Software deprecation

End of support for NeuronCore Groups (NCG)

10/27/2021 - Before the introduction of Neuron Runtime 2.x, NeuronCore Group (NCG) has been used by Neuron Runtime 1.x to define an execution group of one or more NeuronCores where models can be loaded and executed. It also provided separation between processes.

With the introduction of Neuron Runtime 2.x, the strict separation of NeuronCores into groups is no longer needed and NeuronCore Groups (NCG) is deprecated. Neuron Runtime 2.x enables each process to own a set of NeuronCores, and within each process, Neuron Runtime 2.x supports loading and executing multiple models on separate , different or overlapping sets of NeuronCores.

Please note that NEURONCORE_GROUP_SIZES environment variable is in the process of being deprecated, and for a transition period NEURONCORE_GROUP_SIZES can be used to preserve the old NeuronCore Group behavior. The frameworks internally would convert NEURONCORE_GROUP_SIZES to use runtime’s new mode of mapping models to NeuronCores.

For more information see details about NEURON_RT_VISIBLE_CORES at Neuron Runtime Configuration and and Migrate your application to Neuron Runtime 2.x (libnrt.so).

Announcing end of support for NEURONCORE_GROUP_SIZES

10/27/2021 - NEURONCORE_GROUP_SIZES environment variable is in the process of being deprecated, future Neuron releases may no longer support the NEURONCORE_GROUP_SIZES environment variable. Please start using NEURON_RT_VISIBLE_CORES instead.

See End of support for NeuronCore Groups (NCG), Neuron Runtime Configuration and Migrate your application to Neuron Runtime 2.x (libnrt.so) for more information.

End of support for Neuron Conda packages in Deep Learning AMI starting Neuron 1.14.0

05/28/2021 - Starting with Neuron SDK 1.14.0, we will no longer support conda packages to install Neuron SDK framework in DLAMI and we will no longer update conda packages used to install Neuron SDK framework (Neuron conda packages) with new versions.

Starting with Neuron SDK 1.14.0, pip packages (Neuron pip packages) will be used to install Neuron SDK framework in DLAMI conda environment. To upgrade Neuron SDK framework DLAMI users should use pip upgrade commands instead of conda update commands. Instructions are available in this blog and in Neuron SDK documentation (https://awsdocs-neuron.readthedocs-hosted.com/en/latest/neuron-intro/neuron-install-guide.html#deep-learning-ami-dlami).

Starting with Neuron SDK 1.14.0, run one of the following commands to upgrade to latest Neuron framework of your choice:

  • To upgrade Neuron PyTorch:

source activate aws_neuron_pytorch_p36
pip config set global.extra-index-url https://pip.repos.neuron.amazonaws.com
pip install --upgrade torch-neuron neuron-cc[tensorflow] torchvision
  • To upgrade Neuron TensorFlow:

source activate aws_neuron_tensorflow_p36
pip config set global.extra-index-url https://pip.repos.neuron.amazonaws.com
pip install --upgrade tensorflow-neuron tensorboard-neuron neuron-cc
  • To upgrade Neuron MXNet:

source activate aws_neuron_mxnet_p36
pip config set global.extra-index-url https://pip.repos.neuron.amazonaws.com
pip install --upgrade mxnet-neuron neuron-cc

For more information please check the blog.

End of support for Ubuntu 16 starting Neuron 1.14.0

05/01/2021 - Ubuntu 16.04 entered end of life phase officially in April 2021 (see https://ubuntu.com/about/release-cycle) and will not receive any public software or security updates. Starting with Neuron SDK 1.14.0, Ubuntu 16 is no longer supported for Neuron, users who are using Ubuntu 16 are requested to migrate to Ubuntu18 or Amazon Linux 2.

Customers who choose to upgrade libc on Ubuntu 16 to work with Neuron v1.13.0 (or higher versions) are highly discouraged from doing that since Ubuntu 16 will no longer receive public security updates.

End of support for classic TensorBoard-Neuron starting Neuron 1.13.0 and introducing Neuron Plugin for TensorBoard

05/01/2021 - Starting with Neuron SDK 1.13.0, we are introducing Neuron Plugin for TensorBoard and we will no longer support classic TensorBoard-Neuron. Users are required to migrate to Neuron Plugin for TensorBoard.

Starting with Neuron SDK 1.13.0, if you are using TensorFlow-Neuron within DLAMI Conda environment, attempting to run tensorboard with the existing version of TensorBoard will fail. Please update the TensorBoard version before installing the Neuron plugin by running pip install TensorBoard --force-reinstall, for installation instructions see Neuron Plugin for TensorBoard.

Users who are using Neuron SDK releases before 1.13.0, can find classic TensorBoard-Neuron documentation at Neuron 1.12.2 documentation.

For more information see see Neuron Plugin for TensorBoard Release Notes and Neuron Plugin for TensorBoard.

End of support for Python 3.5

2/24/2021 - As Python 3.5 reached end-of-life in October 2020, and many packages including TorchVision and Transformers have stopped support for Python 3.5, we will begin to stop supporting Python 3.5 for frameworks, starting with PyTorch-Neuron version [1.1.7.0] in this release. You can continue to use older versions with Python 3.5.

End of support for ONNX

11/17/2020 - ONNX support is limited and from this version onwards we are not planning to add any additional capabilities to ONNX. We recommend running models in TensorFlow, PyTorch or MXNet for best performance and support.

End of support for PyTorch 1.3

7/16/2020 - Starting this release we are ending the support of PyTorch 1.3 and migrating to PyTorch 1.5.1, customers are advised to migrate to PyTorch 1.5.1.