International Business Machines Corporation

Hye-Young McCreary

Andrew J Geissler

Malcolm Scott Ware

Andreas Bieswanger

Freeman Leigh Rawson, Iii

Michael Stephen Floyd

Raymond J Harrington

Soraya Ghiasi

Freeman L Rawson

Ajay Dholakia

Thomas M Brey

Guillermo Jesus Silva

Karthick Rajamani

Malcolm Scott Allen-Ware

Juan C Rubio

Thomas W Keller, Jr

Steven Paul Hartman

A partition that is executed by multiple processing nodes. Each node includes multiple cores and each of the cores has a frequency that can be set. A first frequency range is provided to the cores. Each core, when executing the identified partition, sets its frequency within the first frequency range. Frequency metrics are gathered from the cores running the partition by the nodes. The gathered frequency metrics are received and analyzed by a hypervisor that determines a second frequency range to use for the partition, with the second frequency range being different from the first frequency range. The second frequency range is provided to the cores at the nodes executing the identified partition. When the cores execute the identified partition, they use a frequencies within the second frequency range.

Partition level power management using fully asynchronous cores with software that has limited asynchronous support

A method, computer program product, and apparatus for managing power management in a data processing system are presented. A core is activated and configured to operate at a frequency in response to a request to increase a processing capacity. A determination whether a use of power resulting from activating the core meets a policy for the use of the power is made. A set of parameters is adjusted to meet the policy for the use of power in response to a determination that the use of power does not meet the policy. A determination whether a number of operations performed by a set of cores is made. An indication that the request to increase the processing capacity is unavailable is made in response to the number of operations having not increased.

Power management for processing capacity upgrade on demand

Mechanisms for dynamically managing core pools are provided. The mechanisms measure the amount of a computational load on a computing device. One way of measuring the load may consist of executing a number of instructions, in a unit of time, with numerous cores of the computing device. The mechanisms compare the number of instructions executed with specific thresholds. Depending on whether the number of instructions is higher or lower than the thresholds, the computing devices may respond by activating and deactivating cores of the computing devices. By limiting execution of instructions of the computing device to a smaller number of cores and switching one or more cores to a lower power state, the devices may conserve power.

Dynamic core pool management

A method, computer program product, and apparatus for managing power in a data processing system are presented. A core is activated in the data processing system and configured to operate at a frequency in response to receiving a request to increase a processing capacity of a set of resources in the data processing system. A determination whether a use of power resulting from activating the core configured to operate at the frequency meets a policy for the use of the power in the data processing system is made. A set of parameters associated with devices in the set of resources are adjusted to meet the policy for the use of power in the data processing system in response to a determination that the use of power does not meet the policy. A determination whether a number of operations performed per unit of time by a set of cores associated with the set of resources increased after activating the core is made. An indication that the request to increase the processing capacity of the set of resources is unavailable is made in response to a determination that the number of operations performed per unit of time by the set of cores associated with the set of resources has not increased.

A method for dynamic power and performance calibration of a data processing system is provided in the illustrative embodiments. A synthesized program loaded in the data processing system is executed responsive to detecting an event in the data processing system. The synthesized program is configured to generate a set of data that is indicative of the data processing system's power-performance characteristics under varying conditions of operation. Using the set of data, a determination is made of a performance limit on an operation of the data processing system under present operating conditions of the data processing system. A parameter of the data processing system is calibrated to operate the data processing system within the performance limit.

Dynamic power and performance calibration of data processing systems

A method, system, and computer usable program product for power distribution considering cooling nodes in a data processing environment. A power demand of a data processing environment is determined for a period. The data processing environment includes a set of computing nodes and cooling nodes. A determination is made that the power demand will exceed a limit on electrical power available to the data processing environment for the period if the computing nodes and the cooling nodes in the data processing environment are operated in a first configuration. A first amount of power is redistributed from a cooling node in the data processing environment to a computing node in the data processing environment such that a temperature related performance threshold of a subset of computing nodes is at least met.

Power distribution considering cooling nodes

Dynamic frequency and voltage scaling for a computer processor, including retrieving information specifying a nominal operating point of frequency and voltage and an operating range of frequency and voltage for the processor; creating, by the power management module dynamically at run time in dependence upon the retrieved information, a table of frequency, voltage pairs, each pair specifying an operating point of frequency and voltage for the processor, each pair disposed upon a line drawn in frequency-voltage space through the nominal operating point between the minimum operating point and the maximum operating point, the distance between each pair defined in dependence upon a minimum change in power supply voltage supported by the power supply; and selecting and applying, by the power management module from the table, an operating voltage and frequency for the processor in dependence upon current operating conditions of the processor.

Dynamic frequency and voltage scaling for a computer processor

A system for identifying a subset of sensors to sample to reduce the frequency of sensor access. The system determines rise times and records values for the sensors in the system. A time criticality of the sensors is determined based on the rise times. The system processes the sensors by first creating sensor subsets based on one or more constraints on the sensors. The system monitors the values of the sensors in a sensor subset and flags a sensor when it makes a determination that, prior to a next scheduled sampling of the sensor subset, the value of a sensor in the monitored sensor subset will exceed a threshold constraint. The system moves those flagged sensors to a second sensor subset which complies with the sensor's constraints.

Sensor subset selection for reduced bandwidth and computation requirements

Disclosed is a computer implemented method, computer program product, and apparatus for determining a safe lower bound for a commonly powered data processing system. A power management module operates the data processing system using at least one nominal operating parameter during an exploration periodicity, with the at least one nominal operating parameter being clock speed. The power management module determines whether a calibration period is occurring. The power management module calibrates the data processing system up to a measurement interval duration expiration. The power management module may repeat operating the data processing system using the at least one nominal operating parameter.

Power cap lower bound exploration in a server environment

A mechanism is provided for unified management of power, performance, and thermals in computer systems. This mechanism incorporates elements to effectively address all aspects of managing computing systems in an integrated manner, instead of independently. The mechanism employs an infrastructure for real-time measurements feedback, an infrastructure for regulating system activity, component operating levels, and environmental control, a dedicated control structure for guaranteed response/preemptive action, and interaction and integration components. The mechanism provides interfaces for user-level interaction. The mechanism also employs methods to address power/thermal concerns at multiple timescales. In addition, the mechanism improves efficiency by adopting an integrated approach, rather than treating different aspects of the power/thermal problem as individual issues to be addressed in a piecemeal fashion.

Unified management of power, performance, and thermals in computer systems

A system and method of parameter measurement for a distributed computer system, the method including selecting a master unit; selecting slave units operably connected to the master unit on a bus, the slave units having a slave clock; determining slave unit latencies between the master unit and the slave units; generating slave unit synchronizing signals for the slave units, the slave unit synchronizing signals being adjusted for the slave unit latencies; synchronizing the slave clocks in response to the slave unit synchronizing signals; and measuring an operating parameter at the slave units at a synchronously determined time.

Growing community of inventors

Liberty Hill, TX, United States of America

Hye-Young McCreary