CXL3.0 physical layer verification method, platform, chip and medium

Abstract

The invention relates to a CXL3.0 physical layer verification method, a platform, a chip and a medium. The method comprises the following steps: acquiring an RTL code written by a designer according to a design specification; generating verification excitation; the verification excitation is input into the RTL code through the verification platform to normally drive the design to be tested, and coverage rate information under the verification excitation is collected; and judging whether a coverage rate requirement is met or not, if the coverage rate requirement is met, ending the verification process, and if the coverage rate requirement is not met, optimizing verification excitation by utilizing a genetic firefly algorithm, and carrying out next verification again. Compared with the prior art, the method has the advantages of improving verification excitation convergence efficiency and the like.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of chip verification, in particular to an improved verification excitation UVM-based CXL3.0 physical layer verification method, platform, chip and medium. BACKGROUND

[0002] With the development of microelectronic technology, the complexity and functionality of digital chip design are increasingly complex and diverse, and the difficulty of chip function verification is also greatly increased. As a core way to judge whether the expected function of the chip is realized, chip verification directly affects the success rate of tape-out and has become a key link in the front-end process of the chip. Compute Express Link (CXL) 3.0 is a new generation of heterogeneous computing interconnection protocol, which runs on the PCI Express (PCIe) 6.0 physical layer and aims to solve emerging data-centric workloads by implementing consistency and memory semantics. In the AI era, the CXL technology has a very broad prospect. The CXL3.0 protocol adds new functions on the basis of the CXL2.0 protocol, and the traditional verification platform for the CXL2.0 protocol has low reusability and cannot meet the new verification requirements. Therefore, a new verification platform and reuse method need to be designed to meet the challenge. The UVM (Universal Verification Methodology) verification methodology, as the verification methodology with the highest usage rate in the industry, has helped to improve the efficiency of chip verification.

[0003] However, the traditional verification platform and the random generation excitation method have the problems of insufficient reusability and low convergence efficiency, which affects the overall verification coverage. SUMMARY

[0004] The purpose of the application is to provide an improved verification excitation UVM-based CXL3.0 physical layer verification method, platform, chip and medium.

[0005] The purpose of the application can be achieved by the following technical solutions:

[0006] An improved verification excitation UVM-based CXL3.0 physical layer verification method, comprising the following steps:

[0007] S1, obtaining the RTL code written by the designer according to the design specification;

[0008] S2, generating verification excitation;

[0009] S3, inputting the verification excitation into the RTL code through the verification platform to normally drive the design under test, and collecting the coverage information under the verification excitation;

[0010] S4, judging whether the coverage requirement is reached, if the coverage requirement is reached, the verification process is ended, if the coverage requirement is not reached, the genetic firefly algorithm is used to optimize the verification incentive, and the step S3 is returned to perform the next verification.

[0011] In the step S4, the collected coverage information is used as the input data of the fitness function of the genetic firefly algorithm, and the genetic firefly algorithm is used to generate the new required verification incentive according to the fitness function.

[0012] In the step S4, the verification incentive is binary coded and used as the individual to be optimized in the genetic firefly algorithm.

[0013] The genetic firefly algorithm includes the following steps:

[0014] Step 1) initializing the related parameters of the genetic firefly algorithm: the parameters of the genetic algorithm are initialized, the next generation population generated by the genetic algorithm is used as the first generation population of the firefly algorithm, and the initial parameters of the firefly algorithm are set;

[0015] Step 2) calculating the individual fitness of the genetic firefly algorithm based on the coverage information;

[0016] Step 3) performing the selection, crossover and mutation operations in the genetic algorithm to generate the first population with better overall fitness;

[0017] Step 4) using the firefly algorithm to optimize the population generated by the genetic algorithm again, updating the individual position through the calculation of the fluorescence brightness and the attraction, and generating the second population;

[0018] Step 5) merging the first and second populations;

[0019] Step 6) judging whether the iteration termination condition is met, if the iteration termination condition is met, the optimal solution and the optimal individual are output, otherwise, the step 2) is returned to perform the next iteration.

[0020] The initialization of the parameters of the genetic algorithm is specifically: the population number is set to 30, the evolution iteration number is set to 400, the crossover probability is set to 0.8, and the mutation probability is set to 0.2.

[0021] The initial parameters of the firefly algorithm are specifically: the light absorption coefficient is set to 1, and the attraction coefficient is set to 2.

[0022] The method further includes an optimized register coverage method, specifically: a python script is used to grab the register data in the verification process of the step S3, a register database is created through a csv file, and the coverage data is extracted from a txt file to automatically perform the comparison.

[0023] A verification platform, which is a reusable verification platform built by a verification personnel according to a verification specification based on a UVM verification methodology, is used to realize the improved verification excitation based on UVM CXL3.0 physical layer verification method.

