CHARMWORKS, INC. — Department of Energy SBIR Phase II: 02c

Description

High Performance Computing technology has advanced significantly, and advanced parallel computers are being built, with DOE leadership. Yet, the American engineering and manufacturing industry has not started leveraging parallel computing at a significant level, partly because of the challenging nature of modern simulation software. US competitiveness in manufacturing may suffer due to its ineffective utilization of parallel computers. The Charm++ parallel programming system, developed at a University with DOE and other federal funding, has demonstrated promise in handling complex dynamic parallel applications in computational science. It automates resource management, tolerates failures, and can manage power/thermal issues. The objective of this project is transform Charm++ into a robust and easy to use system, with new runtime capabilities targeted to industry applications. The Phase I effort focused on analyzing the parallel computing needs of manufacturing/engineering industry, and refactoring the Charm++ software to retain its unique and innovative features that are of importance to this industry, while streamlining the software. Many features were identified as unnecessary; often vestiges of past academic research, and carefully removed. Performance estimation and tuning was carried out for common platforms used in industry including Infiniband and cloud based clusters. New features were implemented to eliminate unexpected errors, and enhance ease of use. During Phase II, extensive innovations and improvements are being made to the Charm++ system to (a) improve ease of use for new Charm++ programmers, (b) increase robustness of the software, and (c) to extend the runtime capabilities to address industry needs. New capabilities will include improved accelerator support, Fortran support, further automation and auto-tuning of functions such as load balancing, resilience and power management. In addition, we will demonstrate the utility of Charm++ by developing a new application in collaboration with an ISV.Commercial Applications and Other Benefits: The SBIR project effort is aimed at producing a commercially viable product based on Charm++ that incorporates innovative research on object-oriented parallel programming with adaptive run-times. It will be the main commercial application to result directly from this effort. In addition, it is expected that libraries developed using Charm++ product will be an additional category of commercial applications. Some of these may come from Charm works, Inc., but it is expected that many, in future, will come from other companies. End-user applications are probably the largest cate- gory of commercial applications enabled by this project effort. This project aims to achieve a broad impact on the state of HPC adoption among manufacturing and engineering industry.