CN-121996488-A - Automatic test system, method, related equipment and program product

CN121996488ACN 121996488 ACN121996488 ACN 121996488ACN-121996488-A

Abstract

The embodiment of the invention provides an automatic test system, an automatic test method, related equipment and a program product, wherein the automatic test system comprises a test scheduling module, the test scheduling module is used for determining a current test case to be executed in a to-be-tested queue, marking the current test case when judging that the current residual resources in the current test environment cannot support the execution of the current test case, updating the next test case into the current test case, sorting the test cases based on the resource requirements of the test cases when judging that the test cases are marked, and determining the current test case to be executed when the current residual resources are increased, wherein the test case in the to-be-tested queue is the test case of a heterogeneous accelerator card of the current test which is matched with the current test environment. The automatic test system provided by the embodiment of the invention can improve the resource utilization rate of the heterogeneous accelerator card for executing the test case.

Inventors

JI XIAOXU

Assignees

海光信息技术股份有限公司

Dates

Publication Date: 20260508
Application Date: 20260128

Claims (16)

1. An automatic test system is characterized by being applied to heterogeneous accelerator cards and comprising a test scheduling module; The test scheduling module is used for determining current test cases to be executed in a to-be-tested queue, marking the current test cases when judging that current residual resources cannot support the execution of the current test cases in a current test environment of an automatic test system aiming at the current test cases, updating the next test cases to be executed in the to-be-tested queue into the current test cases, sorting the test cases based on resource requirements of the test cases when judging that the test cases in the to-be-tested queue are marked, and determining the current test cases to be executed in the test cases after sorting when the current residual resources are increased, wherein the test cases in the to-be-tested queue are test cases of heterogeneous accelerator cards which are matched with the current test environment and are used for testing currently.
2. The automated test system of claim 1, wherein the test scheduling module comprises a test case management module; the test case management module is used for acquiring all test cases supported by the automatic test system, establishing a mapping table according to the supporting relation between each test case and the heterogeneous accelerator card as well as the test environment, screening the test cases from the mapping table based on the heterogeneous accelerator card currently tested and the current test environment, and forming a queue to be tested based on the screened test cases.
3. The automated test system of claim 2, further comprising a test environment construction module for constructing a base test environment of the test environment to provide to the test case management module to form the current test environment; The test environment construction module is used for constructing a basic test environment of the test environment, and comprises: acquiring a compatibility check basis of a local test environment of the automatic test system; acquiring heterogeneous accelerator cards of different types included in the automatic test system; Detecting whether a local test environment of the automatic test system supports testing of the heterogeneous accelerator card or not according to the compatibility check basis of each model of heterogeneous accelerator card, if so, taking the local test environment as a basic test environment, and if not, acquiring corresponding matched test environment construction information according to the model of the heterogeneous accelerator card, and constructing a basic test environment based on the test environment construction information and the local test environment.
4. The automated test system of claim 3, wherein the test environment build information comprises drivers for heterogeneous accelerator cards, kits, and dependent software packages for test environments corresponding to heterogeneous accelerator cards.
5. The automatic test system of claim 3 wherein after the forming of the queue to be tested based on the screened test cases, the test case management module is further configured to initialize a resource pool based on the heterogeneous accelerator card currently being tested and the allocable resources in the basic test environment, form a current test environment for the heterogeneous accelerator card currently being tested, and use the resources in the initialized resource pool as the resources used by the test cases of the queue to be tested.
6. The automated test system of claim 1, wherein the test scheduling module comprises a performance information acquisition module and a fault diagnosis and restoration module; the performance information acquisition module is used for acquiring the performance state information of the current test environment and sending the performance state information of the current test environment to the fault diagnosis and recovery module in the process of executing the current test case by the test scheduling module; The fault diagnosis and recovery module is used for judging whether the current test environment is abnormal or not based on the performance state information, and processing the abnormality when the current test environment is abnormal.
7. The automated test system of claim 6, wherein after processing the anomaly, the fault diagnosis and restoration module is further configured to detect anomaly post-processing status information for the current test environment, determine whether the current test environment supports continuation of a subsequent test based on the anomaly post-processing status information, and if not, end the test and output a warning signal.
8. The automated test system of claim 7, wherein the performance state information of the current test environment comprises device performance state information, operating system performance state information, and heterogeneous accelerator card performance state information, the anomalies comprising drive issues and/or cross-chip connection issues; The method for processing the corresponding driving problem comprises reloading the driving; the corresponding method for processing the cross chip connection problem is to rebuild chip connection.
9. The automated test system of claim 6, further comprising a test result summary module; the test result summarizing module is used for collecting test results obtained by executing the current test cases in the current test environment; The performance information acquisition module is also used for sending the performance state information to the test result summarizing module; the fault diagnosis and recovery module is also used for sending the state information after the exception processing to the test result summarizing module; and the test result summarizing module is also used for summarizing and outputting the performance state information, the state information after exception processing and the test result.
10. The automated test system of claim 9, wherein the test scheduling module is further configured to, after the test result is obtained, release resources used by the test case from which the test result was obtained to the resource pool and increase the current remaining resources.
11. An automatic test method applied to an automatic test system as claimed in any one of claims 1 to 10, comprising: Determining a current test case to be executed in a queue to be tested; Aiming at the current test case, when judging that the current test case can not be supported by current residual resources in the current test environment of an automatic test system, marking the current test case, updating the next test case to be executed in the to-be-tested queue into the current test case, and when judging that the test cases in the to-be-tested queue are marked, sorting the test cases based on the resource requirements of the test cases, and when the current residual resources are increased, determining the current test case to be executed in the test cases after sorting, wherein the test cases in the to-be-tested queue are the test cases of the heterogeneous accelerator card for the current test, which are matched with the current test environment.
12. The automatic test method of claim 11 wherein in executing the current test case, the automatic test method further comprises: Based on the performance state information, judging whether the current test environment is abnormal or not; If so, the exception is handled.
13. The automatic test method of claim 12 further comprising, after the step of handling the exception, detecting exception-handling status information for the current test environment, determining whether the current test environment after handling the exception supports continued subsequent testing based on the exception-handling status information, ending the test if not, and outputting a warning signal.
14. A computer program product comprising a computer program, characterized in that the computer program, when executed, implements the automatic test method according to any of claims 11-13.
15. An electronic device comprising a memory, the memory storing a program, and a processor, the processor invoking the program stored in the memory to perform the automatic test method of any of claims 11-13.
16. A storage medium storing a program which, when executed, implements the automatic test method of any one of claims 11-13.

