Three Papers Accepted at DS-RT'24

10 July 2024

We are thrilled to announce that three of our research papers have been accepted for presentation at the 28th International Symposium on Distributed Simulation and Real Time Applications (DS-RT 2024). This recognition underscores our ongoing commitment to advancing the field of distributed simulation and real-time applications.

  1. “PARSIR: Efficient Parallel Simulation with Speculative Execution on Heterogeneous Platforms” This paper introduces PARSIR, a framework designed to enhance the efficiency of parallel discrete-event simulations through speculative execution on heterogeneous platforms. PARSIS dynamically adapts to the optimal execution scheme in systems with multiòple NUMA nodes, thereby minimizing latency and maximizing throughput. The study demonstrates significant performance gains in various simulation scenarios, showcasing the potential of heterogeneous computing.

  2. “Sampling Policies for Near-Optimal Device Choice in Parallel Simulations on CPU/GPU Platforms” Focusing on the complex task of device selection in parallel discrete-event simulations, this paper proposes heuristic-based sampling policies to achieve near-optimal performance. The study employs a meta-simulation framework to evaluate these heuristics under various stochastic workloads, aiming to strike a balance between execution time and energy consumption. The results highlight the effectiveness of these policies in approximating optimal device choices.

  3. “Online Analytics with Local Operator Rebinding for Stream Processing” In this paper, we present a novel approach to enhancing the performance of stream processing systems, one of the domains of interest of the DOMAIN project, through online analytics and local operator rebinding. This method dynamically adjusts the binding of operators to processing elements based on real-time data analytics, optimizing the processing flow and resource utilization. The experimental results show substantial improvements in the throughput and latency of stream processing applications. More activities in this direction will be covered by the project in the upcoming months.

We look forward to presenting these papers at DS-RT 2024 and engaging with the global research community to share our findings and explore new avenues for collaboration. Stay tuned for more updates from the symposium.