__lasx_xvsrar_w
LASX intrinsic · group: Shift
Synopsis
__m256i __lasx_xvsrar_w (__m256i a, __m256i b)
#include <lasxintrin.h>
Instruction: xvsrar.w xr, xr, xr
CPU Flags: LASX
Description
Arithmetic right shift with rounding each signed 32-bit element in a by the corresponding element of b modulo 32, store the result in dst.
Operation
for (int i = 0; i < 8; i++) {
if ((b.word[i] & 0x1f) == 0) {
dst.word[i] = a.word[i];
} else {
dst.word[i] = ((s32)a.word[i] >> (b.word[i] & 0x1f)) +
(((s32)a.word[i] >> ((b.word[i] & 0x1f) - 1)) & 0x1);
}
}
Tested on real machine.
Latency and Throughput
| CPU | µarch | Latency | Throughput (IPC) |
|---|---|---|---|
| 3C5000 | LA464 | 3 | 2 |
| 3A6000 | LA664 | 3 | 2 |
| 3C6000 | LA664 | 3 | 2 |