Using the C++ Sender Asynchronous Programming Model in the Implementation of Efficient Epidemiological Simulations

ABSTRACT This paper evaluates the C++ Sender Programming Model for heterogeneous parallel computing, using an Agent‐Based epidemiological simulator as a case study. Conventional simulations rely on proprietary frameworks like CUDA , creating vendor lock‐in . To assess the Sender Model as a portable alternative, we implement an SEIRS dengue simulator that assigns tasks to CPU or GPU . Asynchronous execution is 3.8 times faster than synchronous methods. Our analysis also identified bottlenecks such as GPU stalls from memory operations and stochastic workloads, informing future optimization efforts.