diff --git a/ggml-qnn.cpp b/ggml-qnn.cpp
index 6cfb996771f97d..3248e244a31c24 100644
--- a/ggml-qnn.cpp
+++ b/ggml-qnn.cpp
@@ -1580,7 +1580,7 @@ int qnn_instance::qnn_init(const QnnSaver_Config_t ** saver_config) {
         size_t candidate_size = 0;
         uint8_t * rpc_buffer = nullptr;
         const int SIZE_IN_MB = (1 << 20);
-        size_t probe_slots[] = { 1024, 1536, 2048 - 48, 2048};
+        size_t probe_slots[] = {1024, 1536, 2048 - 48, 2048};
         size_t probe_counts  = sizeof(probe_slots) / sizeof(size_t);
         for (size_t idx = 0; idx < probe_counts; idx++) {
             rpc_buffer = static_cast<uint8_t *>(alloc_rpcmem(probe_slots[idx] * SIZE_IN_MB, 4));
@@ -1589,10 +1589,8 @@ int qnn_instance::qnn_init(const QnnSaver_Config_t ** saver_config) {
                 break;
             } else {
                 candidate_size = probe_slots[idx];
-                if (nullptr != rpc_buffer) {
-                    free_rpcmem(rpc_buffer);
-                    rpc_buffer = nullptr;
-                }
+                free_rpcmem(rpc_buffer);
+                rpc_buffer = nullptr;
             }
         }
         if (candidate_size > _rpcmem_capacity)