(int16x8_t a, int16x8_t b);     // VCGE.S16 q0, q0, q0
uint32x4_t vcgeq_s32(int32x4_t a, int32x4_t b);     // VCGE.S32 q0, q0, q0
uint32x4_t vcgeq_f32(float32x4_t a, float32x4_t b); // VCGE.F32 q0, q0, q0
uint8x16_t vcgeq_u8(uint8x16_t a, uint8x16_t b);    // VCGE.U8 q0, q0, q0 
uint16x8_t vcgeq_u16(uint16x8_t a, uint16x8_t b);   // VCGE.U16 q0, q0, q0
uint32x4_t vcgeq_u32(uint32x4_t a, uint32x4_t b);   // VCGE.U32 q0, q0, q0

Vector compare less-than or equal

uint8x8_t  vcle_s8(int8x8_t a, int8x8_t b);         // VCGE.S8 d0, d0, d0 
uint16x4_t vcle_s16(int16x4_t a, int16x4_t b);      // VCGE.S16 d0, d0, d0
uint32x2_t vcle_s32(int32x2_t a, int32x2_t b);      // VCGE.S32 d0, d0, d0
uint32x2_t vcle_f32(float32x2_t a, float32x2_t b);  // VCGE.F32 d0, d0, d0
uint8x8_t  vcle_u8(uint8x8_t a, uint8x8_t b);       // VCGE.U8 d0, d0, d0 
uint16x4_t vcle_u16(uint16x4_t a, uint16x4_t b);    // VCGE.U16 d0, d0, d0
uint32x2_t vcle_u32(uint32x2_t a, uint32x2_t b);    // VCGE.U32 d0, d0, d0
uint8x16_t vcleq_s8(int8x16_t a, int8x16_t b);      // VCGE.S8 q0, q0, q0 
uint16x8_t vcleq_s16(int16x8_t a, int16x8_t b);     // VCGE.S16 q0, q0, q0
uint32x4_t vcleq_s32(int32x4_t a, int32x4_t b);     // VCGE.S32 q0, q0, q0
uint32x4_t vcleq_f32(float32x4_t a, float32x4_t b); // VCGE.F32 q0, q0, q0
uint8x16_t vcleq_u8(uint8x16_t a, uint8x16_t b);    // VCGE.U8 q0, q0, q0 
uint16x8_t vcleq_u16(uint16x8_t a, uint16x8_t b);   // VCGE.U16 q0, q0, q0
uint32x4_t vcleq_u32(uint32x4_t a, uint32x4_t b);   // VCGE.U32 q0, q0, q0

Vector compare greater-than

uint8x8_t  vcgt_s8(int8x8_t a, int8x8_t b);         // VCGT.S8 d0, d0, d0 
uint16x4_t vcgt_s16(int16x4_t a, int16x4_t b);      // VCGT.S16 d0, d0, d0
uint32x2_t vcgt_s32(int32x2_t a, int32x2_t b);      // VCGT.S32 d0, d0, d0
uint32x2_t vcgt_f32(float32x2_t a, float32x2_t b);  // VCGT.F32 d0, d0, d0
uint8x8_t  vcgt_u8(uint8x8_t a, uint8x8_t b);       // VCGT.U8 d0, d0, d0 
uint16x4_t vcgt_u16(uint16x4_t a, uint16x4_t b);    // VCGT.U16 d0, d0, d0
uint32x2_t vcgt_u32(uint32x2_t a, uint32x2_t b);    // VCGT.U32 d0, d0, d0
uint8x16_t vcgtq_s8(int8x16_t a, int8x16_t b);      // VCGT.S8 q0, q0, q0 
uint16x8_t vcgtq_s16(int16x8_t a, int16x8_t b);     // VCGT.S16 q0, q0, q0
uint32x4_t vcgtq_s32(int32x4_t a, int32x4_t b);     // VCGT.S32 q0, q0, q0
uint32x4_t vcgtq_f32(float32x4_t a, float32x4_t b); // VCGT.F32 q0, q0, q0
uint8x16_t vcgtq_u8(uint8x16_t a, uint8x16_t b);    // VCGT.U8 q0, q0, q0 
uint16x8_t vcgtq_u16(uint16x8_t a, uint16x8_t b);   // VCGT.U16 q0, q0, q0
uint32x4_t vcgtq_u32(uint32x4_t a, uint32x4_t b);   // VCGT.U32 q0, q0, q0

