Fast and Flexible Compilation Techniques for Effective Speculative Polyhedral Parallelization

Juan Manuel Martinez Caamaño 1, 2, 3
1 CAMUS - Compilation pour les Architectures MUlti-coeurS
Inria Nancy - Grand Est, ICube - Laboratoire des sciences de l'ingénieur, de l'informatique et de l'imagerie
Abstract : In this thesis, we present our contributions to APOLLO: an automatic parallelization compiler that combines polyhedral optimization with Thread-Level-Speculation, to optimize dynamic codes on-the-fly. Thanks to an online profiling phase and a speculation model about the target's code behavior, Apollo is able to select an optimization and to generate code based on it. During optimized code execution, Apollo constantly verifies the validity of the speculation model. The main contribution of this thesis is a code generation mechanism that is able to instantiate any polyhedral transformation, at runtime, without incurring a major time-overhead. This mechanism is currently in use inside Apollo. We called it Code-Bones. It provides significant performance benefits when compared to other approaches.
Document type :
Theses
Complete list of metadatas

Cited literature [31 references]  Display  Hide  Download

https://hal.inria.fr/tel-01377758
Contributor : Philippe Clauss <>
Submitted on : Friday, October 7, 2016 - 3:19:57 PM
Last modification on : Saturday, October 27, 2018 - 1:23:50 AM
Long-term archiving on : Friday, February 3, 2017 - 6:51:40 PM

Identifiers

  • HAL Id : tel-01377758, version 1

Citation

Juan Manuel Martinez Caamaño. Fast and Flexible Compilation Techniques for Effective Speculative Polyhedral Parallelization. Computation and Language [cs.CL]. Université de Strasbourg, 2016. English. ⟨tel-01377758⟩

Share

Metrics

Record views

455

Files downloads

338