diff --git a/paddle/fluid/framework/parallel_executor.cc b/paddle/fluid/framework/parallel_executor.cc
index b53a6f43fbd1f23e69d23ad0fcc54d5c25d352a3..751b10eeeed10828c08ada4173300c07f81c093e 100644
--- a/paddle/fluid/framework/parallel_executor.cc
+++ b/paddle/fluid/framework/parallel_executor.cc
@@ -253,6 +253,9 @@ void ParallelExecutor::FeedAndSplitTensorIntoLocalScopes(
       t->set_lod(lod_tensors[j].lod());
     }
   }
+  for (auto &p : member_->places_) {
+    platform::DeviceContextPool::Instance().Get(p)->Wait();
+  }
 }
 
 ParallelExecutor::~ParallelExecutor() {