Avoid using saved CUDA graph if scale changes and reset nodes/params on update

Fixes #9451

Author: agray3

ggml/src/ggml-cuda.cu

@@ -2451,6 +2451,7 @@ static void set_ggml_graph_node_properties(ggml_tensor * node, ggml_graph_node_p
     for (int i = 0; i < GGML_MAX_SRC; i++) {
         graph_node_properties->src_address[i] = node->src[i] ? node->src[i]->data : nullptr;
     }
+    memcpy(graph_node_properties->op_params, node->op_params, GGML_MAX_OP_PARAMS);
 }
 
 static bool ggml_graph_node_has_matching_properties(ggml_tensor * node, ggml_graph_node_properties * graph_node_properties) {
@@ -2482,6 +2483,16 @@ static bool ggml_graph_node_has_matching_properties(ggml_tensor * node, ggml_gra
             return false;
         }
     }
+
+    if (node->op == GGML_OP_SCALE) {
+        for (size_t i = 0; i < GGML_MAX_OP_PARAMS / sizeof(int32_t); i++) {
+            int32_t op_param = ((const int32_t *)(node->op_params))[i];
+            int32_t previous_op_param = ((const int32_t *)(graph_node_properties->op_params))[i];
+            if (op_param != previous_op_param) {
+                return false;
+            }
+        }
+    }
     return true;
 }
 
@@ -2694,7 +2705,9 @@ GGML_CALL static enum ggml_status ggml_backend_cuda_graph_compute(ggml_backend_t
             CUDA_CHECK(cudaGraphGetNodes(cuda_ctx->cuda_graph->graph, nullptr, &cuda_ctx->cuda_graph->num_nodes));
             // Subsequent call with non-null argument gets nodes
             cuda_ctx->cuda_graph->nodes.resize(cuda_ctx->cuda_graph->num_nodes);
+            cuda_ctx->cuda_graph->nodes.clear();
             cuda_ctx->cuda_graph->params.resize(cuda_ctx->cuda_graph->num_nodes);
+            cuda_ctx->cuda_graph->params.clear();
             if (cuda_ctx->cuda_graph->num_nodes > 0) {
                 CUDA_CHECK(cudaGraphGetNodes(cuda_ctx->cuda_graph->graph, cuda_ctx->cuda_graph->nodes.data(), &cuda_ctx->cuda_graph->num_nodes));
 

ggml/src/ggml-cuda/common.cuh

@@ -569,6 +569,7 @@ struct ggml_graph_node_properties {
     int64_t ne[GGML_MAX_DIMS];
     size_t nb[GGML_MAX_DIMS];
     void * src_address[GGML_MAX_SRC];
+    int32_t op_params[GGML_MAX_OP_PARAMS / sizeof(int32_t)];
 };
 
 struct ggml_cuda_graph {