Texturizing PPCG: Supporting Texture Memory in a Polyhedral Compiler

Patwardhan, Abhishek A and Upadrasta, Ramakrishna (2016) Texturizing PPCG: Supporting Texture Memory in a Polyhedral Compiler. In: 23rd IEEE International Conference on High Performance Computing, Data, and Analytics, 19-22 December 2016, Hyderabad.

Preview

Text
texPPCG.pdf - Accepted Version
Download (127kB) | Preview

Abstract

In this paper, we discuss techniques to transform sequential programs to texture/surface memory optimized CUDA programs. We achieve this by using PPCG, an automatic paral- lelizing compiler based on the Polyhedral model. We implemented a static analysis in PPCG which validates the semantics of the texturized transformed program. Depending on the results of the analysis, our algorithm chooses to use texture and/or surface memory, and alters the Abstract Syntax Tree accordingly. We also modified the code-generation phase of PPCG to take care of various subtleties. We evaluated the texturization algorithm on the PolyBench (4.2.1 beta) benchmark and observed up to 1.6x speedup with a geometric mean of 1.103X. The title and at many places, the paper uses term Texture memory. But, the optimizations are for Texture and Surface memory.

[error in script]

IITH Creators: