Distributed runtime diagnostics in hierarchical parallel environments

• US 8,060,783 B2
• Filed: 02/20/2009
• Issued: 11/15/2011
• Est. Priority Date: 02/20/2009
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: McAllister, Jeffrey S., Mullins, Timothy J., Ramirez, Nelson
• Primary Examiner(s): Riad; Amine
• Application Number:
  US12/389,926
• Publication Number:
  US 20100218045A1
• Time in Patent Office: 998 Days
• Field of Search: 714/31
• US Class Current: 714/31
• CPC Class Codes: G06F 11/3644 by instrumenting at runtime
• CPC Class Codes: G06F 11/366 using diagnostics G06F11/0703 takes precedence

Abstract

A technique is disclosed for distributed runtime diagnostics in hierarchical parallel environments. In one embodiment, a user is allowed to configure, during runtime, a processing element on which to perform diagnostics, an algorithm for the processing element to execute, a data set for the algorithm to execute against, a diagnostic function for the processing element to execute, a condition for executing the diagnostic function, and visualization parameters for memory local to the processing element. As a result, runtime diagnostics can be performed with sufficient degree of control and customization to aid debugging in a hierarchical parallel environment.

Claims

1. A computer-implemented method for performing distributed runtime diagnostics in a hierarchical parallel environment containing a plurality of processing elements, comprising:

receiving, at a user interface, user input specifying at least a first processing element of the hierarchical parallel environment for diagnostics;

receiving, at the user interface, user input specifying an algorithm for the first processing element to execute using a set of data;

receiving, at the user interface, user input specifying a condition, wherein the condition specifies when, during the algorithm, to execute the diagnostic function;

transmitting, via a communication interface to the first processing element, the specified algorithm, the set of data, the diagnostic function, and the specified condition; and

invoking the diagnostic function on the first processing element when the condition is satisfied.

2. The computer-implemented method of claim 1, further comprising:

receiving user input specifying a second processing element of the hierarchical parallel environment for diagnostics;

transmitting, to the second processing element, the specified algorithm, the set of data, a second diagnostic function, and the specified condition; and

invoking the second diagnostic function on the second processing element when the condition is satisfied.

3. The computer-implemented method of claim 1, further comprising:

receiving user input specifying a data size and a sampling rate; and

displaying a visualization based on the specified data size and the specified sampling rate.

4. The computer-implemented method of claim 1, wherein the specified algorithm is instrumented with a diagnostic worker function and wherein at least one of the specified algorithm, the specified data set, and the specified diagnostic function is loaded into a memory for use by the first processing element via the message.

5. The computer-implemented method of claim 4, wherein the hierarchical parallel environment contains a master processing element, wherein the message is sent by the master processing element, and wherein the computer-implemented method further comprises:

receiving user input specifying a data size and a sampling rate;

returning, by the first processing element, data of the specified data size to the master processing element at the specified sampling rate; and

displaying a visualization based on the returned data.

6. The computer-implemented method of claim 4, wherein, the diagnostic function computes a checksum.

7. The computer-implemented method of claim 1, wherein the specified condition includes at least one of a line number, a loop iteration number, a program counter, and the occurrence of an event, and wherein the diagnostic worker function allocates a region of memory local to the first processing element for storing the specified algorithm, the specified data set, the specified diagnostic function, and the specified condition.

8. A non-transitory computer readable storage medium containing a program which, when executed, performs an operation for performing distributed runtime diagnostics in a hierarchical parallel environment containing a plurality of processing elements, comprising:

receiving user input specifying at least a first processing element of the hierarchical parallel environment for diagnostics;

receiving user input specifying an algorithm for the first processing element to execute using a set of data;

receiving user input specifying a condition, wherein the condition specifies when, during the algorithm, to execute the diagnostic function;

transmitting, to the first processing element, the specified algorithm, the set of data, the diagnostic function, and the specified condition; and

invoking the diagnostic function on the first processing element when the condition is satisfied.

9. The computer readable storage medium of claim 8, wherein the program further comprises:

receiving user input specifying a second processing element of the hierarchical parallel environment for diagnostics;

transmitting, to the second processing element, the specified algorithm, the set of data, the diagnostic function, and the specified condition; and

invoking the diagnostic function on the second processing element when the condition is satisfied.

