发明名称 Predicting energy savings
摘要 A mechanism is provided for estimating energy/power consumption of a fixed-frequency operating mode while system is running in dynamic power management mode. For each time interval in a plurality of time intervals within a time period: a first processor identifies a modeled total nominal power value for at least one second processor during a current time interval, stores the modeled total nominal power value for the current time interval in a storage, identifies a dynamic power management mode power value for the at least one second processor in the data processing system during the current interval, and stores the dynamic power management mode power value for the current time interval in the storage. Responsive to the time period expiring, a comparison is produced of a plurality of modeled total nominal power values and a plurality of dynamic power management mode power values over the time period.
申请公布号 US9329670(B2) 申请公布日期 2016.05.03
申请号 US201213488822 申请日期 2012.06.05
申请人 International Business Machines Corporation 发明人 Allen-Ware Malcolm S.;Huang Wei;Kassem Fadi M.;Rajamani Karthick;Rosedahl Todd J.;Silva Guillermo J.
分类号 G01F1/32;G06F1/32 主分类号 G01F1/32
代理机构 代理人 Lammes Francis;Walder, Jr. Stephen J.;Stock William J.
主权项 1. A method, in a data processing system, for estimating energy/power consumption of a fixed-frequency operating mode while the data processing system is running in a dynamic power management mode, the method comprising: for each time interval in a plurality of time intervals within a time period: identifying, by a first processor, a modeled total nominal power value for a second processor in the data processing system during a current time interval, wherein identifying the modeled total nominal power value for the second processor in the data processing system during the current time interval comprises: identifying, by the first processor, an aggregated activity estimate value for the second processor within the data processing system that indicates the power being used by the second processor in executing activities of a workload in the current time interval as detected by a power sensor associated with the second processor;multiplying, by the first processor, the aggregated activity estimate value with a frequency scaling factor thereby producing a frequency-scaled activity proxy counter value that is indicative of the activities being executed by the second processor as a value of a specified fixed-frequency value;utilizing, vital processor data for the second processor, determining, by the first processor, a power scaling factor for the second processor;multiplying, by the first processor, the power scaling factor with the frequency-scaled activity proxy counter value in order to obtain a modeled active power value;utilizing a current operating temperature of the data processing system as detected by a thermal sensor associated with the data processing system, identifying, by the first processor, a shipped temperature dependent idle power value for the second processor;adding, by the first processor, the identified shipped temperature dependent idle processor power value to the modeled active power value to obtain a modeled processor power value; andadding, by the first processor, the modeled processor power value for the second processor to at least one of: a measured fan power value for a fan in the data processing system for the current time interval as detected by a power sensor associated with the fan, a measured power value for a memory device in the data processing system as detected by a power sensor associated with the memory device, a measured power value for an input/output (I/O) device in the data processing system as detected by a power sensor associated with the I/O device, or a measured power value for a service processor in the data processing system as detected by a power sensor associated with the service processor in order to obtain the modeled total nominal power value for the second processor during the current time interval;storing, by the first processor, the modeled total nominal power value for the current time interval in a storage, wherein, for the plurality of time intervals, the first processor stores a plurality of modeled total nominal power values;identifying, by the first processor, a dynamic power management mode power value for the second processor in the data processing system during the current interval; andstoring, by the first processor, the dynamic power management mode power value for the current time interval in the storage, wherein, for the plurality of time intervals, the first processor stores a plurality of dynamic power management mode power values; and responsive to the time period expiring, producing, by the first processor, a comparison of the plurality of modeled total nominal power values and the plurality of dynamic power management mode power values over the time intervals in the plurality of time intervals in the time period.
地址 Armonk NY US