From: A. Unique TensorFlower <gardener@tensorflow.org>
Date: Thu, 4 Jan 2018 00:06:24 +0000 (-0800)
Subject: Update RNN/LSTM performance docs
X-Git-Tag: v1.6.0-rc0~304^2~5^2~43
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=9f816c90b621b59286a3b39faf213384d0563401;p=platform%2Fupstream%2Ftensorflow.git

Update RNN/LSTM performance docs

PiperOrigin-RevId: 180730491
---

diff --git a/tensorflow/docs_src/performance/performance_guide.md b/tensorflow/docs_src/performance/performance_guide.md
index 0fa1fc38c2..4850e62c12 100644
--- a/tensorflow/docs_src/performance/performance_guide.md
+++ b/tensorflow/docs_src/performance/performance_guide.md
@@ -224,11 +224,7 @@ On NVIDIA GPUs, the use of @{tf.contrib.cudnn_rnn} should always be preferred
 unless you want layer normalization, which it doesn't support.  It is often at
 least an order of magnitude faster than @{tf.contrib.rnn.BasicLSTMCell} and
 @{tf.contrib.rnn.LSTMBlockCell} and uses 3-4x less memory than
-@{tf.contrib.rnn.BasicLSTMCell}.  Unfortunately, @{tf.contrib.cudnn_rnn} is not
-compatible with @{tf.train.SyncReplicasOptimizer} so you should either use a
-different synchronization mechanism (consider an all-reduce based strategy) or
-use the @{tf.contrib.rnn.LSTMBlockFusedCell} (at a significant performance
-penalty).
+@{tf.contrib.rnn.BasicLSTMCell}.
 
 If you need to run one step of the RNN at a time, as might be the case in
 reinforcement learning with a recurrent policy, then you should use the