Speaker
Description
Meeting the data processing requirements of next-generation high-energy physics collider experiments, such as the Circular Electron-Positron Collider (CEPC), poses a significant challenge for data acquisition systems, particularly in the real-time triggering, online selection, and flexible processing of enormous event rates. Conventional online computing architectures based on static pipelines exhibit limitations in flexibility, scalability, and the rapid deployment of offline algorithms.
This report presents a novel data acquisition and online computing architecture centered around a distributed in-memory cache pool. By establishing a globally shared cache pool, this architecture decouples front-end electronics readout from back-end high-performance online processing modules, enabling asynchronous communication. A core scheduling system manages the full lifecycle of online processing algorithms and enables dynamic resource allocation and isolation. This design ensures system real-time performance and stability while significantly enhancing flexibility for algorithm updates and module integration, effectively supporting the direct migration and application of complex offline algorithms in the online environment.
The core components of this architecture have been fully developed. Notably, this design has been successfully deployed and validated within the Large High Altitude Air Shower Observatory (LHAASO). Field tests confirm that the system fulfills real-time processing demands under extreme data throughput, demonstrating the architecture's effectiveness and engineering feasibility in addressing the future data challenges of large-scale facilities like CEPC.
