__lsx_vextrins_h
LSX intrinsic · group: Misc
Synopsis
__m128i __lsx_vextrins_h (__m128i a, __m128i b, imm0_255 imm)
#include <lsxintrin.h>
Instruction: vextrins.h vr, vr, imm
CPU Flags: LSX
Description
Copy the 16-bit element of b selected by imm[2:0] into the element of a selected by imm[6:4]; other elements remain unchanged.
Operation
for (int i = 0; i < 8; i++) {
dst.half[i] = (i == ((imm >> 4) & 7)) ? b.half[imm & 7] : a.half[i];
}
// Expands to:
if (0) {
dst.half[0] = (0 == ((imm >> 4) & 7)) ? (b.half[imm & 7]) : (a.half[0]);
dst.half[1] = (1 == ((imm >> 4) & 7)) ? (b.half[imm & 7]) : (a.half[1]);
dst.half[2] = (2 == ((imm >> 4) & 7)) ? (b.half[imm & 7]) : (a.half[2]);
dst.half[3] = (3 == ((imm >> 4) & 7)) ? (b.half[imm & 7]) : (a.half[3]);
dst.half[4] = (4 == ((imm >> 4) & 7)) ? (b.half[imm & 7]) : (a.half[4]);
dst.half[5] = (5 == ((imm >> 4) & 7)) ? (b.half[imm & 7]) : (a.half[5]);
dst.half[6] = (6 == ((imm >> 4) & 7)) ? (b.half[imm & 7]) : (a.half[6]);
dst.half[7] = (7 == ((imm >> 4) & 7)) ? (b.half[imm & 7]) : (a.half[7]);
}
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 |