Description

Automatic test system, method, related equipment and program product Technical Field The embodiment of the invention relates to the field of chip testing, in particular to an automatic testing system, an automatic testing method, related equipment and a program product. Background With the rapid development of artificial intelligence, high-Performance Computing (HPC) and large-scale deep learning models, heterogeneous acceleration computing architecture has become a core solution for improving the computational intensity. The heterogeneous acceleration computing card is a core hardware component of a heterogeneous acceleration computing architecture, and is widely applied to scenes such as a data center, scientific simulation, AI (ARTIFICIAL INTELLIGENCE ) training reasoning and the like by virtue of the parallel computing capability of the heterogeneous acceleration computing card. However, heterogeneous accelerator cards themselves are iteratively accelerated, the driven versions of heterogeneous accelerator cards are fragmented, and software tools associated with testing of heterogeneous accelerator cards are frequently updated, resulting in testing of heterogeneous accelerator cards facing multiple systematic challenges. In the related art, when testing heterogeneous accelerator cards, there is a problem of low resource utilization. Therefore, a test system capable of improving the resource utilization of heterogeneous accelerator cards executing test cases is needed. Disclosure of Invention The embodiment of the invention provides an automatic test system, an automatic test method, related equipment and a program product, which are used for improving the resource utilization rate of heterogeneous accelerator cards for executing test cases. In order to achieve the above purpose, the embodiment of the present invention provides the following technical solutions. In a first aspect, an embodiment of the present invention provides an automatic test system applied to heterogeneous accelerator cards, including a test scheduling module; The test scheduling module is used for determining current test cases to be executed in a to-be-tested queue, marking the current test cases when judging that current residual resources cannot support the execution of the current test cases in a current test environment of an automatic test system aiming at the current test cases, updating the next test cases to be executed in the to-be-tested queue into the current test cases, sorting the test cases based on resource requirements of the test cases when judging that the test cases in the to-be-tested queue are marked, and determining the current test cases to be executed in the test cases after sorting when the current residual resources are increased, wherein the test cases in the to-be-tested queue are test cases of heterogeneous accelerator cards which are matched with the current test environment and are used for testing currently. In a second aspect, an embodiment of the present invention provides an automatic testing method, applied to the automatic testing system according to the first aspect, including: Determining a current test case to be executed in a queue to be tested; Aiming at the current test case, when judging that the current test case can not be supported by the current residual resources in the current test environment of the automatic test system, marking the current test case, and updating the next test case to be executed in the to-be-tested queue into the current test case; and when judging that the test cases in the to-be-tested queue are marked, sequencing the test cases based on the resource requirements of the test cases, and when the current residual resources are increased, determining the current test case to be executed in the test cases after sequencing, wherein the test cases in the to-be-tested queue are the test cases of the heterogeneous accelerator card which is adapted to the current test environment and is currently tested. In a third aspect, embodiments of the present invention provide a computer program product comprising a computer program which, when executed, implements the automatic test method as described in the second aspect. In a fourth aspect, an embodiment of the present invention provides an electronic device, including a memory, where a program is stored, and a processor, where the processor invokes the program stored in the memory, to perform the automatic test method according to the second aspect. In a fifth aspect, an embodiment of the present invention provides a storage medium storing a program that when executed implements the automatic test method according to the second aspect. The embodiment of the invention provides an automatic test system, which is applied to a heterogeneous acceleration card and comprises a test scheduling module, wherein the test scheduling module is used for determining current test cases to be executed in a to-be-tested queue, marking the cu