__lsx_vfmadd_s
LSX intrinsic · group: Fused Multiply-Add
Synopsis
__m128 __lsx_vfmadd_s (__m128 a, __m128 b, __m128 c)
#include <lsxintrin.h>
Instruction: vfmadd.s vr, vr, vr, vr
CPU Flags: LSX
Description
Compute packed single precision floating point FMA(Fused Multiply-Add): multiply elements in a and b, accumulate to elements in c and store the result in dst.
Operation
for (int i = 0; i < 4; i++) {
dst.fp32[i] = a.fp32[i] * b.fp32[i] + c.fp32[i];
}
Latency and Throughput
| CPU | µarch | Latency | Throughput (IPC) |
|---|---|---|---|
| 3C5000 | LA464 | 5 | 2 |
| 3A6000 | LA664 | 5 | 2 |
| 3C6000 | LA664 | 5 | 2 |
| 2K1000LA | LA264 | 4 | 1 |
| 2K3000 | LA364E | 4 | 1 |