Large-Scale Accelerator System Energy Performance Optimization

    公开(公告)号:US20230119235A1

    公开(公告)日:2023-04-20

    申请号:US17968048

    申请日:2022-10-18

    Applicant: Google LLC

    Abstract: A method and system for controlling performance of a workload partitioned among a plurality of accelerator chips of a multi-chip system. One or more processors may receive performance speed data for each of the accelerator chips, obtain a model of the partitioned workload, determine a portion of the workload that is either overworked or underworked based on the model of the partitioned workload and the performance speed data for each of the plurality of accelerator chips, and adjust a performance speed of an accelerator chip that performs the portion of the partitioned workload that is either overworked or underworked.

    Dynamic Power-Aware Workload Scheduler

    公开(公告)号:US20250103128A1

    公开(公告)日:2025-03-27

    申请号:US18371012

    申请日:2023-09-21

    Applicant: Google LLC

    Abstract: Systems and methods for managing power allocation by connecting a power capping control loop to a workload scheduler. The work scheduler may receive a workload for execution by one or more of a plurality of machines, assign the workload to one or more designated machines of the plurality of machines, determine a respective power quota for each of the one or more designated machines, instruct a programmable power capping control loop to control operation of each of the one or more designated machines according to its respective power quota; and update, after assigning the workload to the one or more designated machines, a record indicating (i) available power of a domain including the plurality of machines and/or (ii) available machines within the domain.

Patent Agency Ranking