Static versioning in the polyhedral model
Abstract
An approach is presented to enhancing the optimization process in a polyhedral compiler by introducing compile-time versioning, i.e., the production of several versions of optimized code under varying assumptions on its run-time parameters. We illustrate this process by enabling versioning in the polyhedral processor placement pass. We propose an efficient code generation method and validate that versioning can be useful in a polyhedral compiler by performing benchmarking on a small set of deep learning layers defined for dynamically-sized tensors.
- Inventors:
- Issue Date:
- Research Org.:
- Reservoir Labs Inc., San Diego, CA (United States)
- Sponsoring Org.:
- USDOE Office of Science (SC), Advanced Scientific Computing Research (ASCR)
- OSTI Identifier:
- 2221950
- Patent Number(s):
- 11693636
- Application Number:
- 17/526,611
- Assignee:
- Reservoir Labs, Inc. (San Diego, CA)
- DOE Contract Number:
- SC0017071; SC0018480; SC0019522
- Resource Type:
- Patent
- Resource Relation:
- Patent File Date: 11/15/2021
- Country of Publication:
- United States
- Language:
- English
Citation Formats
Meister, Benoit J., and Dattatri, Adithya. Static versioning in the polyhedral model. United States: N. p., 2023.
Web.
Meister, Benoit J., & Dattatri, Adithya. Static versioning in the polyhedral model. United States.
Meister, Benoit J., and Dattatri, Adithya. Tue .
"Static versioning in the polyhedral model". United States. https://www.osti.gov/servlets/purl/2221950.
@article{osti_2221950,
title = {Static versioning in the polyhedral model},
author = {Meister, Benoit J. and Dattatri, Adithya},
abstractNote = {An approach is presented to enhancing the optimization process in a polyhedral compiler by introducing compile-time versioning, i.e., the production of several versions of optimized code under varying assumptions on its run-time parameters. We illustrate this process by enabling versioning in the polyhedral processor placement pass. We propose an efficient code generation method and validate that versioning can be useful in a polyhedral compiler by performing benchmarking on a small set of deep learning layers defined for dynamically-sized tensors.},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = {2023},
month = {7}
}
Save to My Library
You must Sign In or Create an Account in order to save documents to your library.
Works referenced in this record:
Runtime multi-versioning and specialization inside a memoized speculative loop optimizer
conference, February 2020
- Lazcano, Raquel; Madroñal, Daniel; Juarez, Eduardo
- Proceedings of the 29th International Conference on Compiler Construction
Polyhedral Source-to-Source Compiler
conference, June 2016
- Adamski, Dominik; Jablonski, Grzegorz; Perek, Piotr
- 2016 MIXDES - 23rd International Conference Mixed Design of Integrated Circuits and Systems
Unified optimization of iterative analytical query processing
patent, February 2021
- Jeong, Chanho; Lee, Sung Yeol; Lee, Chungmin
- US Patent Document 10,922,315
CAnDL: a domain specific language for compiler analysis
conference, February 2018
- Ginsbach, Philip; Crawford, Lewis; O'Boyle, Michael F. P.
- Proceedings of the 27th International Conference on Compiler Construction
System and Method for Advanced Polyhedral Loop Transformations of Source Code in a Compiler
patent-application, March 2009
- Eichenberger, Alexandre E.; O'Brien, John K. P.; O'Brien, Kathryn M.
- US Patent Application 11/861449; 20090083724