diff --git a/onnxruntime/core/common/threadpool.cc b/onnxruntime/core/common/threadpool.cc
index 9292b4a8c7..6cdcb3add7 100644
--- a/onnxruntime/core/common/threadpool.cc
+++ b/onnxruntime/core/common/threadpool.cc
@@ -39,23 +39,9 @@ void ThreadPool::ParallelFor(int32_t total, std::function<void(int32_t)> fn) {
     fn(0);
     return;
   }
+
   // TODO: Eigen supports a more efficient ThreadPoolDevice mechanism
   // We will simply rely on the work queue and stealing in the short term.
-  if (total > NumThreads()) {
-    //The dispatcher thread will be idle at here
-    Barrier barrier(static_cast<unsigned int>(total));
-    std::function<void(int32_t)> handle_iteration = [&barrier, &fn](int iteration) {
-      fn(iteration);
-      barrier.Notify();
-    };
-
-    for (int32_t id = 0; id < total; ++id) {
-      Schedule([=, &handle_iteration]() { handle_iteration(id); });
-    }
-
-    barrier.Wait();
-    return;
-  }
   Barrier barrier(static_cast<unsigned int>(total - 1));
   std::function<void(int32_t)> handle_iteration = [&barrier, &fn](int iteration) {
     fn(iteration);
@@ -65,7 +51,7 @@ void ThreadPool::ParallelFor(int32_t total, std::function<void(int32_t)> fn) {
   for (int32_t id = 1; id < total; ++id) {
     Schedule([=, &handle_iteration]() { handle_iteration(id); });
   }
-  //reuse the current thread for one task
+
   fn(0);
   barrier.Wait();
 }
diff --git a/onnxruntime/core/mlas/lib/mlasi.h b/onnxruntime/core/mlas/lib/mlasi.h
index c19ac16694..6ba13c728d 100644
--- a/onnxruntime/core/mlas/lib/mlasi.h
+++ b/onnxruntime/core/mlas/lib/mlasi.h
@@ -545,7 +545,7 @@ MlasGetMaximumThreadCount(
     MLAS_UNREFERENCED_PARAMETER(ThreadPool);
 #else
     if (ThreadPool != nullptr) {
-        return ThreadPool->NumThreads();
+        return ThreadPool->NumThreads() + 1;
     }
 #endif