ggml : fixes after sync (ggml/983)

ggml : remove test-backend-buffer

ggml : fix CUDA build warnings

diff --git a/ggml/src/ggml-cuda.cu b/ggml/src/ggml-cuda.cu
index bcb39766be8413dad87af1d586038e50b6c5805a..5b6f605b0089bed5df81919187aee20fcf6180b1 100644 (file)
--- a/ggml/src/ggml-cuda.cu
+++ b/ggml/src/ggml-cuda.cu
@@ -2448,6 +2448,7 @@ static void ggml_backend_cuda_synchronize(ggml_backend_t backend) {
     GGML_UNUSED(backend);
 }
 
+#ifdef USE_CUDA_GRAPH
 static void set_ggml_graph_node_properties(ggml_tensor * node, ggml_graph_node_properties * graph_node_properties) {
     graph_node_properties->node_address = node->data;
     graph_node_properties->node_op = node->op;
@@ -2498,6 +2499,7 @@ static bool ggml_graph_node_has_matching_properties(ggml_tensor * node, ggml_gra
 
     return true;
 }
+#endif
 
 static enum ggml_status ggml_backend_cuda_graph_compute(ggml_backend_t backend, ggml_cgraph * cgraph) {
     ggml_backend_cuda_context * cuda_ctx = (ggml_backend_cuda_context *)backend->context;