Method and apparatus for dynamic power management in a processor system

• US 7,240,223 B2
• Filed: 05/07/2003
• Issued: 07/03/2007
• Est. Priority Date: 05/07/2003
• Status: Active Grant

• Current Assignee: Apple Inc.
• Original Assignee: Apple Inc.
• Inventors: Cox, Keith, de Cesare, Josh, Culbert, Michael
• Primary Examiner(s): Cao; Chun
• Assistant Examiner(s): Bae; Ji H.
• Application Number:
  US10/434,515
• Publication Number:
  US 20040225902A1
• Time in Patent Office: 1,518 Days
• Field of Search: 713/300, 713/320, 713/322
• US Class Current: 713/300
• CPC Class Codes: G06F 1/3203 Power management, i.e. event-based initiation of a power-saving mode
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A dynamic power management system includes an operating system (OS) that causes a processor to operate in one of multiple run states that have different performance and/or power dissipation levels. The OS selects the run state in response to processor information (e.g., processor load) being monitored by the OS. The OS can predict future states of the processor information based on sampled processor information. The OS can take an average of the predicted and actual samples for comparison with a threshold to select a run state. The OS can track the number of consecutive saturated samples that occur during a selected window of samples. The OS can predict future processor information samples based on the number of consecutive saturated samples.

Claims

1. A computing system, comprising:

a processor;

a system memory to store an operating system, the operating system to cause the processor to operate in a state selected from a plurality of states as a function of a sample history and a sample prediction that includes one or more actual and future samples of processor information, each state of the plurality of states having a different performance level, wherein the operating system is to predict one or more future samples of processor information, and, wherein the one or more future samples are predicted as a function of saturated samples.

2. The system of claim 1 wherein future samples are predicted as a function of consecutive saturated samples.

3. The system of claim 1 wherein a threshold value of a sample defining a transition between a saturated sample and a non saturated sample is adjustable.

4. The system of claim 1 wherein future samples are predicted as a number of future saturated samples.

5. The system of claim 4 wherein when a most recent sample is a saturated sample, the number of future saturated samples is equal to a number of consecutive saturated samples, the consecutive samples including the most recent sample.

6. The system of claim 4 wherein when a most recent sample is not a saturated sample, the number of future saturated samples is about half of a number of consecutive saturated samples, the consecutive sample being received sequentially previous to the most recent sample.

7. A method performed by an operating system of a computing platform, the method comprising:

receiving samples of information related to the processor'"'"'s operation; and

causing the operating system to configure the processor to operate in a state selected from a plurality of states as a function of a sample history and a sample prediction having one or more actual and future samples of the information, each state of the plurality of states having a different performance level, wherein the one or more future samples are predicted as a function of saturated samples.

8. The method of claim 7 wherein a threshold value defining a transition between a saturated sample and a non saturated sample is adjustable.

9. The method of claim 7 wherein future samples are predicted as a function of consecutive saturated samples.

10. The method of claim 7 wherein future samples are predicted as a number of future saturated samples.

11. The method of claim 10 wherein when a most recent sample is a saturated sample, the number of future saturated samples is equal to a number of consecutive saturated samples, the consecutive samples including the most recent sample.

12. The method of claim 10 wherein when a most recent sample is not a saturated sample, the number of future saturated samples is about half of a number of consecutive saturated samples, the consecutive sample being received sequentially previous to the most recent sample.

13. An apparatus having a processor, which runs an operating system, the apparatus, comprising:

means for receiving samples of information related to the processor'"'"'s operation; and

means for configuring the processor to operate in a state selected from a plurality of states as a function of a sample history and a sample prediction containing one or more actual and predicted samples of the information, each state of the plurality of states having a different performance level, means for predicting one or more future samples of processor information, and wherein the one or more future samples are predicted as a function of saturated samples.

14. The apparatus of claim 13 wherein a threshold value defining a transition between a saturated sample and a non saturated sample is adjustable.

15. The apparatus of claim 13 wherein future samples are predicted as a function of consecutive saturated samples.

16. The apparatus of claim 13 wherein future samples are predicted as a number of future saturated samples.

17. The apparatus of claim 16 wherein when a most recent sample is a saturated sample, the number of future saturated samples is equal to a number of consecutive saturated samples, the consecutive samples including the most recent sample.

18. The apparatus of claim 16 wherein when a most recent sample is not a saturated sample, the number of future saturated samples is about half of a number of consecutive saturated samples, the consecutive sample being received sequentially previous to the most recent sample.

19. A machine readable medium storing executable instructions, which when executed by a data processing system which runs an operating system, cause the data processing system to perform a method, comprising:

receiving samples of information related to the processor'"'"'s operation; and

causing the operating system to configure the processor to operate in a state selected from a plurality of states as a function of a sample history and a sample prediction containing one or more actual and predicted samples of the information, each state of the plurality of states having a different performance level; and

predicting one or more future samples of processor information, wherein the one or more future samples are predicted as a function of saturated samples.

20. The machine readable medium of claim 19 wherein a threshold value defining a transition between a saturated sample and a non saturated sample is adjustable.

21. The machine readable medium of claim 19 wherein future samples are predicted as a function of consecutive saturated samples.

22. The machine readable medium of claim 19 wherein future samples are predicted as a number of future saturated samples.

23. The machine readable medium of claim 22 wherein when a most recent sample is a saturated sample, the number of future saturated samples is equal to a number of consecutive saturated samples, the consecutive samples including the most recent sample.

24. The machine readable medium of claim 22 wherein when a most recent sample is not a saturated sample, the number of future saturated samples is about half of a number of consecutive saturated samples, the consecutive sample being received sequentially previous to the most recent sample.