Swizzle Inventor
Authors:
Open Access
- 4 April 2019
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 65-78
Abstract
Utilizing memory and register bandwidth in modern architectures may require swizzles-non-trivial mappings of data and computations onto hardware resources-such as shuffles. We develop Swizzle Inventor to help programmers implement swizzle programs, by writing program sketches that omit swizzles and delegating their creation to an automatic synthesizer. Our synthesis algorithm scales to real-world programs, allowing us to invent new GPU kernels for stencil computations, matrix transposition, and a finite field multiplication algorithm (used in cryptographic applications). The synthesized 2D convolution and finite field multiplication kernels are on average 1.5-3.2x and 1.1-1.7x faster, respectively, than expert-optimized CUDA kernels.Keywords
Funding Information
- Intel Corporation
- Defense Advanced Research Projects Agency (CONIX CMU 1042741-394324 AM01)
- Nokia
- National Science Foundation (ACI OAC--1535191)
- Mozilla
- Qualcomm
This publication has 25 references indexed in Scilit:
- Communication-minimizing 2D convolution in GPU registers2013
- From relational verification to SIMD loop synthesis2013
- Polyhedral parallel code generation for CUDAACM Transactions on Architecture and Code Optimization, 2013
- High-performance code generation for stencil computations on GPU architectures2012
- A compiler framework for optimization of affine loop nests for gpgpus2008
- Parallel Programmability and the Chapel LanguageThe International Journal of High Performance Computing Applications, 2007
- Sketching stencils2007
- Programming by sketching for bit-streaming programs2005
- An affine partitioning algorithm to maximize parallelism and minimize communication1999
- Uniform techniques for loop optimization1991