Abstract
This chapter focuses on code generation from parallelization techniques. It briefly recalls the well mastered techniques based on linear transformations of polytopes and their scanning with loop nests. To complete the code generation, communications between virtual processors have to be synthesized from the dependence analysis; this can be achieved by applying the same linear transformation as the one going from the initial iteration space to the space-time domain. However, such a translation can present some insufficiencies and the dependence modeling must capture the whole information that is needed to generate communication code such as broadcast instructions, in order to perform it with the intrinsic efficiency of a data-parallel programming model for example. This enhances the interest of defining an appropriate dependence modeling and of relevant parallel code generation techniques from systems of affine recurrence equations. The chapter shows that it is able to generate an efficient communication code, synthesizing either local memory management or point to point communication, or broadcast. However, it is important to note that broadcasts are detected at the code generation level, by using basic linear algebraic tests.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.