Skip to content

DALI v1.41.0
Compare
Choose a tag to compare
@stiepan stiepan released this 29 Aug 19:14
· 82 commits to main since this release
v1.41.0
040b354

Key Features and Enhancements

This DALI release includes the following key features and enhancements:

  • Added support for CUDA 12.6. (#5596)
  • Added fn.experimental.warp_perspective operator. (#5542, #5575)
  • Added fn.random.beta random variate sampling operator. (#5550, #5571)
  • Added fn.io.file.read operator that supports loading files from dynamically specified paths. (#5552, #5572)
  • Enabled support for more simple types in fn.python_function, fn.ones, and fn.zeros. (#5598)
  • Removed unnecessary copy of tensor arguments fed into GPU operators. (#5590)

Fixed Issues

  • Reverted the fn.decoders.image* to use legacy decoders due to performance regression in nvImageCodec. (#5582, #5578, #5586)
  • Optimized S3 downloading in TFRecord reader. (#5554)
  • Added missing validation for number of inputs in argument promotion. (#5592)
  • Added missing header to support compilation with GCC 14. (#5594)
  • Fixed empty batch handling when copying batch from cpu to gpu. (#5567)

Improvements

  • Executor 2.0: ExecGraph (#5587)
  • Enable more Python types to be supported by the DALI python function (#5598)
  • Remove usages of std::call_once. (#5599)
  • Move to CUDA 12.6 (#5596)
  • Remove MakeContiguous before CPU inputs of GPU ops. (#5590)
  • nvImageCodec related fixes (#5586)
  • Mark PropagateError as [[noreturn]] (#5589)
  • Make test_beta_distribution compatible with Python 3.8 (#5571)
  • Add default_batch_size to IterationData. (#5588)
  • Add thread_setup callback to tasking::Executor (#5581)
  • Fix librosa deprecated usage (#5579)
  • Bring back the legacy image decoder operator (#5578)
  • Extract librosa's effects.trim and stft to DALI test utils, to avoid issues with breaking changes (#5568)
  • Remove libjpeg and libtiff deps (#5569)
  • Add warp_perspective operator (#5542)
  • Remove legacy image decoder (#5559)
  • Optimize S3 downloading for TFRecord reader (#5554)
  • Add io.file.read operator (#5552)
  • Add fn.random.beta random variate (#5550)
  • Reduce the batch size in the TensorFlow RN50 L3 test (#5565)
  • Use MakeContiguous when copying CPU->CPU. (#5562)
  • Update the DALI EfficientNet example to be compatible with the latest NumPy (#5561)

Bug Fixes

  • Fixes problems with fetching LFS objects during nvImageCodec conda build (#5603)
  • Fix the --python-tag option passed to python setup.py bdist_wheel command (#5600)
  • Revert "Reintroduce "Move old ImageDecoder to legacy module and make the nvImageCodec based ImageDecoder the default" (#5470)" (#5582)
  • Adding cstdint header to support GCC 14 compilation (#5594)
  • Add missing validation for input count in argument promotion (#5592)
  • Don't return pointers to a local variable in dali_operator_test. (#5585)
  • Fix operator trace caching (#5580)
  • Fix readlink usage - readlink doens't null-terminate strings. (#5577)
  • Fix WarpPerspective::GetFillValue (#5575)
  • Prevent stack-use-after-scope (#5572)
  • Add missing #include <optional> in nvcvop.h (#5570)
  • Fix MakeContiguous sample_dim for empty batches. (#5567)
  • Set affinity by device UUID. (#5566)
  • Unchecked return value from CUDA library (#5564)

Breaking API changes

  • DALI 1.39 was the final release to support the MXNet integration.

Deprecated features

No features were deprecated in this release.

Known issues:

  • The following operators: experimental.readers.fits, experimental.decoders.video, and experimental.inputs.video do not currently support checkpointing.
  • 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 frequency that is less than 10-15 frames, the returned frames might be out of sync.
  • Experimental VideoReaderDecoder does not support open GOP.
    It will not report an error and might produce invalid frames. VideoReader uses a heuristic approach to detect open GOP and should work in most common cases.
  • The DALI TensorFlow plugin might not be compatible with TensorFlow versions 1.15.0 and later.
    To use DALI with the TensorFlow version that does not have a prebuilt plugin binary shipped with DALI, make sure that the compiler that is used to build TensorFlow exists on the system during the plugin installation. (Depending on the particular version, you can use GCC 4.8.4, GCC 4.8.5, or GCC 5.4.)
  • In experimental debug and eager modes, the GPU external source is not properly synchronized with DALI internal streams.
    As a workaround, you can manually synchronize the device before returning the data from the callback.
  • Due to some known issues with meltdown/spectra mitigations and DALI, DALI shows best performance when running in Docker with escalated privileges, for example:
    • privileged=yes in Extra Settings for AWS data points
    • --privileged or --security-opt seccomp=unconfined for bare Docker.

Binary builds

NOTE: DALI builds for CUDA 12 dynamically link the CUDA toolkit. To use DALI, install the latest CUDA toolkit.

CUDA 11.0 and CUDA 12.0 builds use CUDA toolkit enhanced compatibility. 
They are built with the latest CUDA 11.x/12.x toolkit respectively but they can run on the latest, 
stable CUDA 11.0/CUDA 12.0 capable drivers (450.80 or later and 525.60 or later respectively).
However, using the most recent driver may enable additional functionality. 
More details can be found in enhanced CUDA compatibility guide.

Install via pip for CUDA 12.0:
pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/ nvidia-dali-cuda120==1.41.0
pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/ nvidia-dali-tf-plugin-cuda120==1.41.0

or just:

pip install nvidia-dali-cuda120==1.41.0
pip install nvidia-dali-tf-plugin-cuda120==1.41.0

For CUDA 11:
pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/ nvidia-dali-cuda110==1.41.0
pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/ nvidia-dali-tf-plugin-cuda110==1.41.0

or just:

pip install nvidia-dali-cuda110==1.41.0
pip install nvidia-dali-tf-plugin-cuda110==1.41.0

Or use direct download links (CUDA 12.0):

Or use direct download links (CUDA 11.0):

FFmpeg source code:

  • This software uses code of FFmpeg licensed under the LGPLv2.1 and its source can be downloaded here

Libsndfile source code:

Assets 2
Loading