Compiling SIMT Programs on Multi- and Many-Core Processors with Wide Vector Units: A Case Study with CUDA

Authors: Hancheng Wu (North Carolina State University), John Ravi (North Carolina State University), Michela Becchi (North Carolina State University)

Abstract: There has been an increasing interest in SIMT programming tools for multi- and manycore (co)processors with wide vector extensions. In this work, we study the effective implementation of a SIMT programming model (a subset of CUDA C) on Intel platforms with 512-bit vector extensions (hybrid MIMD/SIMD architectures). We first propose a set of compiler techniques to transform programs written using a SIMT programming model into code that leverages both the x86 cores and the vector units of a hybrid MIMD/SIMD architecture, thus providing programmability, high system utilization and portability. We then evaluate the proposed techniques on various hybrid systems using microbenchmarks and real-world applications. Finally, we point out the main challenges in supporting the SIMT model on hybrid systems.

Best Poster Finalist (BP): no

Poster: pdf
Poster summary: PDF
Reproducibility Description Appendix: PDF


Back to Poster Archive Listing