metal : cap threadgroups size of set_rows

2026-03-05 14:33:24 +02:00 · 2025-11-10 16:17:09 +02:00
1 changed files with 5 additions and 0 deletions
--- a/ggml/src/ggml-metal/ggml-metal-ops.cpp
+++ b/ggml/src/ggml-metal/ggml-metal-ops.cpp
@@ -1036,6 +1036,11 @@ int ggml_metal_op_set_rows(ggml_metal_op_t ctx, int idx) {

    nth = std::min(nth, nk0);

+    if (nth*nrptg > ggml_metal_pipeline_max_theads_per_threadgroup(pipeline)) {
+        nth = ggml_metal_pipeline_max_theads_per_threadgroup(pipeline);
+        nrptg = 1;
+    }
+
    ggml_metal_kargs_set_rows args = {
        /*.nk0  =*/ nk0,
        /*.ne01 =*/ ne01,