Adaptive thermal ceiling control system
A system performs adaptive thermal ceiling control at runtime. The system includes computing circuits and a thermal management module. When detecting a runtime condition change that affects power consumption in the system, the thermal management module determines an adjustment to the thermal ceiling of a computing circuit, and increases the thermal ceiling of the computing circuit according to the adjustment.
1. A system performing adaptive thermal ceiling control, comprising:
a plurality of computing circuits; and
a thermal management module coupled to the plurality of computing circuits, the thermal management module operative to:
detect a runtime condition change that changes power consumption in the system; and
determine an adjustment to a thermal ceiling of a computing circuit among the plurality of computing circuits, wherein the thermal management module when determining the adjustment is further operative to:
determine a safety margin based on a temperature increase rate of the computing circuit,
subtract the safety margin from a temperature upper bound of the computing circuit to obtain a calculated thermal ceiling value,
when the thermal ceiling is lower than the calculated thermal ceiling value, raise the thermal ceiling to the calculated thermal ceiling value, and
maintain the thermal ceiling when raising the thermal ceiling violates a minimal value of the safety margin.
2. The system of claim 1 , wherein the runtime condition change includes a change to dynamic power consumption and leakage power consumption.
3. The system of claim 1 , wherein the thermal management module is further operative to detect the runtime condition change based on measurements and predictions of power consumption.
4. The system of claim 3 , wherein the predictions of power consumption include a prediction of dynamic power consumption based on one or more of: voltage, frequency, system configuration, workload, and execution time.
5. The system of claim 3 , wherein the predictions of power consumption include a prediction of leakage power consumption based on one or more of: voltage, process variation, system configuration, and power gating.
6. The system of claim 1 , wherein the runtime condition change includes a change in the number of activated computing circuits in the system.
7. The system of claim 1 , wherein the thermal ceiling of each computing circuit is adjusted independent of one another.
8. The system of claim 1 , wherein the thermal management module is further operative to compare a present temperature of the computing circuit with the thermal ceiling to determine whether to throttle performance of the computing circuit.
9. The system of claim 8 , wherein the performance of the computing circuit is throttled at a throttling temperature which changes dynamically during runtime.
10. The system of claim 1 , wherein the thermal ceiling changes dynamically with changes in runtime conditions.
11. A method for adaptively adjusting a thermal ceiling in a system, comprising:
detecting a runtime condition change that changes power consumption in the system; and
determining an adjustment to the thermal ceiling of a computing circuit in the system, wherein determining the adjustment further comprises:
determining a safety margin based on a temperature increase rate of the computing circuit,
subtracting the safety margin from a temperature upper bound of the computing circuit to obtain a calculated thermal ceiling value,
when the thermal ceiling is lower than the calculated thermal ceiling value, raising the thermal ceiling to the calculated thermal ceiling value, and
maintaining the thermal ceiling when raising the thermal ceiling violates a minimal value of the safety margin.
12. The method of claim 11 , wherein the runtime condition change includes a change to dynamic power consumption and leakage power consumption.
13. The method of claim 11 , wherein detecting the runtime condition change is based on measurements and predictions of power consumption.
14. The method of claim 13 , wherein the predictions of power consumption include a prediction of dynamic power consumption based on one or more of: voltage, frequency, system configuration, workload, and execution time.
15. The method of claim 13 , wherein the predictions of power consumption include a prediction of leakage power consumption based on one or more of: voltage, process variation, system configuration, and power gating.
16. The method of claim 11 , wherein the runtime condition change includes a change in the number of activated computing circuits in the system.
17. The method of claim 11 , wherein the system includes a plurality of computing circuits, and the thermal ceiling of each computing circuit is adjusted independent of one another.
18. The method of claim 11 , further comprising:
comparing a present temperature of the computing circuit with the thermal ceiling to determine whether to throttle performance of the computing circuit.
19. The method of claim 18 , wherein the performance of the computing circuit is throttled at a throttling temperature which changes dynamically during runtime.
20. The method of claim 11 , wherein the thermal ceiling changes dynamically with changes in runtime conditions.