Power management using reward-based sleep state selection

Patent metadata

Field	Value
Publication number	US-9507403-B2
Application number	US-201113996086-A
Country	US
Kind code	B2
Filing date	Dec 27, 2011
Priority date	Dec 27, 2011
Publication date	Nov 29, 2016
Grant date	Nov 29, 2016

Abstract

Systems and methods may provide for conducting a reward determination for a plurality of sleep states to obtain a plurality of reward determinations with respect to a device. In addition, a sleep state may be selected for the device from the plurality of sleep states based at least in part on the plurality of reward determinations. In one example, false entry and missed opportunity probabilities may be determined for stochastic interrupts, wherein the reward determination is conducted based at least in part on the false entry and missed opportunity probabilities.

First claim

We claim:

1. A system comprising:
a controller having logic to,
determine a false entry probability;
conduct a reward determination for a plurality of sleep states to obtain a plurality of reward determinations with respect to the system; and
select a sleep state for the system from the plurality of sleep states based at least in part on the plurality of reward determinations; and

a power management unit to place the system in the selected sleep state,
wherein the false entry probability is a percentage probability that one or more stochastic interrupts will occur before an energy break-even time associated with a sleep state, and is based at least in part on an estimated stochastic interrupt arrival rate.

2. The system of claim 1 , wherein the logic is to,
determine a missed opportunity probability, wherein the reward determination is based at least in part on the false entry probability and the missed opportunity probability.

3. The system of claim 2 , wherein the missed opportunity probability is a probability that all probabilistic interrupts will occur after the energy break even time associated with the sleep state.

4. The system of claim 2 , wherein the logic is to use a stochastic process to determine the false entry probability and the missed opportunity probability.

5. The system of claim 1 , wherein the logic is to,
determine a false entry penalty, and
determine a missed opportunity penalty, wherein the reward determination is based at least in part on the false entry penalty and the missed opportunity penalty.

6. The system of claim 1 , wherein the logic is to determine a correct state gain, wherein the reward determination is based at least in part on the correct state gain.

7. The system of claim 1 , wherein the logic is to,
distinguish one or more deterministic interrupts from one or more stochastic interrupts,
determine an idle period with respect to at least one of the one or more deterministic interrupts,
determine the energy break even time for the sleep state, and
bypass the reward determination if the energy break even time exceeds the idle period.

8. The system of claim 7 , wherein the logic is to identify a plurality of deterministic interrupt sources, wherein the idle period is to be determined for two or more interrupt sources individually.

9. The system of claim 8 , wherein the logic is to,
group two or more of the plurality of deterministic sources together for consideration, and
identify one or more dependencies associated with the grouped two or more of the plurality of deterministic sources.

10. A non-transitory computer readable storage medium comprising a set of instructions which, if executed by a controller, cause a computer to:
determine a false entry probability;
conduct a reward determination for a plurality of sleep states to obtain a plurality of reward determinations with respect to a device; and
select a sleep state for the device from the plurality of sleep states based at least in part on the plurality of reward determinations,
wherein the false entry probability is a percentage probability that one or more stochastic interrupts will occur before an energy break-even time associated with a sleep state, and is based at least in part on an estimated stochastic interrupt arrival rate.

11. The medium of claim 10 , wherein the instructions, if executed, cause a computer to:
determine a missed opportunity probability, wherein the reward determination is conducted based at least in part on the false entry probability and the missed opportunity probability.

12. The medium of claim 11 , wherein the missed opportunity probability is a probability that all probabilistic interrupts will occur after the energy break even time associated with the sleep state.

13. The medium of claim 11 , wherein the instructions, if executed, cause a computer to use a stochastic process to determine the false entry probability and the missed opportunity probability.

14. The medium of claim 10 , wherein the instructions, if executed, cause a computer to:
determine a false entry penalty; and
determine a missed opportunity penalty, wherein the reward determination is based at least in part on the false entry penalty and the missed opportunity penalty.

15. The medium of claim 10 , wherein the instructions, if executed, cause a computer to determine a correct state gain, wherein the reward determination is based at least in part on the correct state gain.

16. The medium of claim 10 , wherein the instructions, if executed, cause a computer to:
distinguish one or more deterministic interrupts from one or more stochastic interrupts;
determine an idle period with respect to at least one of the one or more deterministic interrupts;
determine the energy break even time for the sleep state; and
bypass the reward determination if the energy break even time exceeds the idle period.

17. The medium of claim 16 , wherein the instructions, if executed, cause a computer to identify a plurality of deterministic interrupt sources, wherein the idle period is to be determined for two or more interrupt sources individually.

18. The medium of claim 17 , wherein the instructions, if executed, cause a computer to:
group two or more of the plurality of deterministic sources together for consideration; and
identify one or more dependencies associated with the grouped two or more of the plurality of deterministic sources.

19. An apparatus comprising:
logic to,
determine a false entry probability;
conduct a reward determination for a plurality of sleep states to obtain a plurality of reward determinations with respect to a device; and
select a sleep state for the device from the plurality of sleep states based at least in part on the plurality of reward determinations,

wherein the false entry probability is a percentage probability that one or more stochastic interrupts will occur before an energy break-even time associated with a sleep state, and is based at least in part on an estimated stochastic interrupt arrival rate.

20. The apparatus of claim 19 , wherein the logic is to,
determine a missed opportunity probability, wherein the reward determination is conducted based at least in part on the false entry probability and the missed opportunity probability.

21. The apparatus of claim 20 , wherein the missed opportunity probability is a probability that all probabilistic interrupts will occur after the energy break even time associated with the sleep state.

22. The apparatus of claim 20 , wherein the logic is to use a stochastic process to determine the false entry probability and the missed opportunity probability.

23. The apparatus of claim 19 , wherein the logic is to,
determine a false entry penalty, and
determine a missed opportunity penalty, wherein the reward determination is based at least in part on the false entry penalty and the missed opportunity penalty.

24. The apparatus of claim 19 , wherein the logic is to determine a correct state gain, wherein the reward determination is based at least in part on the correct state gain.

25. The apparatus of claim 19 , wherein the logic is to,
distinguish one or more deterministic interrupts from one or more stochastic interrupts,
determine an idle period with respect to at least one of the one or more deterministic interrupts,
determine the energy break even time for the sleep state, and
bypass the reward determinati

Assignees

• Min Alexander W
• Wang Ren
• Tsai Jr-Shian
• Ergin Mesut A
• Tai Tsung-Yuan C
• Mavila Rajith K
• Iyer Prakash N
• Intel Corp

Inventors

• Min Alexander W
• Wang Ren
• Tsai Jr-Shian
• Ergin Mesut A
• Tai Tsung-Yuan C
• Mavila Rajith K
• Iyer Prakash N

Classifications

• G06F1/3228 Primary

  Monitoring task completion, e.g. by use of idle timers, stop commands or wait commands · CPC title