System and method of workload-dependent reliability projection and monitoring for microprocessor chips and systems

• US 20050257078A1
• Filed: 04/21/2004
• Published: 11/17/2005
• Est. Priority Date: 04/21/2004
• Status: Active Grant

• Current Assignee: GlobalFoundries, Inc.
• Original Assignee: International Business Machines Corporation
• Inventors: Rivers, Jude A., Bose, Pradip, Srinivasan, Jayanth
• Granted Patent: US 7,506,216 B2
• Time in Patent Office: Days
• US Class Current: 714/1
• CPC Class Codes: G06F 11/008 Reliability or availability analysis
• CPC Class Codes: G06F 11/3419 by assessing time
• CPC Class Codes: G06F 11/348 Circuit details, i.e. tracer hardware
• CPC Class Codes: G06F 2201/88 Monitoring involving counting

Abstract

A system and method for projecting reliability includes a module, such as a chip, which includes workload inputs, which account for activity on the chip. A reliability module interacts with the chip to determine a reliability measurement for the chip based upon the workload inputs such that functions of the chip are altered based upon the reliability measurement. The reliability measurements are employed to rate or improve chip designs or calculate a reliability measure in real-time.

Claims

1. A system for projecting reliability, comprising:

a module which includes workload inputs which account for activity on the module; and

a reliability module interacting with the module to determine a reliability measurement for the module in real-time based upon the workload inputs and a measured operational quantity of the module.

2. The system as recited in claim 1, wherein the activity of the module includes measurement of activity of hardware on the module.

3. The system as recited in claim 2, wherein the measurement of module function is determined by an instruction per cycle analysis.

4. The system as recited in claim 2, wherein the measurement of module function is determined by counts on counters provided on the module.

5. The system as recited in claim 2, wherein the activity of the module includes temperature, power, and frequency measurements.

6. The system as recited in claim 2, wherein the activity of the module includes energy estimates.

7. The system as recited in claim 1, wherein the reliability module is integrated in the module.

8. The system as recited in claim 7, wherein the system further includes an operating system for monitoring the workload and the reliability module includes a software module to determine the reliability measurement off-module.

9. The system as recited in claim 1, wherein the reliability measurement includes a mean time to failure value for the entire module.

10. A method for designing a module or system of modules based upon projected reliability, comprising the steps of:

simulating a module response based upon design information;

estimating module activity based upon a workload trace;

converting the module response and the module activity to failure rate data; and

analyzing a design of the module based upon the failure rate data.

11. The method as recited in claim 10, wherein the step of simulating a module response includes simulating power, temperature, frequency and performance of the module.

12. The method as recited in claim 10, wherein the step of estimating module activity based upon a workload trace includes estimating module activity by employing an instruction per cycle analysis.

13. The method as recited in claim 10, wherein the step of converting the module response and the module activity to failure rate data includes calculating failure rate data based upon a plurality failure mechanisms using data obtained from the module response and the module activity.

14. The method as recited in claim 10, wherein the step of analyzing includes estimating a module-wide mean time to failure.

15. A program storage device readable by machine, tangibly embodying a program of instructions executable by the machine to perform method steps for designing a module or system of modules based upon projected reliability, as recited in claim 10.

16. A method for monitoring a module or system of modules based upon projected reliability, comprising the steps of:

determining physical responses of a module during operation, the responses including module activity;

converting the physical responses of the module to failure rate data by performing reliability evaluations based upon the physical responses; and

determining a reliability projection based on the failure rate data.

17. The method as recited in claim 16, wherein the step of determining physical responses of a module during operation includes determining one or more of power, temperature, frequency and performance of the module.

18. The method as recited in claim 16, wherein the step of converting the physical responses of the module to failure rate data includes calculating failure rate data based upon a plurality failure mechanisms using data obtained from the physical responses and module activity.

19. The method as recited in claim 16, wherein the module activity is determined by estimating module activity using an instruction per cycle analysis.

20. The method as recited in claim 16, wherein the module activity is measured by providing counter devices to measure module activity.

21. The method as recited in claim 16, wherein the step of determining a reliability projection based the failure rate data includes estimating a module-wide mean time to failure.

22. The method as recited in claim 16, wherein the step of converting the physical responses of the module to failure rate data by performing reliability evaluations based upon the physical responses includes weighting physical responses in accordance with hardware features of the module.

23. The method as recited in claim 16, wherein the step of determining physical responses of a module includes providing sensors on the module to make measurements.

24. The method as recited in claim 16, wherein the step of converting the physical responses of the module to failure rate data includes providing a reliability module on-module to perform the step of converting.

25. The method as recited in claim 16, wherein the step of converting the physical responses of the module to failure rate data includes providing a software executable reliability module off-module to perform the step of converting.