Research And Design Of RISC Instruction Set Automated Test System

With the increase in the size and complexity of microprocessor hardware,functional testing has become a bottleneck in the design process.The instruction set architecture divides the software and hardware of the entire computer system and is the core of the processor.The testing of the instruction set naturally becomes the most important part of microprocessor verification.As the mainstream technology of current processor architecture,RISC has become one of the major time-consuming phases of processor design.Therefore,the research on RISC instruction set automatic test system is of great significance to improve the efficiency of processor development and shorten the development cycle.Based on the in-depth study of RISC instruction set architecture and existing automated test methods,combined with the characteristics of self-developed processor cores,an RISC instruction set automated test system was established.The system mainly includes four parts: automatic generation of test plans,automatic generation of test cases,static compilation,and automatic generation of test reports.According to the needs of the RISC processor core and the characteristics of the instruction set,the equivalence class partition and boundary value analysis methods are used to extract the parameter constraints,and the test points are further planned.Finally,the instruction set test plan is automatically generated.The test plan automatic generation module not only improves the completeness of the test but also avoids many human factors.The test case automatic generation module analyzes the test cases,standardizes the test case composition structure,and divides a complete test case into multiple standardized test points.A single test point is summarized into three modules: data preparation,instructions to be tested,and results comparison.Sub-modules generate test points.Finally,multiple test points form a complete test case.The test case automatic generation module mainly stipulates test cases,reduces the workload of testers writing test codes,and avoids unnecessary errors caused by manual writing.After the test case is generated,it needs to compile the binary code that the processor can recognize.Therefore,the main function of the static compiling module is to convert the assembly code into a binary form.In addition,it also includes the function of code format check and the number of lines of code.The test report generation module reads the test platform output result file and the expected value file separately to complete the test report.This module can directly locate the wrong position and completely avoid the process of manually comparing the waveform data,which has a significant effect on improving the test efficiency.In this paper,the single-instruction test and the command correlation test are completed using the built automation test system.The efficiency of the automated test system and the completeness of test cases are analyzed.It takes at least a week for the traditional test method to complete a version of the logic test,and the automated test system can complete a version of the logic test in only 1 or 2 days.The completeness of test point coverage is 20%~30% higher than traditional test methods.The system can effectively improve the completeness of the test point under the premise of guaranteeing the test efficiency.