[0024] A chip, which comprises a verification module, is used to realize the improved verification excitation based on UVM CXL3.0 physical layer verification method.

[0025] A storage medium, which stores a program, is used to realize the improved verification excitation based on UVM CXL3.0 physical layer verification method.

[0026] Compared with the prior art, the present application has the following beneficial effects:

[0027] (1) The genetic firefly hybrid algorithm is added in the verification excitation generation method, and compared with the traditional random excitation method, the verification excitation convergence efficiency is improved by 61.9%.

[0028] (2) The optimized register coverage method is added in the verification method, the register data in the verification process is grabbed by using a python script, a register database is created through a csv file, and then the coverage data is extracted from a txt file, so that automatic comparison is realized, and the coverage is improved.

[0029] (3) The reusable verification platform for verifying the CXL protocol physical layer comprises top-level environment, monitor, driver, sequencer, detector and other reusable and portable verification components, so that the verification efficiency of the CXL IP can be greatly improved, the optimized function coverage and code coverage model are adopted, the completeness of the CXL protocol verification is improved, and the risk and cost of the chip implementation are reduced. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 It is a method flowchart of the present application;

[0031] Figure 2 It is a genetic firefly algorithm flowchart of the present application;

[0032] Figure 3 It is a verification excitation binary encoding schematic diagram of the present application;

[0033] Figure 4 It is an algorithm convergence performance curve diagram in an embodiment;

[0034] Figure 5 It is a code coverage result diagram in an embodiment;

[0035] Figure 6A functional coverage result graph in one embodiment.

[0036] Figure 7 A verification platform system block diagram of the present application; DETAILED DESCRIPTION

[0037] The present application will be described in detail below with reference to the accompanying drawings and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present application, and detailed implementation modes and specific operation processes are given, but the protection scope of the present application is not limited to the following embodiments.

[0038] The present embodiment provides a UVM-based CXL3.0 physical layer verification method for improving verification stimuli, as shown in the following steps. Figure 1

[0039] S1, first, detailed design specifications and verification specifications are formulated according to functional requirement specifications, and then RTL codes written by designers according to the design specifications are obtained.

[0040] S2, verification stimuli are generated.

[0041] S3, the verification stimuli are input into the RTL codes to normally drive the design under test through the verification platform, and coverage information under the verification stimuli is collected;

[0042] S4, the collected coverage is compared with the set coverage target to determine whether the coverage requirement is met, if the coverage requirement is met, the verification process is ended, and if the coverage requirement is not met, the genetic firefly algorithm is used to optimize the verification stimuli, and the next verification is returned to step S3.

[0043] The verification of the present application introduces the genetic firefly algorithm to improve the verification process driven by the coverage, improves the verification convergence speed, and realizes the functional verification of the CXL3.0 protocol physical layer.

[0044] The collected coverage information is used as input data of the fitness function of the genetic firefly algorithm, and new required verification stimuli are generated by the genetic firefly algorithm according to the fitness function after each test.

[0045] As shown in the following figure, Figure 2 The genetic firefly algorithm includes the following steps:

[0046] Step 1) initialize the related parameters of the genetic firefly algorithm: initialize the parameters of the genetic algorithm, set the population number to 30, set the evolution iteration number to 400, set the crossover probability to 0.8, set the mutation probability to 0.2; and the next generation population generated by the genetic algorithm is used as the first generation population of the firefly algorithm, the initial parameters of the firefly algorithm are set, the light absorption coefficient is set to 1, and the attraction coefficient is set to 2;​

[0047] Step 2) calculating the fitness of the individual of the genetic firefly algorithm based on the coverage information;

[0048] Step 3) performing the selection, crossover and mutation operations in the genetic algorithm to generate a first population with better overall fitness;

[0049] Step 4) using the firefly algorithm to optimize the population generated by the genetic algorithm again, updating the position of the individual through the calculation of the fluorescence brightness and attractiveness to generate a second population;

[0050] Step 5) merging the first and second populations;

[0051] Step 6) judging whether the iteration termination condition is met, outputting the optimal solution and the optimal individual when the iteration termination condition is met, otherwise, returning to step 2) for the next iteration.

[0052] The genetic algorithm is used to generate test stimuli more targeted, which can make the functional coverage converge faster in the later stage, and a smaller number of stimuli can achieve the verification goal.

[0053] As shown in Figure 3 , the verification stimuli are binary coded and used as the individual to be optimized in the genetic firefly algorithm.

[0054] In this embodiment, the verification method further includes an optimized register coverage method, specifically: a python script is used to capture the register data in the verification process of step S3, a register database is created through a csv file, and the coverage data is extracted from a txt file to automatically compare and improve the coverage.

[0055] Figure 4 The convergence performance curve of the method of the application is shown, wherein GA represents the genetic algorithm, FA represents the firefly algorithm, and FAGA represents the genetic firefly algorithm of the application. It can be seen that the number of iterations required for the method of the application to converge is greatly reduced compared with the genetic algorithm and the firefly algorithm alone, greatly improving the convergence efficiency of generating verification stimuli.

