From ec6608eb119bd50f666071aa90eec2a42471b7e2 Mon Sep 17 00:00:00 2001
From: Shangtong Zhang <zhangshangtong.cpp@icloud.com>
Date: Thu, 3 Aug 2017 22:32:02 -0600
Subject: [PATCH] Check HAS_OPENMP

---
 .../async_learning_impl.hpp                   | 45 +++++++++++++------
 .../worker/one_step_q_learning_worker.hpp     | 14 ++++--
 src/mlpack/tests/async_learning_test.cpp      |  4 +-
 3 files changed, 44 insertions(+), 19 deletions(-)

diff --git a/src/mlpack/methods/reinforcement_learning/async_learning_impl.hpp b/src/mlpack/methods/reinforcement_learning/async_learning_impl.hpp
index 3a007fc461..2cbb1c5c40 100644
--- a/src/mlpack/methods/reinforcement_learning/async_learning_impl.hpp
+++ b/src/mlpack/methods/reinforcement_learning/async_learning_impl.hpp
@@ -86,8 +86,10 @@ void AsyncLearning<
     tasks.push(i);
 
   // Synchronization for get/put task.
-  omp_lock_t tasksLock;
-  omp_init_lock(&tasksLock);
+  #ifdef HAS_OPENMP
+    omp_lock_t tasksLock;
+    omp_init_lock(&tasksLock);
+  #endif
 
   /**
    * Compute the number of threads for the for-loop. In general, we should use
@@ -95,15 +97,21 @@ void AsyncLearning<
    * compiler. We can switch to OpenMP task once MSVC supports OpenMP 3.0.
    */
   size_t numThreads = 0;
-  #pragma omp parallel reduction(+:numThreads)
+  #ifdef HAS_OPENMP
+    #pragma omp parallel reduction(+:numThreads)
+  #endif
   numThreads++;
   Log::Debug << numThreads << " threads will be used in total." << std::endl;
 
-  #pragma omp parallel for shared(stop, workers, tasks, learningNetwork, \
-      targetNetwork, totalSteps, policy)
+  #ifdef HAS_OPENMP
+    #pragma omp parallel for shared(stop, workers, tasks, learningNetwork, \
+        targetNetwork, totalSteps, policy)
+  #endif
   for (omp_size_t i = 0; i < numThreads; ++i)
   {
-    #pragma omp critical
+    #ifdef HAS_OPENMP
+      #pragma omp critical
+    #endif
     {
       Log::Debug << "Thread " << omp_get_thread_num() <<
           " started." << std::endl;
@@ -111,14 +119,19 @@ void AsyncLearning<
     while (!stop)
     {
       // Assign task to current thread from queue.
-      omp_set_lock(&tasksLock);
-      if (tasks.empty()) {
-        omp_unset_lock(&tasksLock);
-        continue;
-      }
+      #ifdef HAS_OPENMP
+        omp_set_lock(&tasksLock);
+        if (tasks.empty())
+        {
+          omp_unset_lock(&tasksLock);
+          continue;
+        }
+      #endif
       size_t task = tasks.front();
       tasks.pop();
-      omp_unset_lock(&tasksLock);
+      #ifdef HAS_OPENMP
+        omp_unset_lock(&tasksLock);
+      #endif
 
       // Get corresponding worker.
       WorkerType& worker = workers[task];
@@ -130,9 +143,13 @@ void AsyncLearning<
       }
 
       // Put task back to queue.
-      omp_set_lock(&tasksLock);
+      #ifdef HAS_OPENMP
+        omp_set_lock(&tasksLock);
+      #endif
       tasks.push(task);
-      omp_unset_lock(&tasksLock);
+      #ifdef HAS_OPENMP
+        omp_unset_lock(&tasksLock);
+      #endif
     }
   }
 
diff --git a/src/mlpack/methods/reinforcement_learning/worker/one_step_q_learning_worker.hpp b/src/mlpack/methods/reinforcement_learning/worker/one_step_q_learning_worker.hpp
index eaa5ae3e5f..6749fe6d12 100644
--- a/src/mlpack/methods/reinforcement_learning/worker/one_step_q_learning_worker.hpp
+++ b/src/mlpack/methods/reinforcement_learning/worker/one_step_q_learning_worker.hpp
@@ -64,7 +64,7 @@ class OneStepQLearningWorker
   void Initialize(NetworkType& learningNetwork)
   {
     updater.Initialize(learningNetwork.Parameters().n_rows,
-                       learningNetwork.Parameters().n_cols);
+        learningNetwork.Parameters().n_cols);
     // Build local network.
     network = learningNetwork;
   }
@@ -112,7 +112,9 @@ class OneStepQLearningWorker
       return false;
     }
 
-    #pragma omp atomic
+    #ifdef HAS_OPENMP
+      #pragma omp atomic
+    #endif
     totalSteps++;
 
     pending[pendingIndex] = std::make_tuple(state, action, reward, nextState);
@@ -129,7 +131,9 @@ class OneStepQLearningWorker
 
         // Compute the target state-action value.
         arma::colvec actionValue;
-        #pragma omp critical
+        #ifdef HAS_OPENMP
+          #pragma omp critical
+        #endif
         {
           targetNetwork.Predict(
               std::get<3>(transition).Encode(), actionValue);
@@ -171,7 +175,9 @@ class OneStepQLearningWorker
     // Update global target network.
     if (totalSteps % config.TargetNetworkSyncInterval() == 0)
     {
-      #pragma omp critical
+      #ifdef HAS_OPENMP
+        #pragma omp critical
+      #endif
       { targetNetwork = learningNetwork; }
     }
 
diff --git a/src/mlpack/tests/async_learning_test.cpp b/src/mlpack/tests/async_learning_test.cpp
index b210abba4a..c01ed1d9a9 100644
--- a/src/mlpack/tests/async_learning_test.cpp
+++ b/src/mlpack/tests/async_learning_test.cpp
@@ -38,7 +38,9 @@ BOOST_AUTO_TEST_CASE(OneStepQLearningTest)
    * This is for the Travis CI server, in your own machine you shuold use more
    * threads.
    */
-  omp_set_num_threads(1);
+  #ifdef HAS_OPENMP
+    omp_set_num_threads(1);
+  #endif
 
   // Set up the network.
   FFN<MeanSquaredError<>, GaussianInitialization> model(MeanSquaredError<>(),