__lasx_xvsubwev_h_b

LASX intrinsic · group: Integer Computation

Synopsis

__m256i __lasx_xvsubwev_h_b (__m256i a, __m256i b)
#include <lasxintrin.h>
Instruction: xvsubwev.h.b xr, xr, xr
CPU Flags: LASX

Description

Subtract the even-positioned signed 8-bit elements in b from the corresponding elements in a, extending both to 16-bit, and save the 16-bit result in dst.

Operation

for (int i = 0; i < 16; i++) {
  dst.half[i] = (s16)(s8)a.byte[2 * i] - (s16)(s8)b.byte[2 * i];
}

// Expands to:

if (0) {
  dst.half[0] = ((s16)((s8)a.byte[0])) - ((s16)((s8)b.byte[0]));
  dst.half[1] = ((s16)((s8)a.byte[2])) - ((s16)((s8)b.byte[2]));
  dst.half[2] = ((s16)((s8)a.byte[4])) - ((s16)((s8)b.byte[4]));
  dst.half[3] = ((s16)((s8)a.byte[6])) - ((s16)((s8)b.byte[6]));
  dst.half[4] = ((s16)((s8)a.byte[8])) - ((s16)((s8)b.byte[8]));
  dst.half[5] = ((s16)((s8)a.byte[10])) - ((s16)((s8)b.byte[10]));
  dst.half[6] = ((s16)((s8)a.byte[12])) - ((s16)((s8)b.byte[12]));
  dst.half[7] = ((s16)((s8)a.byte[14])) - ((s16)((s8)b.byte[14]));
  dst.half[8] = ((s16)((s8)a.byte[16])) - ((s16)((s8)b.byte[16]));
  dst.half[9] = ((s16)((s8)a.byte[18])) - ((s16)((s8)b.byte[18]));
  dst.half[10] = ((s16)((s8)a.byte[20])) - ((s16)((s8)b.byte[20]));
  dst.half[11] = ((s16)((s8)a.byte[22])) - ((s16)((s8)b.byte[22]));
  dst.half[12] = ((s16)((s8)a.byte[24])) - ((s16)((s8)b.byte[24]));
  dst.half[13] = ((s16)((s8)a.byte[26])) - ((s16)((s8)b.byte[26]));
  dst.half[14] = ((s16)((s8)a.byte[28])) - ((s16)((s8)b.byte[28]));
  dst.half[15] = ((s16)((s8)a.byte[30])) - ((s16)((s8)b.byte[30]));
}

Tested on real machine.

Latency and Throughput

CPU	µarch	Latency	Throughput (IPC)
3C5000	LA464	2	2
3A6000	LA664	2	2
3C6000	LA664	2	2

---

All intrinsics in alphabetical order