利索-全球专利检索

  1. 登录
  2. 注册

搜索结果

114 条记录 (耗时 0.026 秒)

    Dynamic hardware selection for experts in mixture-of-experts model
    1.
    发明授权
    Dynamic hardware selection for experts in mixture-of-experts model 有权

    公开(公告)号：US11893502B2

    公开(公告)日：2024-02-06

    申请号：US15849633

    申请日：2017-12-20

    申请人： Advanced Micro Devices, Inc.

    发明人： Nicholas Malaya ,  Nuwan Jayasena

    IPC分类号： G06N5/022 ,  G06N20/00 ,  G06F7/02

    CPC分类号： G06N5/022 ,  G06N20/00 ,  G06F7/02

    摘要： A system assigns experts of a mixture-of-experts artificial intelligence model to processing devices in an automated manner. The system includes an orchestrator component that maintains priority data that stores, for each of a set of experts, and for each of a set of execution parameters, ranking information that ranks different processing devices for the particular execution parameter. In one example, for the execution parameter of execution speed, and for a first expert, the priority data indicates that a central processing unit (“CPU”) executes the first expert faster than a graphics processing unit (“GPU”). In this example, for the execution parameter of power consumption, and for the first expert, the priority data indicates that a GPU uses less power than a CPU. The priority data stores such information for one or more processing devices, one or more experts, and one or more execution characteristics.

    Approach for reducing side effects of computation offload to memory
    3.
    发明授权
    Approach for reducing side effects of computation offload to memory 有权

    公开(公告)号：US11847055B2

    公开(公告)日：2023-12-19

    申请号：US17364854

    申请日：2021-06-30

    申请人： Advanced Micro Devices, Inc.

    发明人： Shaizeen Aga ,  Nuwan Jayasena

    IPC分类号： G06F12/00 ,  G06F12/084 ,  G06F12/02 ,  G06F12/0862 ,  G06F12/0811

    CPC分类号： G06F12/084 ,  G06F12/0238 ,  G06F12/0811 ,  G06F12/0862

    摘要： A technical solution to the technical problem of how to reduce the undesirable side effects of offloading computations to memory uses read hints to preload results of memory-side processing into a processor-side cache. A cache controller, in response to identifying a read hint in a memory-side processing instruction, causes results of the memory-side processing to be preloaded into a processor-side cache. Implementations include, without limitation, enabling or disabling the preloading based upon cache thrashing levels, preloading results, or portions of results, of memory-side processing to particular destination caches, preloading results based upon priority and/or degree of confidence, and/or during periods of low data bus and/or command bus utilization, last stores considerations, and enforcing an ordering constraint to ensure that preloading occurs after memory-side processing results are complete.

    DEVICE AND METHOD FOR ACCELERATING MATRIX MULTIPLY OPERATIONS
    5.
    发明公开
    DEVICE AND METHOD FOR ACCELERATING MATRIX MULTIPLY OPERATIONS 审中-公开

    公开(公告)号：US20230244751A1

    公开(公告)日：2023-08-03

    申请号：US18297230

    申请日：2023-04-07

    申请人： Advanced Micro Devices, Inc.

    发明人： Shaizeen Aga ,  Nuwan Jayasena ,  Allen H. Rush ,  Michael Ignatowski

    IPC分类号： G06F17/16 ,  G06F7/53 ,  G06F15/80

    CPC分类号： G06F17/16 ,  G06F7/5324 ,  G06F15/8007

    摘要： A processing device is provided which comprises memory configured to store data and a plurality of processor cores in communication with each other via first and second hierarchical communication links. Processor cores of a first hierarchical processor core group are in communication with each other via the first hierarchical communication links and are configured to store, in the memory, a sub-portion of data of a first matrix and a sub-portion of data of a second matrix. The processor cores are also configured to determine a product of the sub-portion of data of the first matrix and the sub-portion of data of the second matrix, receive, from another processor core, another sub-portion of data of the second matrix and determine a product of the sub-portion of data of the first matrix and the other sub-portion of data of the second matrix.

