Large Scale Multi-GPU Based Parallel Traffic Simulation for Accelerated Traffic Assignment and Propagation (2406.08496v2)

Abstract: Traffic propagation simulation is crucial for urban planning, enabling congestion analysis, travel time estimation, and route optimization. Traditional micro-simulation frameworks are limited to main roads due to the complexity of urban mobility and large-scale data. We introduce the Large Scale Multi-GPU Parallel Computing based Regional Scale Traffic Simulation Framework (LPSim), a scalable tool that leverages GPU parallel computing to simulate extensive traffic networks with high fidelity and reduced computation time. LPSim performs millions of vehicle dynamics simulations simultaneously, outperforming CPU-based methods. It can complete simulations of 2.82 million trips in 6.28 minutes using a single GPU, and 9.01 million trips in 21.16 minutes on dual GPUs. LPSim is also tested on dual NVIDIA A100 GPUs, achieving simulations about 113 times faster than traditional CPU methods. This demonstrates its scalability and efficiency for large-scale applications, making LPSim a valuable resource for researchers and planners. Code: https://github.com/Xuan-1998/LPSim

• The paper demonstrates that LPSim leverages a multi-GPU architecture to simulate up to 24 million trips and achieve a speedup of approximately 113× over high-end CPU simulations.
• The paper introduces a novel graph partitioning strategy that minimizes inter-GPU communication and optimizes load balance in large-scale traffic simulations.
• The paper shows that optimized vectorized data management enhances memory efficiency and processing speed, enabling real-time urban traffic dynamic simulations.

Overview of Large Scale Multi-GPU Based Parallel Traffic Simulation Framework

The paper presents the Large Scale Multi-GPU Parallel Computing based Regional Scale Traffic Simulation Framework (LPSim), designed to enhance the efficiency of traffic simulation through the utilization of multi-GPU architecture. Addressing the limitations of traditional CPU-based simulations, LPSim leverages the parallel computation capabilities of GPUs to provide scalable, high-fidelity simulations of extensive urban traffic networks.

The authors developed LPSim to specifically tackle the computational challenges of simulating large-scale microscopic traffic scenarios. These challenges include managing a vast amount of spatiotemporal data and processing individual vehicle movements in fine detail. The framework is capable of simulating tens of millions of individual vehicle dynamics simultaneously, far exceeding the capacity of traditional simulation frameworks.

Key Contributions

1. Multi-GPU Architecture: The framework employs a multi-GPU architecture, distributing the computational load across multiple GPUs. This approach efficiently manages extensive graph data, ensuring load balance and reducing inter-GPU communication overhead. The flexibility of LPSim allows it to simulate up to 24 million trips concurrently, offering unprecedented scalability.
2. Graph Partitioning Strategy: LPSim utilizes a novel graph partitioning strategy tailored for multi-GPU environments. By minimizing communication overhead and ensuring efficient data distribution, the framework optimizes the performance of large-scale simulations. The paper provides a theoretical analysis of this graph partitioning problem, introducing both balanced and unbalanced partitioning methods to optimize computational efficiency.
3. Optimized Data Management: The framework employs vectorized data storage and access mechanisms. This allows for efficient allocation of memory resources within the GPU environment, enhancing data handling and processing speed. By using a device vector to store dynamic vehicle data, LPSim addresses the challenges associated with fixed-size arrays, further optimizing memory usage.
4. Numerical Results and Performance Benchmarking: The empirical results demonstrate the scalability and efficiency of LPSim. The simulation of 9.01 million trips on a Google Cloud instance with two NVIDIA V100 GPUs is completed in 21.16 minutes. Furthermore, using dual NVIDIA A100-PCIE-40GB GPUs, the framework achieves a speedup of approximately 113 times compared to similar simulations on high-end CPUs.

Implications and Future Directions

The implications of LPSim are significant for both theoretical research and practical applications in traffic management and urban planning. The framework's ability to simulate large-scale traffic dynamics in real-time enables more informed decision-making for congestion alleviation, infrastructure investment, and policy development. Moreover, LPSim’s open-source code invites further improvements and adaptations by the academic community, fostering advancements in multi-modal traffic simulation and smarter city planning.

Future work aims to expand the framework’s capability to simulate multimodal transportation systems, capturing the dynamics of various travel modes beyond vehicular traffic, such as bicycles and public transit. Integrating real-time data analytics and machine learning models could further refine LPSim’s predictive capabilities, supporting dynamic traffic management strategies.

In conclusion, LPSim represents a substantial advancement in the field of traffic simulation, providing a robust, scalable solution to the complex challenges posed by large-scale traffic network simulations. Its contribution to urban traffic modeling stands to facilitate more efficient and sustainable urban transport systems.