__lsx_vmaddwev_d_wu_w
LSX intrinsic · group: Integer Computation
Synopsis
__m128i __lsx_vmaddwev_d_wu_w (__m128i a, __m128i b, __m128i c)
#include <lsxintrin.h>
Instruction: vmaddwev.d.wu.w vr, vr, vr
CPU Flags: LSX
Description
Multiply the even-positioned unsigned 32-bit elements in b and the corresponding signed 32-bit elements in c, extending both to 64-bit, and add the product to the 64-bit elements in a.
Operation
for (int i = 0; i < 2; i++) {
dst.dword[i] =
(u64)(u32)b.word[2 * i] * (s64)(s32)c.word[2 * i] + (s64)a.dword[i];
}
// Expands to:
if (0) {
dst.dword[0] =
(((u64)((u32)b.word[0])) * ((s64)((s32)c.word[0]))) + ((s64)a.dword[0]);
dst.dword[1] =
(((u64)((u32)b.word[2])) * ((s64)((s32)c.word[2]))) + ((s64)a.dword[1]);
}
Tested on real machine.
Latency and Throughput
| CPU | µarch | Latency | Throughput (IPC) |
|---|---|---|---|
| 3C5000 | LA464 | 4 | 2 |
| 3A6000 | LA664 | 4 | 2 |
| 3C6000 | LA664 | 4 | 2 |
| 2K1000LA | LA264 | 4 | 1 |
| 2K3000 | LA364E | 4 | 1 |