This is a GPU sharing and scheduling management platform specifically designed for heterogeneous computing environments, aiming to maximize GPU utilization. It provides flexible, reliable, on-demand and elastic multi-heterogeneous GPU virtualization, scheduling and management capabilities, supporting various hardware and virtualization technologies from mainstream manufacturers such as NVIDIA and Ascend, and is suitable for high-performance computing scenarios such as deep learning, data processing and scientific computing.