SWPy: Python Numerical Computing Library Optimization for Domestic Many-core Processors

The NumPy computational library, the most typical scientific computing library in Python, is generally limited to a single-node or multi-threaded CPU-only execution model. With the increasing size of datasets, the increasing complexity of programs, and their growing computational overhead, these resources are far beyond what a single CPU node can provide, and there is a growing need to solve these problems by utilizing more massive computational resources. In this paper, we port the NumPy numerical computation library for domestic heterogeneous multicore processors and make full use of the huge computational resources of sunway multicore processors to optimize the parallel acceleration of common key functions in the NumPy library, forming the heterogeneous multicore Python computation library - SWPy.