GPU Circuit Self-Context Save During Context Unmap

    公开(公告)号:US20240419482A1

    公开(公告)日:2024-12-19

    申请号:US18336420

    申请日:2023-06-16

    Abstract: Systems and methods for efficient context switching in multithread processors are disclosed. A processing system comprises a direct memory access module configured to detect a preemption request generated by the scheduling circuit. Responsive to the preemption request, the direct memory access module determines whether execution of a first task from a plurality of tasks needs to be replaced by execution of a second task. When the replacement is necessitated, the module saves a first plurality of registers associated with the first task at a memory location transmitted by the scheduling circuit and queues the second task for execution. The memory location is transmitted by the scheduling circuit as part of the preemption request.

    Processor Power Management Utilizing Dedicated DMA Engines

    公开(公告)号:US20230098742A1

    公开(公告)日:2023-03-30

    申请号:US17490003

    申请日:2021-09-30

    Abstract: Apparatuses, systems and methods for performing efficient power management for a processing unit. A processing unit includes two partitions, each assigned to a respective power domain with operating parameters, and each with a respective direct memory access
    (DMA) engine. If a controller determines a task type of a received task indicates the task is to be processed by components of the second partition, then the controller assigns the task to the second partition and maintains the operational parameters of the first power domain for the components of the first partition or selects lower performance operational parameters of the first power domain. The processing unit accesses data stored in memory using a DMA engine and operational parameters of the second partition. Additionally, the second partition processes the task using the operational parameters of the second power domain.

Patent Agency Ranking