__lsx_vmaddwev_q_du_d
LSX intrinsic · group: Integer Computation
Synopsis
__m128i __lsx_vmaddwev_q_du_d (__m128i a, __m128i b, __m128i c)
#include <lsxintrin.h>
Instruction: vmaddwev.q.du.d vr, vr, vr
CPU Flags: LSX
Description
Multiply the even-positioned unsigned 64-bit elements in b and the corresponding signed 64-bit elements in c, extending both to 128-bit, and add the product to the 128-bit elements in a.
Operation
for (int i = 0; i < 1; i++) {
dst.qword[i] =
(u128)(u64)b.dword[2 * i] * (s128)(s64)c.dword[2 * i] + (s128)a.qword[i];
}
// Expands to:
if (0) {
dst.qword[0] = (((u128)((u64)b.dword[0])) * ((s128)((s64)c.dword[0]))) +
((s128)a.qword[0]);
}
Tested on real machine.
Latency and Throughput
| CPU | µarch | Latency | Throughput (IPC) |
|---|---|---|---|
| 3C5000 | LA464 | 7 | 1.14 |
| 3A6000 | LA664 | 7 | 1.14 |
| 3C6000 | LA664 | 7 | 1.14 |
| 2K1000LA | LA264 | 6 | 1 |
| 2K3000 | LA364E | 6 | 1 |