|
|
@ -6779,15 +6779,20 @@ static void ggml_compute_forward_sum_f32(
|
|
|
|
const size_t nb02 = src0->nb[2];
|
|
|
|
const size_t nb02 = src0->nb[2];
|
|
|
|
const size_t nb03 = src0->nb[3];
|
|
|
|
const size_t nb03 = src0->nb[3];
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
ggml_float sum = 0;
|
|
|
|
|
|
|
|
float row_sum = 0;
|
|
|
|
|
|
|
|
|
|
|
|
for (int64_t i03 = 0; i03 < ne03; i03++) {
|
|
|
|
for (int64_t i03 = 0; i03 < ne03; i03++) {
|
|
|
|
for (int64_t i02 = 0; i02 < ne02; i02++) {
|
|
|
|
for (int64_t i02 = 0; i02 < ne02; i02++) {
|
|
|
|
for (int64_t i01 = 0; i01 < ne01; i01++) {
|
|
|
|
for (int64_t i01 = 0; i01 < ne01; i01++) {
|
|
|
|
ggml_vec_sum_f32(ne00,
|
|
|
|
ggml_vec_sum_f32(ne00,
|
|
|
|
(float *) (dst->data),
|
|
|
|
&row_sum,
|
|
|
|
(float *) ((char *) src0->data + i01*nb01 + i02*nb02 + i03*nb03));
|
|
|
|
(float *) ((char *) src0->data + i01*nb01 + i02*nb02 + i03*nb03));
|
|
|
|
|
|
|
|
sum += row_sum;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
((float *) dst->data)[0] = sum;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static void ggml_compute_forward_sum(
|
|
|
|
static void ggml_compute_forward_sum(
|
|
|
|