Vector compare less-than

uint8x8_t  vclt_s8(int8x8_t a, int8x8_t b);         // VCGT.S8 d0, d0, d0 
uint16x4_t vclt_s16(int16x4_t a, int16x4_t b);      // VCGT.S16 d0, d0, d0
uint32x2_t vclt_s32(int32x2_t a, int32x2_t b);      // VCGT.S32 d0, d0, d0
uint32x2_t vclt_f32(float32x2_t a, float32x2_t b);  // VCGT.F32 d0, d0, d0
uint8x8_t  vclt_u8(uint8x8_t a, uint8x8_t b);       // VCGT.U8 d0, d0, d0 
uint16x4_t vclt_u16(uint16x4_t a, uint16x4_t b);    // VCGT.U16 d0, d0, d0
uint32x2_t vclt_u32(uint32x2_t a, uint32x2_t b);    // VCGT.U32 d0, d0, d0
uint8x16_t vcltq_s8(int8x16_t a, int8x16_t b);      // VCGT.S8 q0, q0, q0 
uint16x8_t vcltq_s16(int16x8_t a, int16x8_t b);     // VCGT.S16 q0, q0, q0
uint32x4_t vcltq_s32(int32x4_t a, int32x4_t b);     // VCGT.S32 q0, q0, q0
uint32x4_t vcltq_f32(float32x4_t a, float32x4_t b); // VCGT.F32 q0, q0, q0
uint8x16_t vcltq_u8(uint8x16_t a, uint8x16_t b);    // VCGT.U8 q0, q0, q0 
uint16x8_t vcltq_u16(uint16x8_t a, uint16x8_t b);   // VCGT.U16 q0, q0, q0
uint32x4_t vcltq_u32(uint32x4_t a, uint32x4_t b);   // VCGT.U32 q0, q0, q0

Vector compare absolute greater-than or equal

uint32x2_t vcage_f32(float32x2_t a, float32x2_t b);  // VACGE.F32 d0, d0, d0
uint32x4_t vcageq_f32(float32x4_t a, float32x4_t b); // VACGE.F32 q0, q0, q0

Vector compare absolute less-than or equal

uint32x2_t vcale_f32(float32x2_t a, float32x2_t b);  // VACGE.F32 d0, d0, d0
uint32x4_t vcaleq_f32(float32x4_t a, float32x4_t b); // VACGE.F32 q0, q0, q0

Vector compare absolute greater-than

uint32x2_t vcagt_f32(float32x2_t a, float32x2_t b);  // VACGT.F32 d0, d0, d0
uint32x4_t vcagtq_f32(float32x4_t a, float32x4_t b); // VACGT.F32 q0, q0, q0

Vector compare absolute less-than

uint32x2_t vcalt_f32(float32x2_t a, float32x2_t b);  // VACGT.F32 d0, d0, d0
uint32x4_t vcaltq_f32(float32x4_t a, float32x4_t b); // VACGT.F32 q0, q0, q0

Vector test bits

uint8x8_t  vtst_s8(int8x8_t a, int8x8_t b);       // VTST.8 d0, d0, d0 
uint16x4_t vtst_s16(int16x4_t a, int16x4_t b);    // VTST.16 d0, d0, d0
uint32x2_t vtst_s32(int32x2_t a, int32x2_t b);    // VTST.32 d0, d0, d0
uint8x8_t  vtst_u8(uint8x8_t a, uint8x8_t b);     // VTST.8 d0, d0, d0 
uint16x4_t vtst_u16(uint16x4_t a, uint16x4_t b);  // VTST.16 d0, d0, d0
uint32x2_t vtst_u32(uint32x2_t a, uint32x2_t b);  // VTST.32 d0, d0, d0
uint8x8_t  vtst_p8(poly8x8_t a, poly8x8_t b);     // VTST.8 d0, d0, d0 
uint8x16_t vtstq_s8(int8x16_t a, int8x16_t b);    // VTST.8 q0, q0, q0 
uint16x8_t vtstq_s16(int16x8_t a, int16x8_t b);   // VTST.16 q0, q0, q0
uint32x4_t vtstq_s32(int32x4_t a, int32x4_t b);   // VTST.32 q0, q0, q0
uint8x16_t vtstq_u8(uint8x16_t a, uint8x16_t b);  // VTST.8 q0, q0, q0 
uint16x8_t vtstq_u16(uint16x8_t a, uint16x8_t b); // VTST.16 q0, q0, q0
uint32x4_t vtstq_u32(uint32x4_t a, uint32x4_t b); // VTST.32 q0, q0, q0
uint8x16_t vtstq_p8(poly8x16_t a, poly8x16_t b);  // VTST.8 q0, q0, q0 
Non-ConfidentialPDF file icon PDF versionARM DUI0472M
Copyright © 2010-2016 ARM Limited or its affiliates. All rights reserved.