METHOD OF DEBUGGING A PROCESSOR

• US 20200134248A1
• Filed: 12/20/2019
• Published: 04/30/2020
• Est. Priority Date: 04/10/2018
• Status: Active Grant

• Current Assignee: Advanced Micro Devices, Inc.
• Original Assignee: Advanced Micro Devices, Inc.
• Inventors: Schieve, Eric W.
• Granted Patent: US 11,093,676 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 2115/10 Processors
• CPC Class Codes: G06F 30/30 Circuit design
• CPC Class Codes: G06F 30/33 Design verification, e.g. functional simulation or model checking
• CPC Class Codes: G06F 30/3308 using simulation
• CPC Class Codes: G06F 30/3323 using formal methods, e.g. equivalence checking or property checking

Abstract

Methods for debugging a processor based on executing a randomly created and randomly executed executable on a fabricated processor. The executable may execute via startup firmware. By implementing randomization at multiple levels in the testing of the processor, coupled with highly specific test generation constraint rules, highly focused tests on a micro-architectural feature are implemented while at the same time applying a high degree of random permutation in the way it stresses that specific feature. This allows for the detection and diagnosis of errors and bugs in the processor that elude traditional testing methods. The processor Once the errors and bugs are detected and diagnosed, the processor can then be redesigned to no longer produce the anomalies. By eliminating the errors and bugs in the processor, a processor with improved computational efficiency and reliability can be fabricated.

Claims

1. A method for debugging a processor comprising:

generating a predetermined number of randomized subtests according to a respective template file for each of a plurality of test types;

creating a test library of respective test binaries for each of the plurality of test types;

executing one or more randomly selected subsets of test binaries from the test library, wherein each of the randomly selected subset of test binaries are executed on a number of physical threads of a processor according to a control data structure for each test binary, wherein the executing is performed via startup firmware; and

detecting one or more errors in executing the randomly selected subset of test binaries from the test library.

2. The method of claim 1, wherein the startup firmware executes without an operating system (OS).

3. The method of claim 1, wherein the executing requires no operating system (OS), hard drive, or input/output (I/O) device.

4. The method of claim 1, wherein the executing is performed without user intervention.

5. The method of claim 1, wherein the startup firmware is loaded from basic input output system (BIOS) read-only memory (ROM).

6. The method of claim 1, wherein the executing includes randomly selecting particular randomized subtests from each test binary of the subset of test binaries and running each of the selected particular randomized subtests.

7. The method of claim 1, wherein the randomly selected subsets are selected by creating a temporary pool of arbitrarily chosen subtests and randomly selecting subsets from the temporary pool.

8. The method of claim 1, wherein the executing includes randomly shifting physical code of the one or more randomly selected subtests by a variable number of bytes in the control data structure after each of the particular randomized subtests is run.

9. The method of claim 1, wherein the control data structure specifies a timeout for executing the one or more randomly selected subsets of test binaries.

10. The method of claim 1, wherein the respective template file of the particular test type includes a weighting file that specifies a branch prediction behavior within the template that is used to generate the predetermined number of randomized subtests.

11. A method for debugging a processor comprising:

generating a predetermined number of randomized subtests according to a respective template file for each of a plurality of test types;

compiling the predetermined number of randomized subtests into a test binary for each of the plurality of test types;

creating a test library using the test binaries for each of the plurality of test types;

executing one or more randomly selected subsets of test binaries from the test library, wherein each of the randomly selected subset of test binaries are executed on a number of physical threads of a processor according to a control data structure for each test binary, wherein the executing is performed via startup firmware;

detecting one or more errors in executing the randomly selected subset of test binaries from the test library; and

redesigning the processor based on the detected one or more errors.

12. The method of claim 11, wherein the startup firmware executes without an operating system (OS).

13. The method of claim 11, wherein the executing requires no operating system (OS), hard drive, or input/output (I/O) device.

14. The method of claim 11, wherein the executing is performed without user intervention.

15. The method of claim 11, wherein the startup firmware is loaded from basic input output system (BIOS) read-only memory (ROM).

16. The method of claim 11, wherein the executing includes randomly selecting particular randomized subtests from each test binary of the subset of test binaries and running each of the selected particular randomized subtests.

17. The method of claim 11, wherein the randomly selected subsets are selected by creating a temporary pool of arbitrarily chosen subtests and randomly selecting subsets from the temporary pool.

18. The method of claim 11, wherein the executing includes randomly shifting physical code of the one or more randomly selected subtests by a variable number of bytes in the control data structure after each of the particular randomized subtests is run.

19. The method of claim 11, wherein the control data structure specifies a timeout for executing the one or more randomly selected subsets of test binaries.

20. The method of claim 11, wherein the respective template file of the particular test type includes a weighting file that specifies a branch prediction behavior within the template that is used to generate the predetermined number of randomized subtests.