METHOD AND APPARATUS FOR EFFICIENT STATISTICAL PROFILING OF VIDEO GAME AND SIMULATION SOFTWARE

• US 20100083234A1
• Filed: 07/31/2009
• Published: 04/01/2010
• Est. Priority Date: 09/30/2008
• Status: Active Grant

• Current Assignee: Nintendo Company Limited
• Original Assignee: Nintendo of America, Inc. (Nintendo Company Limited)
• Inventors: RABIN, Steve
• Granted Patent: US 9,495,279 B2
• Time in Patent Office: Days
• US Class Current: 717/127
• CPC Class Codes: G06F 11/3466 Performance evaluation by tracing or monitoring
• CPC Class Codes: G06F 11/3636 by tracing the execution of the program
• CPC Class Codes: G06F 2201/88 Monitoring involving counting

Abstract

Efficient statistical profiling in embedded computing devices, such as video games, uses a hybrid random distribution of sampling points for more accurate reconstruction of executing code. Transmission of only function start addresses and corresponding representation of the call graph data reduces the memory overhead and increases communication speed.

Claims

1. A system for monitoring and testing executing software of the type that generates moving image frames on a display device at a predetermined refresh rate comprising;

a processor that executes code to provide a display on said display device;

a statistical sampling profiler, which performs statistical sampling of the code being executed to provide profiling data, wherein at least the first sampling point in each frame has a random temporal offset relative to the refresh rate; and

a profile analyzer unit that receives the profiling data from the statistical sampling profiler and provides information about the dynamic operation of the executing code.

2. The system of claim 1, wherein for each frame, the corresponding random offset has a range of [0, n], wherein n is the time between sampling points in the frame.

3. The system of claim 2, wherein the range for the random tempered offset is chosen based on the offset values of neighboring previous frames, so that the sampling distribution varies significantly from frame to frame.

4. The system of claim 1, wherein the code provides real-time video game play or a real-time simulation.

5. A system for monitoring and testing software being executed in an embedded computing device, comprising;

a first computer processor unit, which executes commands based on a computer code comprising a set of functions;

a second computer unit connected to the first computer processor unit and storing said computer code, with each of said functions having an address in the memory of said second computer unit;

a display device which receives output data from the first computer processor unit and displays information related to said data at a predetermined frame rate;

a statistical sampling profiler unit, which is connected to the first computer processor unit and performs sampling of the computer code being executed, wherein the sampling occurs during each frame period, and wherein a list of the starting addresses of all the functions of the computer code being executed is transmitted from the second computer unit side to the first computer processor unit before the sampling begins, and wherein during sampling, each of the entries of the sample call stack resulting from the sampling is transformed to represent the starting address of the function that was sampled; and

a profile analyzer unit that receives the data from the statistical sampling profiler unit and provides information about the dynamic operation of the computer software.

6. The system of claim 5, wherein the starting address transformation for each entry is performed in log N steps of binary search of the list of the starting addresses, where N is the number of addresses in the list.

7. The system of claim 5, wherein the starting address transformation for each entry is performed using a hash table.

8. The system of claim 5, wherein the sample call stack comprising the transformed starting addresses of the functions is transformed into a “

first child-next sibling”

binary tree, resulting in a horizontal array of the frequency counts of each node of the tree.

9. The system of claim 5, wherein individual sample data is preserved by only storing a pointer to the leaf node in the call graph tree for each sample taken.

10. The system of claim 5, wherein the first computer processor unit is a video game console, and the second computer unit is a personal computer (PC).

11. A system for monitoring and testing software being executed in an embedded computing device, comprising;

a first computer processor unit, which executes commands based on a computer code comprising a set of functions;

a second computer unit connected to the first computer processor unit and storing said computer code, with each of said functions having an address in the memory of said second computer unit;

a display device which receives output data from the first computer processor unit and displays information related to said data at a predetermined frame rate;

a statistical sampling profiler unit, which is connected to the first computer processor unit and performs sampling of the computer code being executed, wherein the sampling occurs during each frame period, and wherein a list of the starting addresses of all the functions of the computer code being executed is transmitted from the second computer unit side to the first computer processor unit before the sampling begins, and wherein raw sampling data is collected and stored and at the conclusion of the sampling, it is transformed to comprise the starting addresses of the sampled functions of the executed program; and

a profile analyzer unit that receives the data from the statistical sampling profiler unit and provides information about the dynamic operation of the computer software.

12. The system of claim 11, wherein the first computer processor unit is a video game console, and the second computer unit is a personal computer (PC).

13. A method for monitoring and testing software, being executed by a computing device that provides display information to a display device at a predetermined periodic frame rate, said method comprising;

statistically sampling the computer code being executed, beginning at a sampling point having a random temporal offset relative to said frame rate to provide profiling data; and

analyzing the profiling data from the sampling to provide information about the computer software.

14. The method of claim 13, wherein each of the frames has a different range for its corresponding random offset value, with the biggest range being [0, n], wherein n is the time between sampling points in a frame.

15. The method of claim 14, wherein the range for the random offset of a frame is chosen based on the offset values of neighboring previous frames, so that the sampling distribution varies significantly from frame to frame.

16. The method of claim 13, wherein the embedded computing device is a video game console.

17. A method for monitoring and testing software code comprising a set of functions, said software being executed in an embedded computing device connected to a computer unit which stores said computer software, with each of said functions having an address in the memory of said computer unit, wherein the embedded computing device provides display information to a display device at a predetermined periodic frame rate, comprising;

statistically sampling the computer code being executed, wherein the sampling occurs during each frame period, and wherein a list of the starting addresses of all the functions of the computer code being executed is transmitted from the computer unit side to the embedded computing device before the sampling begins and wherein during sampling, each of the entries of the sample call stack resulting from the sampling is transformed to represent the starting address of the function that was sampled; and

analyzing the data from the sampling to provide information about the computer software.

18. The method of claim 17, wherein the starting address transformation for each entry is performed in log N steps of binary search of the list of the starting addresses, where N is the number of addresses in the list.

19. The method of claim 17, wherein the starting address transformation for each entry is performed using a hash table.

20. The method of claim 17, wherein the sample call stack comprising the transformed starting addresses of the functions is transformed into a “

first child-next sibling”

binary tree, resulting in a horizontal array of the frequency counts of each node of the tree.

21. The method of claim 17, wherein individual sample data is preserved by only storing a pointer to the leaf node in the call graph tree for each sample taken.

22. The method of claim 17, wherein the embedded computing device is a video game console, and the computer unit is a personal computer (PC).

23. A method for monitoring and testing software code comprising a set of functions, said software being executed in an embedded computing device connected to a computer unit which stores said computer software, with each of said functions having an address in the memory of said computer unit, wherein the embedded computing device provides display information to a display device at a predetermined periodic frame rate, comprising;

statistically sampling the computer code being executed, wherein the sampling occurs during each frame period, and wherein a list of the starting addresses of all the functions of the computer code being executed is transmitted from the computer unit side to the embedded computing device before the sampling begins and wherein raw sampling data is collected and stored and at the conclusion of the sampling, it is transformed to comprise the starting addresses of the sampled functions of the executed program; and

analyzing the data from the sampling to provide information about the computer software.

24. The method of claim 23, wherein the embedded computing device is a video game console, and the computer unit is a personal computer (PC).

25. A method of profiling a dynamically executing video game program comprising:

sampling said executing program at sampling times that vary but are related to a display refresh rate; and

storing sampling results in a call tree structure and a separate associated call frequency array.