Release 19.08
The container image for NVIDIA Optimized Deep Learning Framework, powered by Apache MXNet, release 19.08, is available on NGC.
Contents of the Optimized Deep Learning Frameworkcontainer
This container image contains the complete source of the version of NVIDIA Optimized Deep Learning Framework, powered by Apache MXNet in /opt/mxnet
. It is pre-built and installed to the Python path.
The container also includes the following:
- Ubuntu 18.04 including Python 3.5
- NVIDIA CUDA 10.1.243 including cuBLAS 10.2.1.243
- NVIDIA cuDNN 7.6.2
- NVIDIA NCCL 2.4.8 (optimized for NVLink™ )
- ONNX exporter 0.1 for CNN classification models
Note:
The ONNX exporter is being continuously improved. You can try the latest changes by pulling from the main branch.
- Amazon Labs Sockeye sequence-to-sequence framework 1.18.99 (for machine translation)
- MLNX_OFED +4.0
- OpenMPI 3.1.4
- Horovod 0.16.4
- Nsight Compute 10.1.168
- Nsight Systems 2019.3.7.9
- TensorRT 5.1.5
- GluonCV Toolkit 0.4
- GluonNLP Toolkit 0.7.1
- DALI 0.12.0 Beta
- Tensor Core optimized example:
- Jupyter and JupyterLab:
Driver Requirements
Release 19.08 is based on NVIDIA CUDA 10.1.243, which requires NVIDIA Driver release 418.87. However, if you are running on Tesla (Tesla V100, Tesla P4, Tesla P40, or Tesla P100), you may use NVIDIA driver release 384.111+ or 410. The CUDA driver's compatibility package only supports particular drivers. For a complete list of supported drivers, see the CUDA Application Compatibility topic. For more information, see CUDA Compatibility and Upgrades.
GPU Requirements
Release 19.08 supports CUDA compute capability 6.0 and higher. This corresponds to GPUs in the Pascal, Volta, and Turing families. Specifically, for a list of GPUs that this compute capability corresponds to, see CUDA GPUs. For additional support details, see Deep Learning Frameworks Support Matrix.
Key Features and Enhancements
This Optimized Deep Learning Framework release includes the following key features and enhancements.
- NVIDIA Optimized Deep Learning Framework, powered by Apache MXNet container image version 19.08 is based on Apache MXNet 1.5.0 and includes upstream commits up through commit 75a9e187d from June 27, 2019.
- Latest version of NVIDIA CUDA 10.1.243 including cuBLAS 10.2.1.243
- Latest version of NVIDIA cuDNN 7.6.2
- Latest version of NVIDIA NCCL 2.4.8
- Latest version of Horovod 0.16.4
- Added GluonCV Toolkit 0.4 and GluonNLP Toolkit 0.7.1 to the container.
- Latest version of DALI 0.12.0 Beta
- Latest version of OpenMPI 3.1.4
- Latest version of Nsight Systems 2019.3.7.9
- Latest version of MLNX_OFED +4.0
- Latest versions of Jupyter Notebook 6.0.0 and JupyterLab 1.0.5
- Included support for length parameter in softmax (PR 15159)
- Improved performance of forward computation of the
softmax
operator - Improved performance of forward computation of the
FullyConnected
operator when using bias. - Added a demo for fast BERT QA inference using Apache MXNet in
/workspace/examples/gluon/bert_inference.
- Improved latency of inference done with networks using the
Dropout
operator. - Added experimental support for CUDA Graphs inside Apache MXNet guarded by the
MXNET_EXPERIMENTAL_ENABLE_CUDA_GRAPH
environment variable. - Ubuntu 18.04 with July 2019 updates
Tensor Core Examples
These examples focus on achieving the best performance and convergence from NVIDIA Volta Tensor Cores by using the latest deep learning example networks for training. Each example model trains with mixed precision Tensor Cores on Volta, therefore you can get results much faster than training without tensor cores. This model is tested against each NGC monthly container release to ensure consistent accuracy and performance over time. This container includes the following tensor core examples.
- The ResNet50 v1.5 model is a slightly modified version of the original ResNet50 v1 model that trains to a greater accuracy. This model script is available on GitHub as well as NVIDIA GPU Cloud (NGC).
Automatic Mixed Precision (AMP)
Training deep learning networks is a very computationally intensive task. Novel model architectures tend to have an increasing number of layers and parameters, which slows down training. Fortunately, new generations of training hardware as well as software optimizations make training these new models a feasible task.
Most of the hardware and software training optimization opportunities involve exploiting lower precision like FP16 in order to utilize the Tensor Cores available on new Volta and Turing GPUs. While training in FP16 showed great success in image classification tasks, other more complicated neural networks typically stayed in FP32 due to difficulties in applying the FP16 training guidelines that are needed to ensure proper model training.
That is where AMP (Automatic Mixed Precision) comes into play- it automatically applies the guidelines of FP16 training, using FP16 precision where it provides the most benefit, while conservatively keeping in full FP32 precision operations unsafe to do in FP16.
The NVIDIA Optimized Deep Learning Framework, powered by Apache MXNet AMP tutorial, located in /opt/mxnet/nvidia-examples/AMP/AMP_tutorial.md
inside this container, shows how to get started with mixed precision training using AMP for Apache MXNet, using by example the SSD network from GluonCV.
For more information about AMP, see the Training With Mixed Precision Guide.
Known Issues
- The Apache MXNet KVStore GPU peer-to-peer communication tree discovery, as of release 18.09, is not compatible with DGX-1V. Only users that set the environment variable
MXNET_KVSTORE_USETREE=1
will experience issues, which will be resolved in a subsequent release. Issue tracked under 13341. - The default setting of the environment variable
MXNET_GPU_COPY_NTHREADS=1
in the container may not be optimal for all networks. Networks with a high ratio of parameters and computation, like AlexNet, may achieve greater multi-GPU training speeds with the settingMXNET_GPU_COPY_NTHREADS=2
. Users are encouraged to try this setting for their own use case. - There is a known issue in the BERT QA demo, where for some values of sequence length the inference may fail with
CUDA Driver: operation failed due to a previous error during capture
error. To test those values of sequence length, change the lineexport MXNET_EXPERIMENTAL_ENABLE_CUDA_GRAPH=1
toexport MXNET_EXPERIMENTAL_ENABLE_CUDA_GRAPH=0
in thetest_bert_inference
script inside the demo directory.