MLIR: A Compiler Infrastructure for the End of Moore's Law

By Chris Lattner, Jacques A. Pienaar, Mehdi Amini, Uday Bondhugula, River Riddle, Albert Cohen, Tatiana Shpeisman, Andy Davis, Nicolas Vasilache, Oleksandr Zinenko. arXiv.org, 2020

325 citations recorded by the source corpus.

This work presents MLIR, a novel approach to building reusable and extensible compiler infrastructure. MLIR aims to address software fragmentation, improve compilation for heterogeneous hardware, significantly reduce the cost of building domain specific compilers, and aid in connecting existing compilers together. MLIR facilitates the design and implementation of code generators, translators and optimizers at different levels of abstraction and also across application domains, hardware targets and execution environments. The contribution of this work includes (1) discussion of MLIR as a research artifact, built for extension and evolution, and identifying the challenges and opportunities posed by this novel design point in design, semantics, optimization specification, system, and engineering. (2) evaluation of MLIR as a generalized infrastructure that reduces the cost of building compilers-describing diverse use-cases to show research and educational opportunities for future programming languages, compilers, execution environments, and computer architecture. The paper also presents the rationale for MLIR, its original design principles, structures and semantics.

Subfields: Artificial Intelligence, Hardware and Architecture, Information Systems and Management

View Paper: https://www.semanticscholar.org/paper/a38e4c9772172beb56c184a9b5c8fbb77cf18457

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Yoshua Bengio
2. Sara Hooker
3. Torsten Hoefler
4. Tal Ben-Nun
5. Sergey Levine
6. Mohit Bansal
7. Samy Bengio
8. Ruslan Salakhutdinov

In Hardware and Architecture

1. Tal Ben-Nun
2. Andrei Ivanov
3. Torsten Hoefler
4. Samy Bengio
5. Tim Harris
6. Nikoli Dryden
7. Shigang Li
8. Ryota Tomioka

In Information Systems and Management

1. Tal Ben-Nun
2. Torsten Hoefler
3. Emily L. Denton
4. Sara Hooker
5. Muntasir Wahed
6. N. Ahmed

Citations in the corpus

• Characterising Bias in Compressed Models (2020, methodology): “The second type uses fixed-point only 8-bit weights (Vanhoucke et al., 2011; Jacob et al., 2018 (Jacob et al., 2018; Lattner et al., 2020).”
• Dynamic Tensor Rematerialization (2020, background): “In other frameworks that feature a compilation step [31, 32, 33], it may be possible to eliminate much of this overhead by generating these structures in a compiler pass.”
• The Deep Learning Compiler: A Comprehensive Survey (2020, background): “dependence graphs [35] (DDG) of generic compilers [30], [36].”
• Data Movement Is All You Need: A Case Study on Optimizing Transformers (2020, methodology): “Other white-box approaches for optimization exist (e.g., Halide (Ragan-Kelley et al., 2013) and MLIR (Lattner et al., 2020)).”
• DistIR: An Intermediate Representation for Optimizing Distributed Neural Networks (2021, methodology): “DistIR uses MLIR’s dialect-based design [19] to be parametric with respect to its primitive operators.”