Performance Implications of Transient Loop-Carried Data Dependences in Automatically Parallelized Loops

摘要

This is the author accepted manuscript. The final version is available from the Association for Computing Machinery via https://doi.org10.1145/2892208.2892214