[0056] Figure 5 and Figure 6 The code coverage result and the functional coverage result of the verification method of the application are shown respectively. It can be seen that the test result of the application is good, which can improve the completeness of the CXL protocol verification and reduce the risk and cost of tape-out.

[0057] The embodiment also provides a verification platform, which is a reusable verification platform built by a verification personnel according to a verification specification and based on a UVM verification methodology, and is used to implement the improved UVM-based CXL3.0 physical layer verification method of verification stimulus. Figure 7 As shown in FIG. 13, the verification platform includes a top-level environment, a monitor, a driver, a sequencer, a detector, and other reusable and portable verification components.

[0058] The embodiment provides a chip, which includes a verification module and is used to implement the improved UVM-based CXL3.0 physical layer verification method of verification stimulus.

[0059] If the above functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the parts of the present application that essentially contribute to the prior art can be embodied in the form of software products. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0060] The preferred embodiments of the present application are described in detail above. It should be understood that those skilled in the art can make many modifications and variations without creative labor based on the concept of the present application. Therefore, any technical solution that can be obtained by logical analysis, reasoning, or limited experiments by those skilled in the art based on the concept of the present application and on the basis of the prior art should be within the protection scope determined by the claims.

Claims

1. An improved UVM-based CXL3.0 physical layer verification method of verification stimulus, characterized in that, The method comprises the following steps: S1, obtaining RTL code written by a designer according to a design specification; S2, generating a verification stimulus; S3, inputting the verification stimulus into the RTL code through a verification platform to normally drive a design under test, and collecting coverage information under the verification stimulus; S4, judging whether a coverage requirement is met, if the coverage requirement is met, the verification process is ended, and if the coverage requirement is not met, a genetic firefly algorithm is used to optimize the verification stimulus, and the step S3 is returned for next verification.

2. The UVM-based CXL3.0 physical layer verification method with improved verification stimulus according to claim 1, wherein, In the step S4, the collected coverage information is used as input data of a fitness function of the genetic firefly algorithm, and a new required verification stimulus is generated by using the genetic firefly algorithm according to the fitness function.

3. The UVM-based CXL3.0 physical layer verification method with improved verification stimulus of claim 1, wherein, In the step S4, the verification stimulus is binary coded and used as an individual to be optimized in the genetic firefly algorithm.

4. The UVM-based CXL3.0 physical layer verification method with improved verification stimulus of claim 1, wherein, The genetic firefly algorithm comprises the following steps: Step 1), initializing related parameters of the genetic firefly algorithm, initializing parameters of a genetic algorithm, taking a next generation population generated by the genetic algorithm as a first generation population of the firefly algorithm, and setting initial parameters of the firefly algorithm; Step 2), calculating individual fitness of the genetic firefly algorithm based on the coverage information; Step 3), performing selection, crossover and mutation operations in the genetic algorithm to generate a first population with better overall fitness; Step 4), optimizing the population generated by the genetic algorithm again by using the firefly algorithm, updating individual positions by calculating fluorescence brightness and attraction, and generating a second population; Step 5), merging the first and second populations; Step 6), judging whether an iteration termination condition is met, outputting an optimal solution and an optimal individual when the iteration termination condition is met, and otherwise, returning to the step 2) for next iteration.

5. The UVM-based CXL3.0 physical layer verification method with improved verification stimulus of claim 4, wherein, The initializing of the parameters of the genetic algorithm specifically comprises: setting a population number to 30, setting an evolution iteration number to 400, setting a crossover probability to 0.8, and setting a mutation probability to 0.

2.

6. The UVM-based CXL3.0 physical layer verification method with improved verification stimulus of claim 4, wherein, The setting of the initial parameters of the firefly algorithm specifically comprises: setting a light absorption coefficient to 1 and setting an attraction coefficient to 2.

7. The UVM-based CXL3.0 physical layer verification method with improved verification stimulus of claim 1, wherein, The method further comprises an optimized register coverage method, specifically comprising: using a python script to capture register data in the verification process of the step S3, creating a register database through a csv file, extracting coverage data from a txt file, and automatically performing comparison.

8. A verification platform, characterized in that, The verification platform is a reusable verification platform built by a verification personnel according to a verification specification and based on a UVM verification methodology, and is used to implement the UVM-based CXL3.0 physical layer verification method for improving the verification stimulus.

9. A chip, characterized by The chip comprises a verification module, and the verification module is used to implement the UVM-based CXL3.0 physical layer verification method for improving the verification stimulus.

10. A storage medium having stored thereon a program, characterized by The program is executed to implement the UVM-based CXL3.0 physical layer verification method for improving the verification stimulus.

Images