Optimize data input pipeline

Co-authored-by: Jiri Simsa <jsimsa@google.com>

official/resnet/imagenet_main.py

@@ -200,12 +200,13 @@ def input_fn(is_training,
     dataset = dataset.shuffle(buffer_size=_NUM_TRAIN_FILES)
 
   # Convert to individual records.
-  # cycle_length = 10 means 10 files will be read and deserialized in parallel.
-  # This number is low enough to not cause too much contention on small systems
-  # but high enough to provide the benefits of parallelization. You may want
-  # to increase this number if you have a large number of CPU cores.
-  dataset = dataset.apply(tf.data.experimental.parallel_interleave(
-      tf.data.TFRecordDataset, cycle_length=10))
+  # cycle_length = 10 means that up to 10 files will be read and deserialized in
+  # parallel. You may want to increase this number if you have a large number of
+  # CPU cores.
+  dataset = dataset.interleave(
+      tf.data.TFRecordDataset,
+      cycle_length=10,
+      num_parallel_calls=tf.data.experimental.AUTOTUNE)
 
   return resnet_run_loop.process_record_dataset(
       dataset=dataset,

official/resnet/resnet_run_loop.py

@@ -83,6 +83,11 @@ def process_record_dataset(dataset,
     tf.compat.v1.logging.info('datasets_num_private_threads: %s',
                               datasets_num_private_threads)
 
+  # Disable intra-op parallelism to optimize for throughput instead of latency.
+  options = tf.data.Options()
+  options.experimental_threading.max_intra_op_parallelism = 1
+  dataset = dataset.with_options(options)
+
   # Prefetches a batch at a time to smooth out the time taken to load input
   # files for shuffling and processing.
   dataset = dataset.prefetch(buffer_size=batch_size)
@@ -94,12 +99,10 @@ def process_record_dataset(dataset,
   dataset = dataset.repeat(num_epochs)
 
   # Parses the raw records into images and labels.
-  dataset = dataset.apply(
-      tf.data.experimental.map_and_batch(
-          lambda value: parse_record_fn(value, is_training, dtype),
-          batch_size=batch_size,
-          num_parallel_batches=num_parallel_batches,
-          drop_remainder=False))
+  dataset = dataset.map(
+      lambda value: parse_record_fn(value, is_training, dtype),
+      num_parallel_calls=tf.data.experimental.AUTOTUNE)
+  dataset = dataset.batch(batch_size, drop_remainder=False)
 
   # Operations between the final prefetch and the get_next call to the iterator
   # will happen synchronously during run time. We prefetch here again to