    APPROACH FOR PERFORMING EFFICIENT MEMORY OPERATIONS USING NEAR-MEMORY COMPUTE ELEMENTS
    6.
    发明公开
    APPROACH FOR PERFORMING EFFICIENT MEMORY OPERATIONS USING NEAR-MEMORY COMPUTE ELEMENTS 审中-公开

    公开(公告)号：US20230195618A1

    公开(公告)日：2023-06-22

    申请号：US17557568

    申请日：2021-12-21

    申请人： Advanced Micro Devices, Inc.

    发明人： Shaizeen Aga ,  Johnathan Alsop ,  Nuwan Jayasena

    IPC分类号： G06F12/06

    CPC分类号： G06F12/06

    摘要： Near-memory compute elements perform memory operations and temporarily store at least a portion of address information for the memory operations in local storage. A broadcast memory command is then issued to the near-memory compute elements that causes the near-memory compute elements to perform a subsequent memory operation using their respective address information stored in the local storage. This allows a single broadcast memory command to be used to perform memory operations across multiple memory elements, such as DRAM banks, using bank-specific address information. In one implementation, the approach is used to process workloads with irregular updates to memory while consuming less command bus bandwidth than conventional approaches. Implementations include using conditional flags to selectively designate address information in local storage that is to be processed with the broadcast memory command.

    Hardware-software collaborative address mapping scheme for efficient processing-in-memory systems
    8.
    发明授权
    Hardware-software collaborative address mapping scheme for efficient processing-in-memory systems 有权

    公开(公告)号：US11487447B2

    公开(公告)日：2022-11-01

    申请号：US17006646

    申请日：2020-08-28

    申请人： Advanced Micro Devices, Inc.

    发明人： Mahzabeen Islam ,  Shaizeen Aga ,  Nuwan Jayasena ,  Jagadish B. Kotra

    IPC分类号： G06F12/00 ,  G06F3/06 ,  G06F12/02

    摘要： Approaches are provided for implementing hardware-software collaborative address mapping schemes that enable mapping data elements which are accessed together in the same row of one bank or over the same rows of different banks to achieve higher performance by reducing row conflicts. Using an intra-bank frame striping policy (IBFS), corresponding subsets of data elements are interleaved into a single row of a bank. Using an intra-channel frame striping policy (ICFS), corresponding subsets of data elements are interleaved into a single channel row of a channel. A memory controller utilizes ICFS and/or IBFS to efficiently store and access data elements in memory, such as processing-in-memory (PIM) enabled memory.

    APPROACH FOR ENFORCING ORDERING BETWEEN MEMORY-CENTRIC AND CORE-CENTRIC MEMORY OPERATIONS
    9.
    发明申请
    APPROACH FOR ENFORCING ORDERING BETWEEN MEMORY-CENTRIC AND CORE-CENTRIC MEMORY OPERATIONS 有权

    公开(公告)号：US20220317926A1

    公开(公告)日：2022-10-06

    申请号：US17219446

    申请日：2021-03-31

    申请人： Advanced Micro Devices, Inc.

    发明人： Shaizeen Aga ,  Nuwan Jayasena ,  Johnathan Alsop

    IPC分类号： G06F3/06

    摘要： Ordering between memory-centric memory operations, referred to hereinafter as “MC-Mem-Ops,” and core-centric memory operations, referred to hereinafter as “CC-Mem-Ops,” is enforced using inter-centric fences, referred to hereinafter as an “IC-fences.” IC-fences are implemented by an ordering primitive or ordering instruction, that cause a memory controller, a cache controller, etc., to enforce ordering of MC-Mem-Ops and CC-Mem-Ops throughout the memory pipeline and at the memory controller by not reordering MC-Mem-Ops (or sometimes CC-Mem-Ops) that arrive before the IC-fence to after the IC-fence. Processing of an IC-fence also causes the memory controller to issue an ordering acknowledgment to the thread that issued the IC-fence instruction. IC-fences are tracked at the core and designated as complete when the ordering acknowledgment is received. Embodiments include a completion level-specific cache flush operation which, when used with an IC-fence, provides proper ordering between cached CC-Mem-Ops and MC-Mem-ops with reduced data transfer and completion times.