__lsx_vfrsqrt_s
LSX intrinsic · group: Floating Point Computation
Synopsis
__m128 __lsx_vfrsqrt_s (__m128 a)
#include <lsxintrin.h>
Instruction: vfrsqrt.s vr, vr
CPU Flags: LSX
Description
Compute reciprocal of square root of single precision floating point elements in a.
Operation
for (int i = 0; i < 4; i++) {
dst.fp32[i] = 1.0 / sqrt(a.fp32[i]);
}
Latency and Throughput
| CPU | µarch | Latency | Throughput (IPC) |
|---|---|---|---|
| 3C5000 | LA464 | 21 | 0.11(1/9) |
| 3A6000 | LA664 | 17 | 0.05(1/19) |
| 3C6000 | LA664 | 21 | 0.11(1/9.5) |
| 2K1000LA | LA264 | 26 | 0(1/62) |
| 2K3000 | LA364E | 26 | 0(1/62) |