We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
1 parent 33e171d commit adf3de4Copy full SHA for adf3de4
ggml.c
@@ -15629,7 +15629,6 @@ static int ggml_get_n_tasks(struct ggml_tensor * node, int n_threads) {
15629
} break;
15630
case GGML_OP_DIAG_MASK_ZERO:
15631
case GGML_OP_DIAG_MASK_INF:
15632
- case GGML_OP_SOFT_MAX:
15633
case GGML_OP_SOFT_MAX_BACK:
15634
case GGML_OP_ROPE:
15635
case GGML_OP_ROPE_BACK:
@@ -15645,6 +15644,10 @@ static int ggml_get_n_tasks(struct ggml_tensor * node, int n_threads) {
15645
15644
{
15646
n_tasks = 1; //TODO
15647
+ case GGML_OP_SOFT_MAX:
15648
+ {
15649
+ n_tasks = MIN(MIN(4, n_threads), ggml_nrows(node->src[0]));
15650
+ } break;
15651
case GGML_OP_CONV_TRANSPOSE_1D:
15652
15653
n_tasks = n_threads;
0 commit comments