Deep Learning Compilers/Optimizers

List of high-performance, deep-learning-oriented IRs/compilers/Optimizers in no particular order:

• XLA (Google): https://www.tensorflow.org/performance/xla/

• taco (MIT): http://tensor-compiler.org/

• DLVM (UIUC): http://dlvm.org/ (No Longer maintained)

• Swift for Tensorflow (Google): https://github.com/tensorflow/swift (DLVM maintainers moved on to this project)

• nGraph (Intel): http://ngraph.nervanasys.com/docs/cpp/

• TVM (DMLC): https://github.com/dmlc/tvm

• Julia's Flux/DataFlow: https://github.com/MikeInnes/DataFlow.jl https://github.com/FluxML/Flux.jl (Flux is a ML framework that uses DataFlow.jl - not a lot of optimization happening yet, but has potential)

• Glow (PyTorch/FaceBook): https://github.com/pytorch/glow

General HPC graph/DAG compiler frameworks that could possibly be applied to deep learning:

• NLAFET – Parallel Numerical Linear Algebra for Extreme Scale Systems: http://www.nlafet.eu/

• OCR - Open Community Runtime (Intel): https://01.org/open-community-runtime