DETECTION OF KNOWN WORKLOAD PATTERNS

    公开(公告)号:US20210311538A1

    公开(公告)日:2021-10-07

    申请号:US17347474

    申请日:2021-06-14

    Abstract: Described are mechanisms and methods for applying Machine Learning (ML) techniques for power management at different levels of a power management stack. An apparatus may comprise a first circuitry, a second circuitry, and a third circuitry. The first circuitry may have a plurality of memory registers. The second circuitry may be operable to establish values for a plurality of features based on samples of values of the plurality of memory registers taken at one or more times within a range of time of predetermined length. The third circuitry may be operable to compare the plurality of features against a plurality of learned parameters for a reference workload.

    HARDWARE APPARATUSES AND METHODS TO PERFORM TRANSACTIONAL POWER MANAGEMENT

    公开(公告)号:US20180059766A1

    公开(公告)日:2018-03-01

    申请号:US15678025

    申请日:2017-08-15

    Abstract: Methods and apparatuses relating to transactional power management are described. In one embodiment, a hardware apparatus includes a hardware processor having a core, a plurality of power domains to transition to one of a plurality of power states in response to a power management command for each power domain, and a power transaction unit to assign a first power management command as a first power transaction and a second power management command as a second power transaction for concurrent execution, perform a commit of the first power transaction and the second power transaction when there is no conflict between the first power transaction and the second power transaction, and perform an abort of the first power transaction and a commit of the second power transaction when there is a conflict between the first power transaction and the second power transaction.

    Power governance of processing unit

    公开(公告)号:US11132046B2

    公开(公告)日:2021-09-28

    申请号:US16650782

    申请日:2017-12-15

    Abstract: Power governance circuitry is provided to control a performance level of a processing unit of a processing platform. The power governance circuitry comprises measurement circuitry to measure a current utilization of the processing unit at a current operating frequency and to determine any change in utilization or power and frequency control circuitry is provided to update the current operating frequency to a new operating frequency by determining a new target quantified power expenditure to be applied in a subsequent processing cycle depending on the determination of any change in utilization or power. A new operating frequency is selected to satisfy the new target quantified power based on a scalability function specifying a variation of a given value of utilization or power with the operating frequency. A processing platform and machine readable instructions are provided to set a new quantified target power of a processing unit.

    Detection of known workload patterns

    公开(公告)号:US11036275B2

    公开(公告)日:2021-06-15

    申请号:US16369643

    申请日:2019-03-29

    Abstract: Described are mechanisms and methods for applying Machine Learning (ML) techniques for power management at different levels of a power management stack. An apparatus may comprise a first circuitry, a second circuitry, and a third circuitry. The first circuitry may have a plurality of memory registers. The second circuitry may be operable to establish values for a plurality of features based on samples of values of the plurality of memory registers taken at one or more times within a range of time of predetermined length. The third circuitry may be operable to compare the plurality of features against a plurality of learned parameters for a reference workload.

Patent Agency Ranking