Giovanni Agosta
Towards a High-Performance Modelica Compiler
Agosta, Giovanni; Baldino, Emanuele; Casella, Francesco; Cherubin, Stefano; Leva, Alberto; Terraneo, Federico
Authors
Emanuele Baldino
Francesco Casella
Stefano Cherubin S.Cherubin@napier.ac.uk
Lecturer
Alberto Leva
Federico Terraneo
Abstract
The use of Modelica as a modelling and simulation language is progressively replacing hand-tuned simulation code written in traditional imperative programming languages. This adoption is fuelled by the availability of libraries to target different application domains, as well as optimizations in Modelica implementations that allow to address larger problems. However, the effort required to extend existing Modelica implementations to support large scale models may not be economically sustainable by the Modelica community. To overcome this barrier, we believe a perspective change is required. Instead of developing, maintaining and optimizing a dedicated codebase, we propose to develop a Modelica implementation by relying on the LLVM state-of-the-art compiler framework. Although this approach will require a higher initial development effort, we believe that it will lead to significantly improved performance as well as lower overall cost. The paper discusses a possible roadmap for such a development, and presents a very early prototype implementation that exploits array structures by avoiding scalar expansion during the code generation process.
Citation
Agosta, G., Baldino, E., Casella, F., Cherubin, S., Leva, A., & Terraneo, F. (2019). Towards a High-Performance Modelica Compiler. In Proceedings of the 13th International Modelica Conference, Regensburg, Germany, March 4–6, 2019 (313-320). https://doi.org/10.3384/ecp19157313
Conference Name | The 13th International Modelica Conference |
---|---|
Conference Location | Regensburg, Germany |
Start Date | Mar 4, 2019 |
End Date | Mar 6, 2019 |
Publication Date | Feb 1, 2019 |
Deposit Date | Jun 19, 2021 |
Pages | 313-320 |
Series Title | Linköping Electronic Conference Proceedings |
Series Number | 157 |
Series ISSN | 1650-3740 |
Book Title | Proceedings of the 13th International Modelica Conference, Regensburg, Germany, March 4–6, 2019 |
ISBN | 9789176851227 |
DOI | https://doi.org/10.3384/ecp19157313 |
Keywords | Modelica Tools, Large-scale model simulation, Compilers, LLVM |
Public URL | http://researchrepository.napier.ac.uk/Output/2781731 |
You might also like
TAFFO: The compiler-based precision tuner
(2022)
Journal Article
Ahead-Of-Real-Time (ART): A Methodology for Static Reduction of Worst-Case Execution Time
(2022)
Conference Proceeding
Front Matter, Table of Contents, Preface, Conference Organization
(2022)
Conference Proceeding
Architecture-aware Precision Tuning with Multiple Number Representation Systems
(2021)
Conference Proceeding
Dynamic Precision Autotuning with TAFFO
(2020)
Journal Article