Power throttling mechanism using instruction rate limiting in high power machine-learning ASICs

    公开(公告)号:US11720158B2

    公开(公告)日:2023-08-08

    申请号:US16818493

    申请日:2020-03-13

    Applicant: Google LLC

    CPC classification number: G06F1/28 G06N20/00

    Abstract: A system contains a machine learning application specific integrated circuit (ASIC) and a power supply unit. The power supply unit and the ASIC are configured to be in data communication through dedicated pins on the ASIC and the power supply unit. The power supply unit detects a present power consumption of the ASIC. Upon determining that a threshold condition has been met, the power supply unit, responsive to the condition sends a digital signal to the ASIC. The ASIC contains a synchronizer which synchronizes the digital signal to be consistent with the ASICs internal clock frequency. A chip manager the synchronized signal and other signals to generate a throttling mask. The throttling mask is sent to a sequencer of the ASIC, which then limits the instruction flow into the processing units of the ASIC based on the mask. This in turn limits the power being consumed by the ASIC.

    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.

    POWER THROTTLING MECHANISM USING INSTRUCTION RATE LIMITING IN HIGH POWER MACHINE-LEARNING ASICs

    公开(公告)号:US20210286419A1

    公开(公告)日:2021-09-16

    申请号:US16818493

    申请日:2020-03-13

    Applicant: Google LLC

    Abstract: A system contains a machine learning application specific integrated circuit (ASIC) and a power supply unit. The power supply unit and the ASIC are configured to be in data communication through dedicated pins on the ASIC and the power supply unit. The power supply unit detects a present power consumption of the ASIC. Upon determining that a threshold condition has been met, the power supply unit, responsive to the condition sends a digital signal to the ASIC. The ASIC contains a synchronizer which synchronizes the digital signal to be consistent with the ASICs internal clock frequency. A chip manager the synchronized signal and other signals to generate a throttling mask. The throttling mask is sent to a sequencer of the ASIC, which then limits the instruction flow into the processing units of the ASIC based on the mask. This in turn limits the power being consumed by the ASIC.

    Active Silicon D2D Bridge
    6.
    发明公开

    公开(公告)号:US20230411297A1

    公开(公告)日:2023-12-21

    申请号:US17841188

    申请日:2022-06-15

    Applicant: Google LLC

    Abstract: A microelectronic system may include a substrate having a first surface, one or more interposers mounted to and electrically connected to the first surface, first and second application specific integrated circuits (ASICs) each at least partially overlying and electrically connected to one of the interposers, a plurality of high-bandwidth memory elements (HBMs) each at least partially overlying and electrically connected to one of the interposers, and an active silicon bridge mounted to and electrically connected to the first surface and providing an electrical connection between the first and second ASICs, the active silicon bridge having active microelectronic devices therein. The microelectronic system may be configured such that the first and second ASICs and the active silicon bridge each have a purely digital CMOS interface therein. A plurality of bumps providing the electrical connection between the ASICs and the active silicon bridge may be configured to receive serial data therethrough.

    Power Throttling Mechanism Using Instruction Rate Limiting in High Power Machine-Learning ASICs

    公开(公告)号:US20230297152A1

    公开(公告)日:2023-09-21

    申请号:US18201974

    申请日:2023-05-25

    Applicant: Google LLC

    CPC classification number: G06F1/28 G06N20/00

    Abstract: A system contains a machine learning application specific integrated circuit (ASIC) and a power supply unit. The power supply unit and the ASIC are configured to be in data communication through dedicated pins on the ASIC and the power supply unit. The power supply unit detects a present power consumption of the ASIC. Upon determining that a threshold condition has been met, the power supply unit, responsive to the condition sends a digital signal to the ASIC. The ASIC contains a synchronizer which synchronizes the digital signal to be consistent with the ASICs internal clock frequency. A chip manager the synchronized signal and other signals to generate a throttling mask. The throttling mask is sent to a sequencer of the ASIC, which then limits the instruction flow into the processing units of the ASIC based on the mask. This in turn limits the power being consumed by the ASIC.

Patent Agency Ranking