5.1. PopTorch


New features

  • Upgraded from PyTorch 1.10 to 1.13.

  • Added support for variables being sharded across replicas.

  • poptorch.set_overlap_for_input and poptorch.set_overlap_for_output can now be applied to tuples, lists, and dicts of tensors.

  • PopTorch now catches aten::lstm directly when compiling with dispatch for PopART, allowing set_available_memory to work with it.

  • Added support for aten::index_fill_.int_Scalar.

  • Added support for dict inputs.

  • Added support for torch.count_nonzero.

  • Support the tanh approximation for GELU.

  • Added support for torch.scatter_reduce operation.

Bug Fixes

  • Fixed clamp_max in cases where the max is large.

  • Fixed shape inference failing on PopART for argsort, GRU and norm ops.

  • Fixed shape inference for strided slices.

  • Fixed casting of groupnorm.

  • Fixed an issue where the alpha and beta arguments were flipped for torch.addmm.

  • Fixed a “not representable” error when using BCEWithLogitsLoss with a dtype of half.

  • Fixed intermittent compilation hang caused by tqdm (progress bar).

Other improvements

  • Fixed in-place modification of slice regions.

  • Documentation typo fixes and clarifications.

  • Improved error message when encountering CPU tensors.

  • Use the IPU DispatchKey instead of the XLA DispatchKey, which means that error messages will now mention IPU rather XLA.

Known issues


Compatibility changes

  • Dropped support for Python 3.6 (in order to upgrade to PyTorch 1.13).

  • Removed support for torch.jit.trace(). For help on migration issues when using the dispatcher frontend, see the Legacy tracing frontend section in the 3.0.0 version of the PyTorch for the IPU: User Guide.

  • Removed support for building on CentOS 7.x.

  • Removed the Autocast API (this was only available when using the tracing frontend).


New features

  • Enable the dispatcher by default.

    The dispatcher frontend replaces the tracing frontend and provides many benefits, including better performance and fewer and simpler PopTorch-specific coding requirements.

    The dispatcher frontend is now enabled by default, so no special changes are required to use it. See the Legacy tracing frontend section in the 3.0.0 version of the PyTorch for the IPU: User Guide for help on specific migration issues.

  • Add support for interactive querying of poptorch.Options.

  • Support Tensor.unfold in PyTorch.

  • Improve error message for aten::item in static graphs.

  • Support dictionary input in PopTorch.

  • Improve error reporting for lists and tuples not being supported for set_overlap_for_input.

  • Support for tensor.new_full.

  • Improved performance of index_put when the indices are a one dimensional vector.

  • Support for torch_scatter.

Bug Fixes

  • Fixed BatchNorm running statistics. It now usies the unbiased estimator to update running_var at training time.

  • Scalar tensor inputs to the graph now working.

  • Fixed expand, when the desired shape contained both added dimensions and -1.

  • Fixed a bug where torch.nn.functional.kl_div would produce NaN.

Other improvements


Known issues


Compatibility changes