TensorFlow Release 16.12
The NVIDIA container image of TensorFlow, release 16.12, is available.
TensorFlow container image version 16.12 is based on TensorFlow 0.12.0.
Contents of TensorFlow
This container image contains the complete source of the version of NVIDIA TensorFlow in /opt/tensorflow. It is pre-built and installed into the /usr/local/[bin,lib] directories in the container image.
The container also includes the following:
- Ubuntu 14.04
- NVIDIA CUDA® 8.0.54
- NVIDIA CUDA® Deep Neural Network library™ (cuDNN) 6.0.5
- NVIDIA® Collective Communications Library ™ (NCCL) 1.6.1 (optimized for NVLink™ )
Key Features and Enhancements
This TensorFlow release includes the following key features and enhancements.
- Supports multi-GPU training
- [BETA] NCCL integration for improved multi-GPU scaling
Note:
Requires explicit use by the model script.
- [BETA] NCCL integration for improved multi-GPU scaling
- Supports recurrent neural networks
- Support for cuDNN recurrent neural networks (RNN) layers
Note:
Requires explicit use by the model script.
- Support for cuDNN recurrent neural networks (RNN) layers
- Better I/O throughput via
libjpeg-turbo, fast iDCT decoding - Support for the non-fused Winograd algorithm for improved convolution performance.
- TensorBoard; a data visualization toolkit
- Several built-in TensorFlow examples
- Ubuntu 14.04 with November 2016 updates
Known Issues
There are no known issues in this release.