Scalable and Intelligent Traffic Modeling for Improved Urban Planning
- Name: Scalable and Intelligent Traffic Modeling for Improved Urban Planning
- EuroHPC machine used: LUMI
- Topic: Computer and Information Sciences
Overview of the project
The Scalable and Intelligent Traffic Modeling for Improved Urban Planning project focused on enhancing the performance, scalability, and operational robustness of the Ruth traffic simulator, a high‑performance tool designed for large‑scale route analysis and traffic modelling. The central aim of the work was to improve the efficiency of computationally intensive components by optimising how workloads are distributed across HPC resources.
To address these goals, the collaboration centered on benchmarking, performance tuning, and enabling more effective distributed execution. This included rethinking the simulator’s communication model and improving its ability to run efficiently on modern HPC clusters.
How did EPICURE support the project and what were the benefits of the support?
EPICURE support enabled the integration of an MPI-focused approach through the use of the ACE (Asynchronous Communication and Execution) library, which was specifically tailored for this purpose. With the integration of ACE, it is now possible to avoid long initialisation times, leverage MPI, and improve the portability and compatibility of the simulator across different HPC clusters. ACE is currently used to perform alternative route computation, which is one of the most computationally expensive components of the traffic simulator, and is implemented as an additional execution variant while the original TCP-based distribution version remains available. Integration with ACE improved both the performance and portability of the Ruth simulator and led to the presentation of the poster “Scalable Alternative Route Computation with ACE” at SC25.
Additional references
SCALABLE ALTERNATIVE ROUTE COMPUTATION WITH ACE: A C++17 LIBRARY FOR HPC TRAFFIC SIMULATIONS (https://sc25.supercomputing.org/proceedings/posters/poster_pages/post115.html)