中科微点-全球专利检索

  1. Login
  2. Sign Up

Search Results

28 records (took 0.021 seconds)

    DISTRIBUTED ADDRESS TRANSLATION IN A MULTI-NODE INTERCONNECT FABRIC
    2.
    发明申请
    DISTRIBUTED ADDRESS TRANSLATION IN A MULTI-NODE INTERCONNECT FABRIC 审中-公开

    公开(公告)号：US20200159669A1

    公开(公告)日：2020-05-21

    申请号：US16198649

    申请日：2018-11-21

    Applicant: NVIDIA Corporation

    Inventor： Samuel Hammond Duncan ,  Sanjeev Jain ,  Mark Douglas Hummel ,  Vyas Venkataraman ,  Olivier Giroux ,  Larry Robert Dennison ,  Alexander Toichi Ishii ,  Hemayet Hossain ,  Nir Haim Arad

    IPC: G06F12/1027

    Abstract: Multiprocessor clusters in a virtualized environment conventionally fail to provide memory access security, which is frequently a requirement for efficient utilization in multi-client settings. Without adequate access security, a malicious process may access what might be confidential data that belongs to a different client sharing the multiprocessor cluster. Furthermore, an inadvertent programming error in the code for one client process may accidentally corrupt data that belongs to the different client. Neither scenario is acceptable. Embodiments of the present disclosure provide access security by enabling each processing node within a multiprocessor cluster to virtualize and manage local memory access and only process access requests possessing proper access credentials. In this way, different applications executing on a multiprocessor cluster may be isolated from each other while advantageously sharing the hardware resources of the multiprocessor cluster.

    IN-NETWORK MESSAGE AGGREGATION FOR EFFICIENT SMALL MESSAGE TRANSPORT
    4.
    发明公开
    IN-NETWORK MESSAGE AGGREGATION FOR EFFICIENT SMALL MESSAGE TRANSPORT 审中-公开

    公开(公告)号：US20230327996A1

    公开(公告)日：2023-10-12

    申请号：US18149924

    申请日：2023-01-04

    Applicant: NVIDIA Corporation

    Inventor： Benjamin Klenk ,  Alan Lynn Davis ,  Larry Robert Dennison

    IPC: H04L47/2441 ,  H04L43/026

    CPC classification number: H04L47/2441 ,  H04L43/026

    Abstract: Aggregation of small payloads from multiple packets may improve bandwidth efficiency of a network, particularly a high-performance compute cluster with thousands of network endpoints and distributed data. Aggregation is context-based and a packet header is reduced because the common components that are shared by the aggregated messages are included once within the header. Execution contexts are explicitly created and destroyed by application programs. Each participating endpoint stores context-specific properties until the context is destroyed, so that the properties are not included in the header. Aggregation may be performed at different hierarchical levels by switches and/or endpoints.

    Distributed batch normalization using partial populations
    5.
    发明授权
    Distributed batch normalization using partial populations 有权

    公开(公告)号：US11341369B2

    公开(公告)日：2022-05-24

    申请号：US16669925

    申请日：2019-10-31

    Applicant: NVIDIA Corporation

    Inventor： Larry Robert Dennison ,  Benjamin Klenk

    IPC: G06K9/62 ,  G06N3/08 ,  G06F7/483 ,  G06F9/38 ,  G06N3/04

    Abstract: A technique for performing data parallel training of a neural network model is disclosed that incorporates batch normalization techniques using partial populations to generate normalization parameters. The technique involves processing, by each processor of a plurality of processors in parallel, a first portion of a sub-batch of training samples allocated to the processor to generate activations for the first portion of the sub-batch. Each processor analyzes the activations and transmits statistical measures for the first portion to an additional processor that reduces the statistical measures from multiple processors to generate normalization parameters for a partial population of the training samples that includes the first portion from each of the plurality of processors. The normalization parameters are then transmitted back to each of the processors to normalize the activations for both the first portion and a second portion of the sub-batch of training samples allocated to each processor.

    SCALABLE IN-NETWORK COMPUTATION FOR MASSIVELY-PARALLEL SHARED-MEMORY PROCESSORS
    6.
    发明申请
    SCALABLE IN-NETWORK COMPUTATION FOR MASSIVELY-PARALLEL SHARED-MEMORY PROCESSORS 有权

    公开(公告)号：US20220029845A1

    公开(公告)日：2022-01-27

    申请号：US17495547

    申请日：2021-10-06

    Applicant: NVIDIA Corporation

    Inventor： Benjamin Klenk ,  Nan Jiang ,  Larry Robert Dennison ,  Gregory M. Thorson

    IPC: H04L12/18 ,  G06F9/50 ,  H04L12/801 ,  H04L12/813 ,  H04L12/927 ,  H04L12/741 ,  H04L29/08

    Abstract: A network device configured to perform scalable, in-network computations is described. The network device is configured to process pull requests and/or push requests from a plurality of endpoints connected to the network. A collective communication primitive from a particular endpoint can be received at a network device. The collective communication primitive is associated with a multicast region of a shared global address space and is mapped to a plurality of participating endpoints. The network device is configured to perform an in-network computation based on information received from the participating endpoints before forwarding a response to the collective communication primitive back to one or more of the participating endpoints. The endpoints can inject pull requests (e.g., load commands) and/or push requests (e.g., store commands) into the network. A multicast capability enables tasks, such as a reduction operation, to be offloaded to hardware in the network device.

