Method for generating automatic codes on multiple DSP (Digital Signal Processor) platform

Abstract

The invention relates to a method for generating automatic codes on a multiple DSP (Digital Signal Processor) platform. The method has the following main steps: 1) aiming at specific codes on the multiple DSP platform, carrying out classification and feature extraction, and summarizing a set of generic code frame template; 2) utilizing an XML (Extensive Makeup Language) description file to model and describe a specific algorithm process developed by a user, comprising a module list, module properties and the interconnected relationship between modules; 3) and finally, extracting necessary information in the XML description file, such as the module list, the module properties, and the interconnected relationship between the modules through an automatic code generator, and generating the final DSP source code based on the generic code frame template in the step 1).

Description

technical field [0001] The invention belongs to the embedded system field, specifically a technical method for automatically generating codes on a multi-DSP platform. Background technique [0002] With the advancement of computer technology, software production has also begun to develop in the direction of automation. Through special code automatic conversion software, the automatic conversion of the upper-level software model into the implementation code can greatly improve the efficiency of software production and liberate people from repetitive labor. Due to the high cost of human resources abroad, code generation technology has long been applied in software development, such as Matlab&Simulink from Mathworks, Labview from NI, Gedae from Gedae, etc. [0003] The multi-DSP platform is a hardware processing structure widely used in signal processing fields such as radar, communication and sonar. The need for high-density, high-throughput real-time signal processing tasks....

AI Technical Summary

Problems solved by technology

This development method has several obvious disadvantages: First, when the developed application program is large in scale and there are many DSP nodes (for example, dozens), the details of the interconnection and communication between the underlying hardware and system configuration need to be taken care of by the user. and intervention, both tedious and error-prone

The second is that this kind of development basically relies on human programming. The program has obvious personal style and a certain degree of randomness, lacks modularization and generalization, and the code developed by one programmer may be difficult for others to understand and accept; Engaged in a lot of repetitive work, the same application processing flow must rewrite the relevant code under different hardware mapping and allocation methods, which leads to a long system development cycle and a large consumption of manpower and material resources

Images