What are the options for generating multi-core code using MLIR? Ideally, I’d like to have statically-parallelized code (fork-join) for transforms such as TensorFlow’s Conv2d. What back-end and what transforms should I look into?
What are the options for generating multi-core code using MLIR? Ideally, I’d like to have statically-parallelized code (fork-join) for transforms such as TensorFlow’s Conv2d. What back-end and what transforms should I look into?