Speaker
Description
The Circular Electron Positron Collider (CEPC) is a next-generation high-energy electron–positron collider. Its Reference Detector will generate large volumes of data, placing demanding requirements on the trigger and data acquisition system. A fully software-based trigger offers flexible deployment, straightforward upgrades, and extensible algorithm support, while introducing substantial computational demands.
This dissertation focuses on online data processing for the CEPC and investigates two key aspects. First, GPU-oriented track-based filtering algorithms are studied and evaluated in terms of physics performance, processing latency, and throughput. Second, a heterogeneous CPU–GPU online computing framework is developed to support time-slice data organization, heterogeneous memory management, and task scheduling, providing a unified and extensible execution environment for online algorithms.
This work provides an algorithmic and software foundation for building a flexible, high-throughput, and sustainable online data processing system for the CEPC.
