MATE: Unified Model for Communication-Tolerant Scientific Applications
MATE is a distributed programming model for developing communication-tolerant scientific applications. MATE employs a combination of mechanisms to reduce or hide the cost of network and intra-node data movement. MATE provides these benefits within a single unified model, as opposed to hybrid (e.g., MPI+X) approaches.
- Source Code: https://bitbucket.org/SergioMartin86/mate/src/master/
- User Manual: Mate User Manual
Publications
Peer-Reviewed Publications
- MATE, a Unified Model for Communication-Tolerant Scientific Applications.
- Toucan - A Translator for Communication Tolerant MPI Applications.
Talks and Reports
- "MATE, a Unified Model for Communication-Tolerant Scientific Applications"
- Doctoral Dissertation. Speaker: Sergio Martin
- University of California, San Diego
- San Diego, CA. December 2018
- Download PDF
- "The MATE Model: Rationale and Preliminary Results."
- Invited Talk. Speaker: Sergio Martin
- Argonne National Laboratory
- Argonne, IL. March 2018
- Download PDF
- "A Hybrid Model for Reducing the Cost of Communication in Large-Scale Applications"
- Thesis Proposal. Speaker: Sergio Martin
- University of California, San Diego
- San Diego, CA. October 2017
- Download PDF
- "Translation and Runtime System-based Techniques for Hiding the Cost of Communication"
- Technical Report Talk. Speaker: Sergio Martin
- Lawrence Berkeley National Laboratory
- Berkeley, CA. August 2017
- Download PDF
- Toucan - IPDPS 17 Talk
- Paper Presentation. Speaker: Sergio Martin
- 31th International Parallel and Distributed Processing Symposium (IPDPS'17)
- Orlando, FL. June 2017.
- Download PDF