diff --git a/paddle/fluid/framework/details/build_strategy.cc b/paddle/fluid/framework/details/build_strategy.cc
index 7d2a081e3b1d0ae3501730bef052d483b23d5a62..45c2c73415264f2df9145262a0bcd3ab26a78506 100644
--- a/paddle/fluid/framework/details/build_strategy.cc
+++ b/paddle/fluid/framework/details/build_strategy.cc
@@ -34,6 +34,8 @@ namespace details {
 static inline bool SeqOnlyAllReduceOps(const BuildStrategy &strategy) {
   // Should fix the allreduce op order if scheduling
   // them in multiple threads or processes to avoid hang.
+  // NOTE: ParallelExecutor would execute this pass on each graph, so
+  // don't need to append it here.
   return (!strategy.enable_sequential_execution_ &&
           strategy.num_trainers_ > 1) &&
          !strategy.enable_parallel_graph_;
@@ -118,7 +120,7 @@ class ParallelExecutorPassBuilder : public ir::PassBuilder {
     }
 
     // Verify that the graph is correct for multi-device executor.
-    auto multi_devices_pass = AppendPass("multi_devices_check_pass");
+    AppendPass("multi_devices_check_pass");
 
     if (SeqOnlyAllReduceOps(strategy)) {
       AppendPass("all_reduce_deps_pass");