中科微点-全球专利检索

  1. Login
  2. Sign Up

Search Results

9 records (took 0.016 seconds)

    METHOD AND SYSTEM FOR ACCELERATION OR OFFLOADING UTILIZING A MULTIPLE INPUT DATA STREAM
    1.
    发明公开
    METHOD AND SYSTEM FOR ACCELERATION OR OFFLOADING UTILIZING A MULTIPLE INPUT DATA STREAM 审中-公开

    公开(公告)号：US20230409226A1

    公开(公告)日：2023-12-21

    申请号：US18461989

    申请日：2023-09-06

    Applicant: Lemon Inc. ,  Douyin Vision Co., Ltd.

    Inventor： Chul LEE ,  Hui ZHANG ,  Bo LI ,  Shan XIAO ,  Ping ZHOU ,  Fei LIU

    IPC: G06F3/06

    CPC classification number: G06F3/0647 ,  G06F3/067 ,  G06F3/0604

    Abstract: Methods and systems for performing a hardware acceleration process that includes a hardware offloading engine. The hardware offloading engine includes an interface for communicating with a host device and a hardware accelerator for performing the acceleration process. The hardware accelerator has a processor configured to: receive from the interface a hardware offloading command from the host device, the hardware offloading command including an operation code, an input pointer, a total input size, and a first input size, parse the operation code, the input pointer, the total input size, and the first input size from the hardware offloading command, retrieve input data based on the input pointer and the total input size, separate the input data into a first data stream and a second data stream based on the first input size, and execute an offloaded operation on the first data stream and the second data stream.

    TECHNIQUES FOR CONTROLLING SIMULATION FOR HARDWARE OFFLOADING SYSTEMS
    2.
    发明公开
    TECHNIQUES FOR CONTROLLING SIMULATION FOR HARDWARE OFFLOADING SYSTEMS 审中-公开

    公开(公告)号：US20240020178A1

    公开(公告)日：2024-01-18

    申请号：US18476004

    申请日：2023-09-27

    Applicant: Lemon Inc. ,  Beijing Volcano Engine Technology Co., Ltd.

    Inventor： Shan XIAO ,  Hui ZHANG ,  Bo LI ,  Chul LEE ,  Ping ZHOU ,  Fei LIU

    IPC: G06F9/50

    CPC classification number: G06F9/5083

    Abstract: Described are examples for simulating performance of a hardware offloading system including receiving, by a simulator that corresponds to a simulated architecture representing the hardware offloading system, input data from a user application for processing by the simulated architecture, preparing, by the simulator, corresponding output data for the input data without computing the corresponding output data by the simulated architecture, and returning, by the simulator, the corresponding output data to the user application after a simulated idle time related to computing the corresponding output data by the simulated architecture.

    METHOD AND SYSTEM FOR ACCELERATION OR OFFLOADING UTILIZING A UNIFIED DATA POINTER
    6.
    发明公开
    METHOD AND SYSTEM FOR ACCELERATION OR OFFLOADING UTILIZING A UNIFIED DATA POINTER 审中-公开

    公开(公告)号：US20240036940A1

    公开(公告)日：2024-02-01

    申请号：US18485418

    申请日：2023-10-12

    Applicant: Lemon Inc. ,  Beijing Volcano Engine Technology Co., Ltd.

    Inventor： Chul LEE ,  Hui ZHANG ,  Shan XIAO ,  Bo LI ,  Ping ZHOU ,  Fei LIU

    IPC: G06F9/50

    CPC classification number: G06F9/5083 ,  G06F9/5027 ,  G06F2209/509

    Abstract: Methods, systems, and devices for performing an acceleration process by offloading an operation. The system includes a hardware offloading engine that includes a hardware accelerator for performing the acceleration process. The hardware accelerator has a processor configured to receive a hardware offloading command, the hardware offloading command including an operation code, an input pointer, and an output pointer, in which at least one of the input pointer or the output pointer includes a unified data pointer that includes one or more bits of memory for identifying the source location for the input data or the destination location of the output data, parse the operation code, the input pointer, and the output pointer from the hardware offloading command, retrieve the input data based on the input pointer, and execute an offloaded operation on the input data based on the operation code.

    MULTI-DIMENSIONAL SOLID STATE DRIVE BLOCK ACCESS
    7.
    发明公开
    MULTI-DIMENSIONAL SOLID STATE DRIVE BLOCK ACCESS 审中-公开

    公开(公告)号：US20230195345A1

    公开(公告)日：2023-06-22

    申请号：US18108351

    申请日：2023-02-10

    Applicant: Lemon Inc.

    Inventor： Chul LEE ,  Ping ZHOU ,  Hui Frankie ZHANG ,  Fei LIU

    IPC: G06F3/06

    CPC classification number: G06F3/064 ,  G06F3/0604 ,  G06F3/0655 ,  G06F3/0679

    Abstract: A method for accessing blocks of a solid state drive is described. A starting position is received, where the starting position identifies a first block of a contiguous block region within a namespace of the solid state drive. The contiguous block region is to be accessed according to a single input/output operation and the namespace comprises two dimensions of logical address space with respective indices for indexing blocks within a corresponding dimension of the logical address space. A first dimensional identifier that identifies a size of the contiguous block region in a first dimension of the namespace is received. A second dimensional identifier that identifies a size of the contiguous block region in a second dimension of the namespace is received. Blocks of the contiguous block region are accessed in response to the single input/output operation according to the starting position, the first dimensional identifier, and the second dimensional identifier.

    METHOD AND SYSTEM OF MANAGING MEMORY USAGE IN A DATA LEARNING OPERATION
    8.
    发明公开
    METHOD AND SYSTEM OF MANAGING MEMORY USAGE IN A DATA LEARNING OPERATION 审中-公开

    公开(公告)号：US20240118818A1

    公开(公告)日：2024-04-11

    申请号：US18512191

    申请日：2023-11-17

    Applicant: Lemon Inc.

    Inventor： Jin ZHOU ,  Tongping LIU ,  Yong FU ,  Ping ZHOU ,  Wei XU ,  Jianjun CHEN

    IPC: G06F3/06

    CPC classification number: G06F3/0631 ,  G06F3/0616 ,  G06F3/064 ,  G06F3/0679

    Abstract: Methods and systems for managing memory usage in data learning operations. The method includes profiling one or more objects used for training in the data learning operation, in which the profiling includes determining an object size, a memory allocation timestamp, and a memory deallocation timestamp; and scheduling the memory usage. The scheduling includes grouping the one or more objects into the one or more groups based on the memory allocation and/or the memory deallocation timestamp of the one or more objects, and arranging the one or more objects in the one or more groups in descending order in a memory space. Two or more objects are provided in the descending order in the memory space, in which one of the objects having an earliest memory allocation timestamp is provided at a first value and the other object having a later memory allocation timestamp is provided at a second value.

Patent Agency Ranking