Neuron SDK Release - August 5, 2020 (#145)

Release notes for Neuron SDK Release - August 5, 2020

Author: micwade-aws

release-notes/README.md

@@ -1,5 +1,8 @@
 # Neuron Release Notes
 
+## August 4, 2020 Release
+Bug fix for a latent issue caused by a race condition in Neuron Runtime leading to possible crashes. The crash was observed under stress load conditons.  All customers are encouraged to update the latest Neuron Runtime package (aws-neuron-runtime), version 1.0.8813.0 or newer.  Other Neuron packages are being updated as well, but are to be considered non-critical updates.
+
 ## July 16, 2020 Release
 This release of Neuron SDK adds support for the OpenPose (posenet) Neural Network. An example of using Openpose for end to end inference is available [here](../src/examples/tensorflow/openpose_demo/).
 

release-notes/conda/conda-mxnet-neuron.md

@@ -2,6 +2,17 @@
 
 This document lists the release notes for the Neuron Conda-MxNet package.
 
+# [1.5.1.1.0.2101.0-2.0.631.0]
+
+Date: 08/05/2020
+
+## Included Neuron Packages
+
+neuron_cc-1.0.17937.0
+
+mxnet_neuron-1.5.1.1.0.2101.0
+
+
 # [1.5.1.1.0.2093.0-2.0.579.0]
 
 Date: 07/16/2020

release-notes/conda/conda-tensorflow-neuron.md

@@ -2,6 +2,19 @@
 
 This document lists the release notes for the Neuron Conda-TensorFlow package.
 
+# [1.15.3.1.0.1953.0_2.0.769.0]
+
+Date: 08/05/2020
+
+## Included Neuron Packages
+
+neuron_cc-1.0.17937.0
+
+tensorboard_neuron-1.15.0.1.0.513.0
+
+tensorflow_neuron-1.15.3.1.0.1889.0
+
+
 # [1.15.3.1.0.1891.0-2.0.706.0]
 
 Date: 07/16/2020

release-notes/conda/conda-torch-neuron.md

@@ -4,6 +4,22 @@ This document lists the release notes for the Neuron Conda-Pytorch package.
 
 ### [Conda Tensorflow Release Notes](../tensorflow-neuron.md)
 
+
+# [1.5.1.1.0.258.0_2.0.871.0]
+
+Date: 08/05/2020
+
+## Included Neuron Packages
+
+[neuron-cc-1.0.17937.0](../neuron-cc.md)
+
+[torch_neuron-1.0.1522.0](../torch-neuron.md)
+
+torch_neuron_base-1.5.1.1.0.258.0
+
+
+
+
 # [1.5.1.1.0.251.0_2.0.783.0]
 
 Date: 07/16/2020

release-notes/mxnet-neuron.md

@@ -2,11 +2,25 @@
 
 This document lists the release notes for MXNet-Neuron framework.
 
-# Known Issues 07/16/2020
+# Known Issues 08/05/2020
 
 * Issue: MXNet Model Server is not able to clean up Neuron RTD states after model is unloaded (deleted) from model server.
   * Workaround: run “`/opt/aws/neuron/bin/neuron-cli reset`“ to clear Neuron RTD states after all models are unloaded and server is shut down.
 
+# [1.5.1.1.0.2101.0]
+
+Date 08/05/2020
+
+## Summary
+
+Various minor improvements.
+
+## Major New Features
+
+## Resolved Issues
+
+
+
 # [1.5.1.1.0.2093.0]
 
 Date 07/16/2020

release-notes/neuron-cc-ops/neuron-cc-ops-mxnet.md

@@ -1,5 +1,10 @@
 # Supported operators [MXNet]
 
+### Neuron Compiler Release [1.0.17937.0]
+
+No changes
+
+
 ### Neuron Compiler Release [1.0.16861.0]
 
 Removed ```log``` (Was erroneously reported as added in previous release. )

release-notes/neuron-cc-ops/neuron-cc-ops-onnx.md

@@ -1,5 +1,10 @@
 # Supported operators [ONNX]
 
+
+### Neuron Compiler Release [1.0.17937]
+
+No changes
+
 ### Neuron Compiler Release [1.0.16861]
 
 No changes

release-notes/neuron-cc-ops/neuron-cc-ops-pytorch.md

@@ -7,6 +7,10 @@ import torch.neuron
 print(*torch.neuron.get_supported_operations(), sep='\n')
 ```
 
+
+### PyTorch Neuron Release [1.0.1522.0]
+* No change
+
 ### PyTorch Neuron Release [1.0.1386.0]
 
 Added the following instructions. Please note, primitives are included in this list from this release.

release-notes/neuron-cc.md

@@ -14,7 +14,7 @@ The supported operators are also listed here:
 * [Neuron-cc MXNet Operators](./neuron-cc-ops/neuron-cc-ops-mxnet.md)
 * [Neuron-cc ONNX Operators](./neuron-cc-ops/neuron-cc-ops-onnx.md)
 
-## Known issues and limitations - updated 07/16/2020
+## Known issues and limitations - updated 08/05/2020
 
 1. **Control flow** Neuron only supports control flow operators which are static at compile time, i.e. static length RNN, top-k, sort, ...
 2. **Size of neural network** The size of neural network is influenced by a) type of neural network (CNN, LSTM, MLP) , b) number of layers, c) sizes of input (dimension of the tensors, batch size, ...). The current Neuron compiler release has a limitation in terms of the size of neural network it could effectively optimize. As a result, we limit CNN models (e.g. ResNet) to have an input size of up to 480x480 fp16/32, batch size=4; LSTM models (e.g. GNMT) are limited to a time step limit of up to 900; MLP models (like BERT) are limited up to sequence-length=128, batch=8.
@@ -28,6 +28,21 @@ The supported operators are also listed here:
 
 
 
+# [1.0.17937.0]
+
+Date 8/05/2020
+
+## Summary
+
+Various minor improvements. 
+
+## Major New Features
+
+## Resolved Issues
+
+
+
+
 # [1.0.16861.0]
 
 Date 7/16/2020

release-notes/neuron-runtime.md

@@ -13,6 +13,21 @@ Use this table to determine the version of Runtime that will support the version
 
 ---
 
+# [1.0.8813.0]
+
+Date: 08/05/2020
+
+## Summary
+Patching a bug from prior versions that could lead to crashes under load.
+
+## Major New Features
+* n/a
+
+## Resolved Issues
+* Fixed a race condition in the runtime that was leading to crashes in some cases of load testing.  
+
+
+
 # [1.0.8444.0]
 
 Date: 07/16/2020