2019 EuroLLVM Developers’ Meeting: T. Shpeisman & C. Lattner “MLIR: Multi-Level Intermediate Repr..”

llvm.org/devmtg/2019-04/
-
MLIR: Multi-Level Intermediate Representation for Compiler Infrastructure - Tatiana Shpeisman (Google), Chris Lattner (Google)
Slides: llvm.org/devmtg/2019-04/slides...
-
This talk will give an overview of Multi-Level Intermediate Representation - a new intermediate representation designed to provide a unified, flexible and extensible intermediate representation that is language-agnostic and can be used as a base compiler infrastructure. MLIR shares similarities with traditional CFG-based three-address SSA representations (including LLVM IR or SIL), but it also introduces notions from the polyhedral domain as first class concepts. The notion of dialects is a core concept of MLIR extensibility, allowing multiple levels in a single representation. MLIR supports the continuous lowering from dataflow graphs to high-performance target specific code through partial specialization between dialects. We will illustrate in this talk how MLIR can be used to build an optimizing compiler infrastructure for deep learning applications.
MLIR supports multiple front- and back-ends and uses LLVM IR as one of its primary code generation targets. MLIR also relies heavily on design principles and practices developed by the LLVM community. For example, it depends on LLVM APIs and programming idioms to minimize IR size and maximize optimization efficiency. MLIR uses LLVM testing utilities such as FileCheck to ensure robust functionality at every level of the compilation stack, TableGen to express IR invariants, and it leverages LLVM infrastructure such as dominance analysis to avoid implementing all the necessary compiler functionalities from scratch. At the same time, it is a brand new IR, both more restrictive and more general than LLVM IR in different aspects of its design. We believe that the LLVM community will find in MLIR a useful tool for developing new compilers, especially in machine learning and other high-performance domains.
-
Videos Filmed & Edited by Bash Films: www.BashFilms.com

[@dontlikethetube]

Seemed like hard questions from the audience! Glad there are crazy brains solving these kinds of problems and a way for those solutions to be high impact across a bunch of language implementations!

[@MrM12LRV]

Such an exciting project : )

[@WenZong]

MLIR introduction in Xiangsheng form

[@8-P]

Is the Tutorial @ 31:30 available somewhere?

[@LLVMPROJ]

All videos are coming. Please subscribe to our channel.

[@8-P]

@@LLVMPROJ The Tutorial is available here: kzbin.info/www/bejne/maqsdIh_j5pprbM Thank you! :)

[@tonyjiang4991]

MLIR is just another level of IR above LLVM IR to help higher level of optimizations introduced due to recent year trend (like DL)

[@razinedrive4757]

keep up

[@kyleschlicht4800]

Pls make a compiler to compile tensorflow into MIT scratch

[@malharjajoo7393]

5:37 - honestly, this slide doesnt really explain "Why do we need MLIR ?" for a beginner ...

[@tom.zhangmingfnegtom.zhang6163]

maybe MLIR is more flexible than TVM. In google AI teams perspective , they think MLIR maybe in the future will be the standard for AI middle-layer framework

[@blanamaxima]

Very annoying presentation.

[@blanamaxima]

true

[@abharti]

kzbin.info/www/bejne/pKLEkIivr5ajeKc