uses channel to replace the traditional buffer

Author: JiayiFeng

paddle/fluid/operators/reader/create_double_buffer_reader_op.cc

@@ -12,42 +12,35 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include <condition_variable>
-#include <mutex>
 #include <thread>
+#include "paddle/fluid/framework/channel.h"
 #include "paddle/fluid/operators/reader/reader_op_registry.h"
 
 namespace paddle {
 namespace operators {
 namespace reader {
 
-static constexpr size_t kDoubleBufferSize = 3;
+static constexpr size_t kDoubleBufferSize = 2;
 
 class DoubleBufferReader : public framework::DecoratedReader {
  public:
   explicit DoubleBufferReader(ReaderBase* reader)
       : DecoratedReader(reader),
-        buffer_(kDoubleBufferSize),
-        write_pos_(0),
-        read_pos_(0) {
-    std::thread prefetch(
-        std::bind(&DoubleBufferReader::PrefetchThreadFunc, this));
+        buffer_(framework::MakeChannel<std::vector<framework::LoDTensor>>(
+            kDoubleBufferSize)) {
+    std::thread prefetch(&DoubleBufferReader::PrefetchThreadFunc, this);
     prefetch.detach();
   }
 
   void ReadNext(std::vector<framework::LoDTensor>* out) override;
-  bool HasNext() const override;
+  void ReInit() override;
+
+  ~DoubleBufferReader() { buffer_->Close(); }
 
  private:
   void PrefetchThreadFunc();
 
-  std::vector<std::vector<framework::LoDTensor>> buffer_;
-  size_t write_pos_;
-  size_t read_pos_;
-
-  std::mutex mtx_;
-  std::condition_variable buffer_not_full_;
-  std::condition_variable buffer_not_empty_;
+  framework::Channel<std::vector<framework::LoDTensor>>* buffer_;
 };
 
 class CreateDoubleBufferReaderOp : public framework::OperatorBase {
@@ -80,44 +73,36 @@ class CreateDoubleBufferReaderOpMaker : public DecoratedReaderMakerBase {
 };
 
 void DoubleBufferReader::ReadNext(std::vector<framework::LoDTensor>* out) {
-  std::unique_lock<std::mutex> lck(mtx_);
-  while (write_pos_ == read_pos_) {
-    buffer_not_empty_.wait(lck);
-  }
-
   out->clear();
-  out->reserve(buffer_[read_pos_].size());
-  // TODO(fengjiayi): This copy shall be reduced.
-  for (size_t i = 0; i < buffer_[read_pos_].size(); ++i) {
-    framework::LoDTensor dst;
-    TensorCopy(buffer_[read_pos_][i], platform::CPUPlace(), &dst);
-    dst.set_lod(buffer_[read_pos_][i].lod());
-    out->push_back(dst);
-  }
-
-  ++read_pos_;
-  if (read_pos_ >= kDoubleBufferSize) {
-    read_pos_ = 0;
-  }
-  buffer_not_full_.notify_all();
+  buffer_->Receive(out);
 }
 
-bool DoubleBufferReader::HasNext() const {
-  return reader_->HasNext() || !buffer_.empty();
+void DoubleBufferReader::ReInit() {
+  reader_->ReInit();
+  buffer_->Close();
+  // The existing prefetch thread will terminate for the buffer_ is closed.
+  buffer_ = framework::MakeChannel<std::vector<framework::LoDTensor>>(
+      kDoubleBufferSize);
+  std::thread prefetch(&DoubleBufferReader::PrefetchThreadFunc, this);
+  prefetch.detach();
 }
 
 void DoubleBufferReader::PrefetchThreadFunc() {
-  while (reader_->HasNext()) {
-    std::unique_lock<std::mutex> lck(mtx_);
-    while (((write_pos_ + 1) % kDoubleBufferSize) == read_pos_) {
-      buffer_not_full_.wait(lck);
+  VLOG(5) << "A new prefetch thread starts.";
+  while (true) {
+    std::vector<framework::LoDTensor> batch;
+    reader_->ReadNext(&batch);
+    if (batch.empty()) {
+      // EOF
+      buffer_->Close();
+      VLOG(5) << "Reached the end of the file. The prefetch thread terminates.";
+      break;
     }
-    reader_->ReadNext(&buffer_[write_pos_]);
-    ++write_pos_;
-    if (write_pos_ >= kDoubleBufferSize) {
-      write_pos_ = 0;
+    if (!buffer_->Send(&batch)) {
+      VLOG(5) << "WARNING: The double buffer channel has been closed. The "
+                 "prefetch thread terminates.";
+      break;
     }
-    buffer_not_empty_.notify_all();
   }
 }