Automated Mapping of Task-Based Programs onto Distributed and Heterogeneous Machines
Name
3581784.3607079.pdf
Size
1.45 MB
Format
Adobe PDF
Checksum (MD5)
1e76ee79ab00f39a38c509a729ac168c
Author(s)
S. F. X. Teixeira, Thiago
Henzinger, Alexandra
Yadav, Rohan
Aiken, Alex
Date Issued
November 12, 2023
Publisher
ACM|The International Conference for High Performance Computing, Networking, Storage and Analysis
Citation
S. F. X. Teixeira, Thiago, Henzinger, Alexandra, Yadav, Rohan and Aiken, Alex. 2023. "Automated Mapping of Task-Based Programs onto Distributed and Heterogeneous Machines."
Version
Final published version
Abstract
In a parallel and distributed application, a mapping is a selection of a processor for each computation or task and memories for the data collections that each task accesses. Finding high-performance mappings is challenging, particularly on heterogeneous hardware with multiple choices for processors and memories. We show that fast mappings are sensitive to the machine, application, and input. Porting to a new machine, modifying the application, or using a different input size may necessitate re-tuning the mapping to maintain the best possible performance.
We present AutoMap, a system that automatically tunes the mapping to the hardware used and finds fast mappings without user intervention or code modification. In contrast, hand-written mappings often require days of experimentation. AutoMap utilizes a novel constrained coordinate-wise descent search algorithm that balances the trade-off between running computations quickly and minimizing data movement. AutoMap discovers mappings up to 2.41× faster than custom, hand-written mappers.
MIT Department
Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory
Terms of Use
Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use.
Persistent DSpace Link
DOI of Published Version
https://doi.org/10.1145/3581784.3607079
