wehub-resource-sync
53 lines
2.0 KiB
C++
53 lines
2.0 KiB
C++
#include <riscv_vector.h>
|
|
|
|
void MNNAxByClampBroadcastUnit(float* C, const float* A, const float* B, size_t width, size_t cStride, size_t aStride, size_t height, const float* parameters) {
|
|
float beta = parameters[1];
|
|
float minF = parameters[2];
|
|
float maxF = parameters[3];
|
|
const ptrdiff_t stride = 4 * sizeof(float);
|
|
|
|
for (int y = 0; y < height; ++y) {
|
|
auto a = A + aStride * y;
|
|
auto b = B + 4 * y;
|
|
auto c = C + cStride * y;
|
|
float b0Beta = b[0] * beta;
|
|
float b1Beta = b[1] * beta;
|
|
float b2Beta = b[2] * beta;
|
|
float b3Beta = b[3] * beta;
|
|
size_t w = width;
|
|
|
|
while (w > 0) {
|
|
size_t vl = __riscv_vsetvl_e32m8(w);
|
|
|
|
vfloat32m8_t data = __riscv_vlse32_v_f32m8(a + 0, stride, vl);
|
|
data = __riscv_vfadd_vf_f32m8(data, b0Beta, vl);
|
|
data = __riscv_vfmax_vf_f32m8(data, minF, vl);
|
|
data = __riscv_vfmin_vf_f32m8(data, maxF, vl);
|
|
__riscv_vsse32_v_f32m8(c + 0, stride, data, vl);
|
|
|
|
data = __riscv_vlse32_v_f32m8(a + 1, stride, vl);
|
|
data = __riscv_vfadd_vf_f32m8(data, b1Beta, vl);
|
|
data = __riscv_vfmax_vf_f32m8(data, minF, vl);
|
|
data = __riscv_vfmin_vf_f32m8(data, maxF, vl);
|
|
__riscv_vsse32_v_f32m8(c + 1, stride, data, vl);
|
|
|
|
data = __riscv_vlse32_v_f32m8(a + 2, stride, vl);
|
|
data = __riscv_vfadd_vf_f32m8(data, b2Beta, vl);
|
|
data = __riscv_vfmax_vf_f32m8(data, minF, vl);
|
|
data = __riscv_vfmin_vf_f32m8(data, maxF, vl);
|
|
__riscv_vsse32_v_f32m8(c + 2, stride, data, vl);
|
|
|
|
data = __riscv_vlse32_v_f32m8(a + 3, stride, vl);
|
|
data = __riscv_vfadd_vf_f32m8(data, b3Beta, vl);
|
|
data = __riscv_vfmax_vf_f32m8(data, minF, vl);
|
|
data = __riscv_vfmin_vf_f32m8(data, maxF, vl);
|
|
__riscv_vsse32_v_f32m8(c + 3, stride, data, vl);
|
|
|
|
a += 4 * vl;
|
|
c += 4 * vl;
|
|
w -= vl;
|
|
}
|
|
}
|
|
}
|
|
|