Polygeist: Raising C to Polyhedral MLIR

We present Polygeist, a new compilation flow that connects the MLIR compiler infrastructure to cutting edge polyhedral optimization tools. It consists of a C and C++ frontend capable of converting a broad range of existing codes into MLIR suitable for polyhedral transformation and a bi-directional c...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:2021 30th International Conference on Parallel Architectures and Compilation Techniques (PACT) S. 45 - 59
Hauptverfasser: Moses, William S., Chelini, Lorenzo, Zhao, Ruizhe, Zinenko, Oleksandr
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 01.09.2021
Schlagworte:
Affine
Automatic Parallelization
Benchmark testing
Bidirectional control
C++
C++ languages
Codes
Compiler
LLVM
MLIR
OpenMP
OpenScop
Optimization
Parallel
Parallel architectures
Pluto
Polly
Polybench
Polygeist
Polyhedral
Raising
Reduction
Splitting
Transformation
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:We present Polygeist, a new compilation flow that connects the MLIR compiler infrastructure to cutting edge polyhedral optimization tools. It consists of a C and C++ frontend capable of converting a broad range of existing codes into MLIR suitable for polyhedral transformation and a bi-directional conversion between MLIR and OpenScop exchange format. The Polygeist/MLIR intermediate representation featuring high-level (affine) loop constructs and n-D arrays embedded into a single static assignment (SSA) substrate enables an unprecedented combination of SSA-based and polyhedral optimizations. We illustrate this by proposing and implementing two extra transformations: statement splitting and reduction parallelization. Our evaluation demonstrates that Polygeist outperforms on average both an LLVM IR-level optimizer (Polly) and a source-to-source state-of-the-art polyhedral compiler (Pluto) when exercised on the Polybench/C benchmark suite in sequential (2.53x vs 1.41x, 2.34x) and parallel mode (9.47x vs 3.26x, 7.54x) thanks to the new representation and transformations.
DOI:10.1109/PACT52795.2021.00011