__lsx_vhsubw_wu_hu

LSX intrinsic · group: Integer Computation

Synopsis

__m128i __lsx_vhsubw_wu_hu (__m128i a, __m128i b)
#include <lsxintrin.h>
Instruction: vhsubw.wu.hu vr, vr, vr
CPU Flags: LSX

Description

Subtract even-positioned unsigned 16-bit elements in b from odd-positioned unsigned 16-bit elements in a, zero-extending both to 32-bit, to get the 32-bit result.

Operation

for (int i = 0; i < 4; i++) {
  dst.word[i] = (u32)(u16)a.half[2 * i + 1] - (u32)(u16)b.half[2 * i];
}

// Expands to:

if (0) {
  dst.word[0] = ((u32)((u16)a.half[1])) - ((u32)((u16)b.half[0]));
  dst.word[1] = ((u32)((u16)a.half[3])) - ((u32)((u16)b.half[2]));
  dst.word[2] = ((u32)((u16)a.half[5])) - ((u32)((u16)b.half[4]));
  dst.word[3] = ((u32)((u16)a.half[7])) - ((u32)((u16)b.half[6]));
}

Tested on real machine.

Latency and Throughput

CPU	µarch	Latency	Throughput (IPC)
3C5000	LA464	2	2
3A6000	LA664	2	2
3C6000	LA664	2	2
2K1000LA	LA264	2	1
2K3000	LA364E	2	1

---

All intrinsics in alphabetical order