A Vectorized Calculation Method for Solving Large-Scale Triangular Linear Equations System Oriented to GPDSP

Abstract

The invention discloses a GPDSP-oriented large-scale triangular linear equations solution vectorized calculation method, which is: the solution of the large-scale triangular linear equations is divided into irregular triangular matrix blocks and vector multiplication by means of block calculation and regularized matrix block and vector multiplication; run the operating system through the CPU core in GPDSP, responsible for the logical division of data and task scheduling, and calculate the irregular triangular matrix block and vector multiplication calculation with less load; through GPDSP The DSP core is responsible for regularized matrix division and vector multiplication, continuously obtains new computing tasks from the CPU core and reports the running results to the CPU core, and the CPU core realizes the coordination of computing tasks between DSP cores through dynamic scheduling. Dynamic load balancing. The invention is simple in principle and convenient in operation, can make full use of the general calculation of the CPU core in the GPDSP and the powerful vector processing array parallel calculation of the DSP core, and the high-bandwidth vector data loading capability, and significantly improves the memory access ratio of the DSP core calculation.

Description

technical field [0001] The present invention mainly relates to a general-purpose digital signal processor (General-Purpose Digital Signal Processor, referred to as GPDSP), in particular to a large-scale triangular linear equation solution vectorized calculation method suitable for GPDSP. Background technique [0002] Calculations for solving trigonometric linear equations are widely used in the fields of high-performance computing and scientific engineering computing. For example, the solution of large-scale dense linear equations is usually LU decomposition or QR decomposition of the coefficient matrix, and then transformed into triangular linear equations to solve, and finally the final linear equations are obtained by solving the triangular linear equations. High Performance Linpack (HPL) is the most important test benchmark of TOP500. HPL tests and evaluates the floating-point performance of high-performance computer systems by solving dense linear algebraic equations. A...

AI Technical Summary

Problems solved by technology

The advantage of the CPU core in GPDSP lies in control and management, and its computing power accounts for a very small proportion in GPDSP, while the vector processing array of the DSP core has powerful vectorized computing power and high-bandwidth vector data loading capability, which is the main component of GPDSP computing performance. Contributors, therefore, the above acceleration method for solving triangular linear equations using CPU and GPU is not suitable for GPDSP, and cannot make full use of the vector array storage and memory access mode of DSP core and the architectural features of vector processing array concurrent vector processing, and it is difficult to make full use of GPDSP The advantages of vector computing

Images