__lsx_vshuf4i_b

LSX intrinsic · group: Shuffling

Synopsis

__m128i __lsx_vshuf4i_b (__m128i a, imm0_255 imm)
#include <lsxintrin.h>
Instruction: vshuf4i.b vr, vr, imm
CPU Flags: LSX

Description

For each group of four consecutive 8-bit elements in a, permute the group according to the four 2-bit fields of imm, save the result to dst.

Examples

__m128i __lsx_vshuf4i_b(__m128i{0xabcdef1314156678, 0x1234123443214321}, 0x12)
= 0x13ef13cd78667815 0x3412343421432121

Operation

for (int i = 0; i < 16; i++) {
  dst.byte[i] = a.byte[(i & ~0x3) + ((imm >> (2 * (i & 0x3))) & 0x3)];
}

// Expands to:

if (0) {
  dst.byte[0] = a.byte[0 + ((imm >> 0) & 0x3)];
  dst.byte[1] = a.byte[0 + ((imm >> 2) & 0x3)];
  dst.byte[2] = a.byte[0 + ((imm >> 4) & 0x3)];
  dst.byte[3] = a.byte[0 + ((imm >> 6) & 0x3)];
  dst.byte[4] = a.byte[4 + ((imm >> 0) & 0x3)];
  dst.byte[5] = a.byte[4 + ((imm >> 2) & 0x3)];
  dst.byte[6] = a.byte[4 + ((imm >> 4) & 0x3)];
  dst.byte[7] = a.byte[4 + ((imm >> 6) & 0x3)];
  dst.byte[8] = a.byte[8 + ((imm >> 0) & 0x3)];
  dst.byte[9] = a.byte[8 + ((imm >> 2) & 0x3)];
  dst.byte[10] = a.byte[8 + ((imm >> 4) & 0x3)];
  dst.byte[11] = a.byte[8 + ((imm >> 6) & 0x3)];
  dst.byte[12] = a.byte[12 + ((imm >> 0) & 0x3)];
  dst.byte[13] = a.byte[12 + ((imm >> 2) & 0x3)];
  dst.byte[14] = a.byte[12 + ((imm >> 4) & 0x3)];
  dst.byte[15] = a.byte[12 + ((imm >> 6) & 0x3)];
}

Tested on real machine.

Latency and Throughput

CPU	µarch	Latency	Throughput (IPC)
3C5000	LA464	1	2
3A6000	LA664	1	4
3C6000	LA664	1	4
2K1000LA	LA264	1	1
2K3000	LA364E	1	1

---

All intrinsics in alphabetical order