__lasx_xvfrecip_d
LASX intrinsic · group: Floating Point Computation
Synopsis
__m256d __lasx_xvfrecip_d (__m256d a)
#include <lasxintrin.h>
Instruction: xvfrecip.d xr, xr
CPU Flags: LASX
Description
Compute reciprocal of double precision floating point elements in a.
Operation
for (int i = 0; i < 4; i++) {
dst.fp64[i] = 1 / a.fp64[i];
}
Latency and Throughput
| CPU | µarch | Latency | Throughput (IPC) |
|---|---|---|---|
| 3C5000 | LA464 | 23 | 0.08(1/12) |
| 3A6000 | LA664 | 23 | 0.25(1/4) |
| 3C6000 | LA664 | 23 | 0.1(1/10.5) |