__lsx_vssrarni_b_h
LSX intrinsic · group: Shift
Synopsis
__m128i __lsx_vssrarni_b_h (__m128i a, __m128i b, imm0_15 imm)
#include <lsxintrin.h>
Instruction: vssrarni.b.h vr, vr, imm
CPU Flags: LSX
Description
Arithmetic right shift with rounding the signed 16-bit elements in a and b by imm, saturate the result to signed 8-bit, write the 8-bit result into the lower half of each 128-bit lane of dst, and zero the upper half.
Operation
for (int i = 0; i < 16; i++) {
if (i < 8) {
s16 temp;
if (imm == 0) {
temp = (s16)b.half[i];
} else {
temp = ((s16)b.half[i] >> imm) + (((s16)b.half[i] >> (imm - 1)) & 1);
}
dst.byte[i] = clamp<s16>(temp, -128, 127);
} else {
s16 temp;
if (imm == 0) {
temp = (s16)a.half[i - 8];
} else {
temp =
((s16)a.half[i - 8] >> imm) + (((s16)a.half[i - 8] >> (imm - 1)) & 1);
}
dst.byte[i] = clamp<s16>(temp, -128, 127);
}
}
Tested on real machine.
Latency and Throughput
| CPU | µarch | Latency | Throughput (IPC) |
|---|---|---|---|
| 3C5000 | LA464 | 4 | 1 |
| 3A6000 | LA664 | 4 | 2 |
| 3C6000 | LA664 | 4 | 2 |
| 2K1000LA | LA264 | 3 | 0.5(1/2) |
| 2K3000 | LA364E | 3 | 0.5(1/2) |