    Distributed batch normalization using estimates and rollback
    7.
    发明授权
    Distributed batch normalization using estimates and rollback 有权

    公开(公告)号：US11170263B2

    公开(公告)日：2021-11-09

    申请号：US16669979

    申请日：2019-10-31

    Applicant: NVIDIA Corporation

    Inventor： Larry Robert Dennison ,  Benjamin Klenk

    IPC: G06K9/62 ,  G06N3/08 ,  G06F7/483 ,  G06F9/38 ,  G06N3/04

    Abstract: A technique utilizing speculative execution and rollback for performing data parallel training of a neural network model is disclosed. Activations for a layer of the neural network model are normalized during a speculative normalization operation using estimated normalization parameters associated with a partial population of a set of training data allocated to a particular processor. Normalization parameters associated with the total population of the set of training data are generated by a distributed reduce operation in parallel with the speculative normalization operation. An optional rollback operation can revert the activations to a pre-normalization state if the estimated normalization parameters for the partial population are subsequently determined to be inaccurate compared to the normalization parameters for the population of the set of training data distributed across a plurality of processors.

    INJECTION LIMITING AND WAVE SYNCHRONIZATION FOR SCALABLE IN-NETWORK COMPUTATION
    8.
    发明申请
    INJECTION LIMITING AND WAVE SYNCHRONIZATION FOR SCALABLE IN-NETWORK COMPUTATION 有权

    公开(公告)号：US20210036881A1

    公开(公告)日：2021-02-04

    申请号：US16938044

    申请日：2020-07-24

    Applicant: NVIDIA Corporation

    Inventor： Benjamin Klenk ,  Nan Jiang ,  Larry Robert Dennison

    IPC: H04L12/18 ,  G06F9/50 ,  H04L12/927 ,  H04L12/813 ,  H04L12/801

    Abstract: A network device configured to perform scalable, in-network computations is described. The network device is configured to process pull requests and/or push requests from a plurality of endpoints connected to the network. A collective communication primitive from a particular endpoint can be received at a network device. The collective communication primitive is associated with a multicast region of a shared global address space and is mapped to a plurality of participating endpoints. The network device is configured to perform an in-network computation based on information received from the participating endpoints before forwarding a response to the collective communication primitive back to one or more of the participating endpoints. An injection policy comprising the issuing of credits enables each endpoint to limit the amount of collective communication primitives injected into the network simultaneously to reduce network congestion caused by increased network traffic due to the multicast capability of the network devices.

    Sparse convolutional neural network accelerator
    9.
    发明授权
    Sparse convolutional neural network accelerator 有权

    公开(公告)号：US10860922B2

    公开(公告)日：2020-12-08

    申请号：US16686931

    申请日：2019-11-18

    Applicant: NVIDIA Corporation

    Inventor： William J. Dally ,  Angshuman Parashar ,  Joel Springer Emer ,  Stephen William Keckler ,  Larry Robert Dennison

    IPC: G06N3/063 ,  G06F7/544 ,  G06N3/04 ,  G06F7/523 ,  G06N3/08

    Abstract: A method, computer program product, and system perform computations using a sparse convolutional neural network accelerator. A first vector comprising only non-zero weight values and first associated positions of the non-zero weight values within a 3D space is received. A second vector comprising only non-zero input activation values and second associated positions of the non-zero input activation values within a 2D space is received. The non-zero weight values are multiplied with the non-zero input activation values, within a multiplier array, to produce a third vector of products. The first associated positions are combined with the second associated positions to produce a fourth vector of positions, where each position in the fourth vector is associated with a respective product in the third vector. The products in the third vector are transmitted to adders in an accumulator array, based on the position associated with each one of the products.

    Distributed address translation in a multi-node interconnect fabric
    10.
    发明授权
    Distributed address translation in a multi-node interconnect fabric 有权

    公开(公告)号：US10769076B2

    公开(公告)日：2020-09-08

    申请号：US16198649

    申请日：2018-11-21

    Applicant: NVIDIA Corporation

    Inventor： Samuel Hammond Duncan ,  Sanjeev Jain ,  Mark Douglas Hummel ,  Vyas Venkataraman ,  Olivier Giroux ,  Larry Robert Dennison ,  Alexander Toichi Ishii ,  Hemayet Hossain ,  Nir Haim Arad

    IPC: G06F9/455 ,  G06F12/1027

    Abstract: Multiprocessor clusters in a virtualized environment conventionally fail to provide memory access security, which is frequently a requirement for efficient utilization in multi-client settings. Without adequate access security, a malicious process may access what might be confidential data that belongs to a different client sharing the multiprocessor cluster. Furthermore, an inadvertent programming error in the code for one client process may accidentally corrupt data that belongs to the different client. Neither scenario is acceptable. Embodiments of the present disclosure provide access security by enabling each processing node within a multiprocessor cluster to virtualize and manage local memory access and only process access requests possessing proper access credentials. In this way, different applications executing on a multiprocessor cluster may be isolated from each other while advantageously sharing the hardware resources of the multiprocessor cluster.

Patent Agency Ranking