Systems, Devices, and/or Methods for Managing Energy Usage

• US 20110252248A1
• Filed: 09/02/2009
• Published: 10/13/2011
• Est. Priority Date: 09/08/2008
• Status: Active Grant

• Current Assignee: Virginia Tech Intellectual Properties, Inc.
• Original Assignee: Virginia Tech Intellectual Properties, Inc.
• Inventors: Turner, Joseph, Cameron, Kirk
• Granted Patent: US 8,918,657 B2
• Time in Patent Office: Days
• US Class Current: 713/300
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202
• CPC Class Codes: G06Q 50/06 Energy or water supply

Abstract

Certain exemplary embodiments can provide a system, machine, device, manufacture, circuit, and/or user interface adapted for, and/or a method and/or machine-readable medium comprising machine-implementable instructions for, activities that can comprise, based on a predicted non-zero future power requirement for performing one or more portions of an expected workload by a predetermined system, the predicted future power requirement based on a past power requirement for performing one or more portions of a known workload by the predetermined system, automatically causing a change from a first power state of the predetermined system to a second power state that does not violate a specification for performing the one or more portions of the expected workload.

Claims

1. A method comprising:

via a predetermined processor;

based on a predicted non-zero future power requirement for performing one or more portions of an expected workload by a predetermined system, said predicted future power requirement based on a past power requirement for performing one or more portions of a known workload by said predetermined system, automatically causing a change from a first power state of said predetermined system to a second power state that does not violate a guaranteed deterministically-bounded explicit specification for performing said one or more portions of the expected workload.

2. The method of claim 1, further comprising:

receiving said specification.

3. The method of claim 1, further comprising:

automatically determining said past power requirement.

4. The method of claim 1, further comprising:

automatically monitoring power utilization for an identified workload.

5. The method of claim 1, further comprising:

automatically transforming information about power utilized when performing an identified workload into a discrete set of workload classes.

6. The method of claim 1, further comprising:

automatically transforming information about power utilized when performing an identified workload into a discrete set of workload classes; and

automatically predicting a next workload class from said set of workload classes.

7. The method of claim 1, further comprising:

transforming information about power utilized when performing an identified workload into a discrete set of workload classes;

predicting a next workload class from said set of workload classes; and

transforming a next workload class into an actual power state.

8. The method of claim 1, further comprising:

transforming information about power utilized when performing an identified workload into a discrete set of workload classes;

predicting a next workload class from said set of workload classes;

transforming a next workload class into an actual power state; and

identifying an amount to modify said actual power state to avoid violating said specification.

9. The method of claim 1, further comprising:

automatically predicting said future power requirement.

10. The method of claim 1, further comprising:

guaranteeing fulfillment of said future power requirement while incurring no greater than a user-specified performance loss.

11. The method of claim 1, further comprising:

automatically fulfilling said future power requirement while incurring no greater than a user-specified performance loss.

12. The method of claim 1, further comprising:

dividing a software process into a plurality of distinct workloads, each of said distinct workloads characterizable by a substantially constant power requirement.

13. The method of claim 1, further comprising:

identifying an available performance slack from an estimate of a current performance loss and an acceptable performance loss.

14. The method of claim 1, further comprising:

co-optimizing a chosen power state and a length of a next timestep.

15. The method of claim 1, further comprising:

co-optimizing a chosen power state and a magnitude of a next sampling rate.

16. The method of claim 1, wherein:

said specification identifies a performance loss limit.

17. The method of claim 1, wherein:

said specification deterministically bounds a user-specified performance loss.

18. The method of claim 1, wherein:

said power state comprises an operating frequency.

19. The method of claim 1, wherein:

said power state comprises a operating voltage.

20. The method of claim 1, wherein:

said predicted future power requirement is obtained via a Markov model.

21. A system comprising:

a processor adapted for;

based on a predicted non-zero future power requirement for performing an expected workload by a predetermined system, said predicted future power requirement based on a past power requirement for performing a known workload by said predetermined system, automatically causing a change from a first power state of said predetermined system to a second power state that is sufficient to perform the expected workload, yet that does not violate a guaranteed deterministically-bounded explicit performance specification.

22. A machine-readable medium comprising machine-implementable instructions for activities comprising:

based on a predicted non-zero future power requirement for performing an expected workload by a predetermined system, said predicted future power requirement based on a past power requirement for performing a known workload by said predetermined system, automatically causing a change from a first power state of said predetermined system to a second power state that is sufficient to perform the expected workload, yet that does not violate a guaranteed deterministically-bounded explicit performance specification.

23. A method comprising:

via a predetermined processor;

based on a predicted non-zero future power requirement for performing an expected workload by a predetermined system, said predicted future power requirement based on a past power requirement for performing a known workload by said predetermined system, dynamically adaptively controlling a power state of said predetermined system to a level that is sufficient to perform the expected workload yet prevents violation of a guaranteed deterministically-bounded explicit performance specification.