llama : fix order of parameters (#8706)

usage of `aclrtGetMemInfo` is correct:

https://www.hiascend.com/doc_center/source/zh/canncommercial/63RC2/inferapplicationdev/aclcppdevg/aclcppdevg_03_0103.html

Co-authored-by: Judd <redacted>

diff --git a/src/llama.cpp b/src/llama.cpp
index bc0183741db4e44e280f0887fe0d548dc41d7edc..77f7d32f84c0b0ee557832a48ef676280d2fbfdf 100644 (file)
--- a/src/llama.cpp
+++ b/src/llama.cpp
@@ -2905,7 +2905,7 @@ static size_t llama_get_device_memory(const llama_model & model, int device) {
 #elif defined(GGML_USE_CANN)
     size_t total;
     size_t free;
-    ggml_backend_cann_get_device_memory(device, &total, &free);
+    ggml_backend_cann_get_device_memory(device, &free, &total);
     return free;
 #else
     return 1;