TensorFlow Release 18.07
The NVIDIA container image of TensorFlow, release 18.07, is available.
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 as a system Python module.
To achieve optimum TensorFlow performance, for image based training, the container includes a sample script that demonstrates efficient training of convolutional neural networks (CNNs). The sample script may need to be modified to fit your application. The container also includes the following:
- Ubuntu 16.04
- NVIDIA CUDA 9.0.176 (see Errata section and 2.1) including CUDA® Basic Linear Algebra Subroutines library™ (cuBLAS) 9.0.425
- NVIDIA CUDA® Deep Neural Network library™ (cuDNN) 7.1.4
- NCCL 2.2.13 (optimized for NVLink™ )
- Horovod™ 0.12.1
- OpenMPI™ 3.0.0
- TensorBoard 1.8.0
- MLNX_OFED 3.4
- OpenSeq2Seq v0.4 at commit 98ad236a.
- TensorRT 4.0.1
- DALI 0.1 Beta
Release 18.07 is based on CUDA 9, which requires NVIDIA Driver release 384.xx.
Key Features and Enhancements
- TensorFlow container image version 18.07 is based on TensorFlow 1.8.0.
- Added support for DALI 0.1 Beta.
- Latest version of CUDA® Basic Linear Algebra Subroutines library™ (cuBLAS) 9.0.425.
- Ubuntu 16.04 with June 2018 updates
Accelerating Inference In TensorFlow With TensorRT (TF-TRT)
For step-by-step instructions on how to use TF-TRT, see Accelerating Inference In TensorFlow With TensorRT User Guide.
- Key Features And Enhancements
Added TensorRT 4.0 API support with extended layer support. This support includes the FullyConnected layer and
Resource management added, where memory allocation is uniformly managed by TensorFlow.
Bug fixes and better error handling in conversion.
TensorRT conversion relies on static shape inference, where the frozen graph should provide explicit dimension on all ranks other than the first batch dimension.
Batchsize for converted TensorRT engines are fixed at conversion time. Inference can only run with batchsize smaller than the specified number.
Current supported models are limited to CNNs. Object detection models and RNNs are not yet supported.
- Known Issues
Input tensors are required to have rank 4 for quantization mode (INT8 precision).
Starting with the next major version of CUDA release, we will no longer provide updated Python 2 containers and will only update Python 3 containers.
There are no known issues in this release.