DALI Release 0.10.0 Beta

The DALI 0.10.0 is a beta release. Hence, for all the features, the functionality and performance will likely be limited.

Using DALI 0.10.0 Beta

The DALI 0.10.0 can be used with the 19.06 NVIDIA GPU Cloud (NGC) optimized container for MXNet, PyTorch, and TensorFlow. Also, the 19.06 container will be shipped with DALI 0.10.0.

To upgrade to DALI 0.10.0 beta from an older version of DALI, follow the installation instructions in the DALI Quick Start Guide.

Refer to the DALI Developer Guide for usage details.

Note: The internal DALI C++ API used for operators implementation, and the C++ API that enables using DALI as a library from native code, are not yet officially supported. Hence these APIs may change in the next release without advance notice.

Key Features and Enhancements

This DALI release includes the following key features and enhancements.

  • Reduced peak memory consumption. DALI tends to do new allocation before releasing the old memory during buffer resize. As it does not copy the old memory content, the old memory can be freed before allocating the new memory.
  • Started publishing DALI nightly builds for CUDA 9 and CUDA 10, and weekly for CUDA 10.
  • Added Python function operator. Now the user can create a Python-based operator that accepts one input and produces one output.

Breaking API Changes

  • None.

Deprecated Features

  • None.

Known Issues

  • The new video reader operator requires NVIDIA VIDEO CODEC SDK support in the platform. Prior to 19.01, the NVIDIA GPU Cloud (NGC) optimized containers lack this functionality in the default configuration. To enable the functionality, run the container with the "video" capability enabled, as shown below:
    -e "NVIDIA_DRIVER_CAPABILITIES=compute,utility,video"
  • The video loader operator requires that the key frames occur at a minimum every 10 to 15 frames of the video stream. If the key frames occur at a lesser frequency, then the returned frames may be out of sync.