diff --git a/paddle/fluid/inference/api/analysis_predictor.cc b/paddle/fluid/inference/api/analysis_predictor.cc
index 5d9d5a3178aaa39f4b80197fb5ac7cd46504bf4f..8b775b0faa6b2ba58ed2a9f330d9619535a49a3d 100644
--- a/paddle/fluid/inference/api/analysis_predictor.cc
+++ b/paddle/fluid/inference/api/analysis_predictor.cc
@@ -185,14 +185,6 @@ bool AnalysisPredictor::PrepareExecutor() {
   return true;
 }
 
-void AnalysisPredictor::SetMkldnnThreadID(int tid) {
-#ifdef PADDLE_WITH_MKLDNN
-  platform::set_cur_thread_id(tid);
-#else
-  LOG(ERROR) << "Please compile with MKLDNN first to use MKLDNN";
-#endif
-}
-
 bool AnalysisPredictor::Run(const std::vector<PaddleTensor> &inputs,
                             std::vector<PaddleTensor> *output_data,
                             int batch_size) {
diff --git a/paddle/fluid/inference/api/analysis_predictor.h b/paddle/fluid/inference/api/analysis_predictor.h
index b5e134ced70f8bf9ef0267bee08ec9836aeb5338..551ca5ba0cd6f555c5b01a212a941908d8ce898a 100644
--- a/paddle/fluid/inference/api/analysis_predictor.h
+++ b/paddle/fluid/inference/api/analysis_predictor.h
@@ -80,8 +80,6 @@ class AnalysisPredictor : public PaddlePredictor {
   framework::Scope *scope() { return scope_.get(); }
   framework::ProgramDesc &program() { return *inference_program_; }
 
-  void SetMkldnnThreadID(int tid);
-
   std::string GetSerializedProgram() const override;
 
   bool MkldnnQuantize();
diff --git a/paddle/fluid/inference/tests/api/analyzer_int8_object_detection_tester.cc b/paddle/fluid/inference/tests/api/analyzer_int8_object_detection_tester.cc
index 19c38270b70dff92085f2f19b98f71d5cb804d4a..ccb50d404379d0b3aea64d04371ceae0fcf6daef 100644
--- a/paddle/fluid/inference/tests/api/analyzer_int8_object_detection_tester.cc
+++ b/paddle/fluid/inference/tests/api/analyzer_int8_object_detection_tester.cc
@@ -144,7 +144,7 @@ std::shared_ptr<std::vector<PaddleTensor>> GetWarmupData(
   int test_data_batch_size = test_data[0][0].shape[0];
   auto iterations = test_data.size();
   PADDLE_ENFORCE(
-      static_cast<int32_t>(num_images) <= iterations * test_data_batch_size,
+      static_cast<size_t>(num_images) <= iterations * test_data_batch_size,
       "The requested quantization warmup data size " +
           std::to_string(num_images) + " is bigger than all test data size.");
 
diff --git a/paddle/fluid/inference/tests/api/tester_helper.h b/paddle/fluid/inference/tests/api/tester_helper.h
index 05936458cefbadecb0ab15afdc3b48d0fd4d64ce..61cf10c31788be87d14c93a344168088390e9275 100644
--- a/paddle/fluid/inference/tests/api/tester_helper.h
+++ b/paddle/fluid/inference/tests/api/tester_helper.h
@@ -418,12 +418,6 @@ void TestMultiThreadPrediction(
       // The inputs of each thread are all the same.
       std::vector<std::vector<PaddleTensor>> outputs_tid;
       auto &predictor = predictors[tid];
-#ifdef PADDLE_WITH_MKLDNN
-      if (use_analysis) {
-        static_cast<AnalysisPredictor *>(predictor.get())
-            ->SetMkldnnThreadID(static_cast<int>(tid) + 1);
-      }
-#endif
       if (FLAGS_warmup) {
         PredictionWarmUp(predictor.get(), inputs, &outputs_tid, num_threads,
                          tid);