Abstract:
Power and performance of a multi-core system is managed dynamically by adjusting power table indices at runtime. Runtime statistics is measured, when an application is executed on a first core of a first type at a first operating point (OPP) in a first time period, and on a second core of a second core type at a second OPP in a second time period. A controller estimates, based on the runtime statistics, a first pair of indices associated with a first OPP for the first core and a second pair of indices associated with a second OPP for the second core. During runtime, the controller incorporates the first pair of indices and the second pair of indices into power table indices; and determines, from the power table indices, selected indices associated with a selected OPP of a core of a selected core type for executing the application.
Abstract:
A switch interconnect is dynamically controlled at runtime to connect power sources to processing units in a multiprocessor system. Each power source is shareable by the processing units and each processing unit has a required voltage for processing a workload. When a system condition is detected at runtime, the switch interconnect is controlled to change a connection between at least one processing unit and a shared power source to maximize power efficiency. The shared power source is one of the power sources that supports multiple processing units having different required voltages.
Abstract:
A method and a computer-readable medium for dynamically managing power of a multi-core processor of a computing system are provided. The multi-core processor generates a dynamic voltage and frequency scaling (DVFS) table, determines a first index by alternatively selecting either a power budget or a required performance thereof, determines a current thread level parallelism (TLP) of the computing system, selects one of entries according to the current TLP and the first index, and configure first cores and second cores thereof according to a first settings and a second settings of the selected entry.
Abstract:
The surface temperature of a portable device is estimated. A sensor detects the internal temperature of the portable device. The internal temperature and an ambient temperature are used as input to a circuit model that describes thermal behaviors of the portable device. Dynamic thermal management may be performed based on the estimated surface temperature.
Abstract:
Power and performance of a multi-core system is managed dynamically by adjusting power table indices at runtime. Runtime statistics is measured, when an application is executed on a first core of a first type at a first operating point (OPP) in a first time period, and on a second core of a second core type at a second OPP in a second time period. A controller estimates, based on the runtime statistics, a first pair of indices associated with a first OPP for the first core and a second pair of indices associated with a second OPP for the second core. During runtime, the controller incorporates the first pair of indices and the second pair of indices into power table indices; and determines, from the power table indices, selected indices associated with a selected OPP of a core of a selected core type for executing the application.
Abstract:
An electronic device has a processing system and a management circuit. The processing system executes an application. The management circuit detects an operating behavior of the application during execution of the application, analyzes the detected operating behavior of the application to generate an application identification result, and configures an application-dependent task according to at least the application identification result.
Abstract:
The present invention provides a method of dynamic thermal management applied to a portable device, wherein the method includes the steps of: obtaining a surface temperature of the portable device; obtaining a junction temperature of a chip of the portable device; and calculating an upper limit of the junction temperature according to the junction temperature and the surface temperature.
Abstract:
The surface temperature of a portable device is estimated. The portable device includes a sensor for detecting the internal temperature of the portable device. The portable device also includes circuitry for estimating the surface temperature, using the internal temperature and an ambient temperature of the portable device as input to a circuit model. The circuit model describes thermal behaviors of the portable device. The circuitry is operative to identify a scenario in which the portable device operates, and determine the ambient temperature using the scenario and at least the internal temperature.
Abstract:
A method and a computer-readable medium for dynamically managing power of a multi-core processor of a computing system are provided. The multi-core processor generates a dynamic voltage and frequency scaling (DVFS) table, determines a first index by alternatively selecting either a power budget or a required performance thereof, determines a current thread level parallelism (TLP) of the computing system, selects one of entries according to the current TLP and the first index, and configure first cores and second cores thereof according to a first settings and a second settings of the selected entry.
Abstract:
Methods and apparatus are provided for adaptive thermal slope control for dynamic thermal management. In one novel aspect, the device monitors and obtains sampling temperatures, calculates a thermal-slope index, determines whether the calculated thermal-slope index is greater than a predefined slope threshold, adjusts a power budget based on a thermal-slope algorithm, and applies the dynamic thermal management (DTM) adaptively based on the adjusted power budget. In one embodiment, fixed slope algorithm is used. The power budget is adjusted such that an adjusted slope of temperatures stays at a constant. In another embodiment, the time prediction algorithm is used. The power budget is adjusted such that a predicted time to reach a predefined thermal threshold stays a constant. In one embodiment, the time-prediction algorithm is a time-to-target-point (T2TP) algorithm. The T2TP is obtained using a linear equation or a LOG equation.