10. The computer readable storage medium of claim 8, wherein the program further comprises:

receiving user input specifying a data size and a sampling rate for visualization purposes; and

displaying visualization information based on the specified data size and the specified sampling rate.

11. The computer readable storage medium of claim 8, wherein the specified algorithm is instrumented with a diagnostic worker function and wherein at least one of the specified algorithm, the specified data set, and the specified diagnostic function is loaded into a memory for use by the first processing element via the message.

12. The computer readable storage medium of claim 11, wherein the hierarchical parallel environment contains a master processing element, wherein the message is sent by the master processing element, and wherein the program further comprises:

receiving user input specifying a data size and a sampling rate;

returning, by the first processing element, data of the specified data size to the master processing element at the specified sampling rate; and

displaying a visualization based on the returned data.

13. The computer readable storage medium of claim 11, wherein, the diagnostic function computes a checksum.

14. The computer readable storage medium of claim 8, wherein the specified condition includes at least one of a line number, a loop iteration number, a program counter, and the occurrence of an event, and wherein the diagnostic worker function allocates a region of memory local to the first processing element for storing the specified algorithm, the specified data set, the specified diagnostic function, and the specified condition.

15. A system, comprising:

a processor; and

a memory containing a program, which when executed by the processor is configured to perform an operation for performing distributed runtime diagnostics in a hierarchical parallel environment containing a plurality of processing elements, comprising;

receiving user input specifying a processing element of the hierarchical parallel environment for diagnostics;

receiving user input specifying an algorithm for the processing element to execute;

receiving user input specifying a data set for the algorithm to execute against;

receiving user input specifying a diagnostic function for the processing element to execute;

receiving user input specifying a condition, wherein the condition specifies when, during the algorithm, to execute the diagnostic function;

loading the specified algorithm, the specified data set, the specified diagnostic function, and the specified condition into a local memory of the specified processing element; and

executing the specified diagnostic function on the specified processing element when the condition is satisfied.

16. The system of claim 15, wherein the program further comprises:

receiving user input specifying a second processing element of the hierarchical parallel environment for diagnostics;

transmitting, to the second processing element, the specified algorithm, the set of data, the diagnostic function, and the specified condition; and

invoking the diagnostic function on the second processing element when the condition is satisfied.

17. The system of claim 15, wherein the program further comprises:

receiving user input specifying a data size and a sampling rate for visualization purposes; and

displaying visualization information based on the specified data size and the specified sampling rate.

18. The system of claim 15, wherein the specified algorithm is instrumented with a diagnostic worker function and wherein at least one of the specified algorithm, the specified data set, and the specified diagnostic function is loaded into a memory for use by the first processing element via the message.

19. The system of claim 18, wherein the hierarchical parallel environment contains a master processing element, wherein the message is sent by master processing element, and wherein the program further comprises:

receiving user input specifying a data size and a sampling rate;

returning, by the first processing element, data of the specified data size to the master processing element at the specified sampling rate; and

displaying a visualization based on the returned data.

20. The system of claim 18, wherein, the diagnostic function computes a checksum.

21. The system of claim 15, wherein the specified condition includes at least one of a line number, a loop iteration number, a program counter, and the occurrence of an event, and wherein the diagnostic worker function allocates a region of memory local to the first processing element for storing the specified algorithm, the specified data set, the specified diagnostic function, and the specified condition.

22. A computer-implemented method for performing distributed runtime diagnostics in a computing environment which includes a hierarchy of processing elements including at least a main processing element and a plurality of subordinate processing element, the method comprising:

invoking, by the main processing element, executable code on one or more of the subordinate processing elements, wherein the executable code on at least a first one of the one or more subordinate processing elements is instrumented to invoke a diagnostic function when specified conditions are satisfied;

invoking the diagnostic function, via the instrumented executable code, during execution of the executable code on the first subordinate processing element, on the subordinate processing element when the conditions are satisfied;

storing, by the first subordinate processing element, a result from obtained from invoking the diagnostic function in a dedicated region of memory local to the first subordinate processing element; and

reporting, by the first subordinate processing element, the result to the main processing element.