System and method for adjusting processor performance based on platform and ambient thermal conditions
A system and method for efficient management of operating modes within an integrated circuit (IC) for optimal power and performance targets. A semiconductor chip includes processing units each of which operates with respective operating parameters. Temperature sensors are included to measure a temperature of the one or more processing units during operation. A power manager determines a calculated power value independent of thermal conditions and current draw. The power manager reads each of a first thermal design power (TDP) value for the processing units and a second TDP value for a platform housing the semiconductor chip. The power manager determines a ratio of the first TDP value to the second TDP value. Additionally, the power manager determines another ratio of the first TDP value to the calculated power value. Using the measured temperature, the ratios and the calculated power value, the power manager determines a manner to adjust the operating parameters.
1. A semiconductor chip comprising:
one or more processing units, each operating with respective operating parameters;
a power manager configured to:
receive a sensor measured temperature of the one or more processing units;
determine whether or not an ambient temperature is to be used when adjusting the respective operating parameters;
use the ambient temperature when adjusting the respective operating parameters in response to detecting a first condition; and
exclude the ambient temperature when adjusting the respective operating parameters in response to detecting a second condition.
2. The semiconductor chip as recited in claim 1 , wherein:
the first condition is the sensor measured temperature does not exceed a temperature threshold; and
the second condition is the sensor measured temperature exceeds the temperature threshold.
3. The semiconductor chip as recited in claim 2 , wherein the power manager is further configured to determine a calculated power based on activity level data from each of the one or more processing units.
4. The semiconductor chip as recited in claim 2 , wherein the temperature threshold is based on a difference between a maximum temperature limit and an under-reporting error of a temperature sensor used to determine the sensor measured temperature.
5. The semiconductor chip as recited in claim 2 , wherein in response to detecting the second condition, the power manager is further configured to:
determine a calculated temperature based on the calculated power value and a worst-case ambient temperature;
determine a first difference between a maximum temperature limit and the calculated temperature; and
adjust the respective operating parameters based at least in part on the first difference.
6. The semiconductor chip as recited in claim 2 , wherein in response to detecting the first condition, the power manager is configured to use the ambient temperature to determine a boosted power-performance state (P-state) for the one or more processing units, wherein said boosted P-state is a higher P-state than a P-state that would be determined without using the ambient temperature.
7. The semiconductor chip as recited in claim 6 , wherein in response to detecting the first condition, the power manager is further configured to:
determine a first boost value based on the maximum temperature limit and the sensor measured temperature;
determine an extra boost value based at least in part on the ambient temperature;
determine a total boost value based on a sum of the first boost value and the extra boost value; and
adjust the respective operating parameters to select the boosted P-state using the total boost value.
8. A method comprising:
receiving a sensor measured temperature of one or more processing units, each operating with respective operating parameters;
determining whether or not an ambient temperature is to be used when adjusting the respective operating parameters;
using the ambient temperature when adjusting the respective operating parameters in response to detecting a first condition; and
excluding the ambient temperature when adjusting the respective operating parameters in response to detecting a second condition.
9. The method as recited in claim 8 , wherein:
the first condition is the sensor measured temperature does not exceed a temperature threshold; and
the second condition is the sensor measured temperature exceeds the temperature threshold.
10. The method as recited in claim 9 , wherein the method further comprises determining a calculated power based on activity level data from each of the one or more processing units.
11. The method as recited in claim 9 , wherein the temperature threshold is based on a difference between a maximum temperature limit and an under-reporting error of a temperature sensor used to determine the sensor measured temperature.
12. The method as recited in claim 9 , wherein in response to detecting the second condition, the method further comprises:
determining a calculated temperature based on the calculated power value and a worst-case ambient temperature;
determining a first difference between a maximum temperature limit and the calculated temperature; and
adjusting the respective operating parameters based at least in part on the first difference.
13. The method as recited in claim 9 , wherein in response to detecting the first condition, the method further comprises using the ambient temperature to determine a boosted power-performance state (P-state) for the one or more processing units, wherein said boosted P-state is a higher P-state than a P-state that would be determined without using the ambient temperature.
14. The method as recited in claim 13 , wherein in response to detecting the first condition, the method further comprises:
determining a first boost value based on the maximum temperature limit and the sensor measured temperature;
determining an extra boost value based at least in part on the ambient temperature;
determining a total boost value based on a sum of the first boost value and the extra boost value; and
adjusting the respective operating parameters to select the boosted P-state using the total boost value.
15. An on-die power management system comprising:
an input adjuster unit; and
an operation parameter selector; and
wherein the input adjuster unit is configured to:
receive a sensor measured temperature of one or more processing units, each operating with respective operating parameters; and
determine whether or not an ambient temperature is to be used when adjusting the respective operating parameters;
wherein the operation parameter selector is configured to:
use the ambient temperature when adjusting the respective operating parameters in response to detecting a first condition; and
exclude the ambient temperature when adjusting the respective operating parameters in response to detecting a second condition.
16. The power management system as recited in claim 15 , wherein:
the first condition is the sensor measured temperature does not exceed a temperature threshold; and
the second condition is the sensor measured temperature exceeds the temperature threshold.
17. The power management system as recited in claim 16 , wherein in response to detecting the first condition, the operation parameter selector is configured to use the ambient temperature to determine a boosted power-performance state (P-state) for the one or more processing units, wherein said boosted P-state is a higher P-state than a P-state that would be determined without using the ambient temperature.