Generating Performance Portable Code Using Rewrite Rules: from High-Level Functional Expressions to High-Performance OpenCL Code
- Submitting institution
-
University of Glasgow
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 11-04917
- Type
- E - Conference contribution
- DOI
-
10.1145/2784731.2784754
- Title of conference / published proceedings
- ICFP 2015 Proceedings of the 20th ACM SIGPLAN International Conference on Functional Programming
- First page
- 205
- Volume
- -
- Issue
- -
- ISSN
- -
- Open access status
- -
- Month of publication
- September
- Year of publication
- 2015
- URL
-
http://eprints.gla.ac.uk/146605/
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
3
- Research group(s)
-
-
- Citation count
- 30
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- ORIGINALITY: The paper addresses performance portability between parallel architectures like CPUs and GPUs. It applies an innovative optimisation approach based on rewrite rules. SIGNIFICANCE: Published in the premier functional programming conference, it has been the highest-cited and downloaded paper of ICFP 2015 in the ACM DL. It has formed the basis for follow-up funding from Google (£36K, Principal Engineer), Microsoft Research (£50K, Principal Research Manager), Huawei (£100K), and Facebook Research (£36K). RIGOUR: The paper introduces a formal data-parallel language and defines semantics-preserving rewrite rules that are experimentally validated to demonstrate performance portability across 3 different processors (multi-core CPUs and GPUs).
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -