公开(公告)号：US11243701B2

公开(公告)日：2022-02-08

申请号：US16914375

申请日：2020-06-28

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Dengben Wu ,  Xiaoxin Xu ,  Junjie Wang

IPC: G06F3/06

Abstract: The present invention provides a data write method and a solid-state drive array. The solid-state drive array is based on a RAID system and includes n solid-state drives. Before to-be-written data is written into the solid-state drive array, the to-be-written data is divided into n data blocks that are in a one-to-one correspondence with the n solid-state drives. After the n data blocks are all stored into the corresponding solid-state drives, FTL update of the n data blocks is performed, to complete write of the to-be-written data. If the solid-state drive array is powered off during storage of the n data blocks, because FTL update of the data blocks is not performed, all the n data blocks fail to be written into the solid-state drive array, thereby ensuring atomicity of write operations of the n data blocks.

公开(公告)号：US20240273364A1

公开(公告)日：2024-08-15

申请号：US18586400

申请日：2024-02-23

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Rongrong Fu ,  Xiaoxin Xu ,  Quanchong Huang

IPC: G06N3/08

CPC classification number: G06N3/08

Abstract: In a data processing method, when performing iterative training on a first neural network deployed on a first edge device, a cloud-side device obtains a first training set based on an application scenario feature of data collected by the first edge device. The first training set includes the data of the first edge device and sample data associated with the application scenario feature. The cloud-side device obtains a trained first neural network based on the first training set and the first neural network deployed on the first edge device. The cloud-side device then deploys the trained first neural network on the first edge device.

公开(公告)号：US20190037043A1

公开(公告)日：2019-01-31

申请号：US16133179

申请日：2018-09-17

Applicant: Huawei Technologies Co., Ltd.

Inventor： Xiaoxin Xu ,  Ligang Chen ,  Yixiang Liao

IPC: H04L29/08 ,  G06F9/4401 ,  G06F9/455

Abstract: A data prefetching method and a prefetching apparatus to improve system service performance, where the data prefetching method includes a first prefetching apparatus receives a data prefetching instruction from a first host. The first prefetching apparatus determines a target data block based on the data prefetching instruction, and obtains identification of a target prefetching apparatus from a second prefetching apparatus. Then the first prefetching apparatus determines a target storage location of the target data block based on the identification, and obtains the target data block based on the target storage location.

公开(公告)号：US20220334798A1

公开(公告)日：2022-10-20

申请号：US17855555

申请日：2022-06-30

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Tengyi LIN ,  Qiuping Pan ,  Shengyu Shen ,  Xiaoxin Xu ,  Wei Zhang

IPC: G06F7/487 ,  G06F7/485

Abstract: This application discloses a floating-point number multiplication computation method, an apparatus, and an arithmetic logic unit. The method includes: obtaining a plurality of to-be-computed first-precision floating-point numbers; decomposing each to-be-computed first-precision floating-point number to obtain at least two second-precision floating-point numbers, a second precision of the second-precision floating-point number is lower than a first precision of the first-precision floating-point number; determining various combinations including two second-precision floating-point numbers obtained by decomposing different first-precision floating-point numbers; inputting the second-precision floating-point numbers in each combination into a second-precision multiplier to obtain an intermediate computation result corresponding to each combination; and determining a computation result for the plurality of to-be-computed first-precision floating-point numbers based on the intermediate computation result corresponding to each combination.

公开(公告)号：US20200326855A1

公开(公告)日：2020-10-15

申请号：US16914375

申请日：2020-06-28

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Dengben Wu ,  Xiaoxin Xu ,  Junjie Wang

IPC: G06F3/06

Abstract: The present invention provides a data write method and a solid-state drive array. The solid-state drive array is based on a RAID system and includes n solid-state drives. Before to-be-written data is written into the solid-state drive array, the to-be-written data is divided into n data blocks that are in a one-to-one correspondence with the n solid-state drives. After the n data blocks are all stored into the corresponding solid-state drives, FTL update of the n data blocks is performed, to complete write of the to-be-written data. If the solid-state drive array is powered off during storage of the n data blocks, because FTL update of the data blocks is not performed, all the n data blocks fail to be written into the solid-state drive array, thereby ensuring atomicity of write operations of the n data blocks.

公开(公告)号：US20230162008A1

公开(公告)日：2023-05-25

申请号：US18150907

申请日：2023-01-06

Applicant: Huawei Technologies Co., Ltd.

Inventor： Yongzhong Wang ,  Fuchun Wei ,  Wei Zhang ,  Xiaoxin Xu

IPC: G06N3/0464 ,  G06N3/091

CPC classification number: G06N3/0464 ,  G06N3/091 ,  G06V10/82

Abstract: A data processing method includes obtaining first data and second data, where the first data and the second data are adjacent sequence data, and a sequence of the first data is prior to a sequence of the second data; padding third data between the first data and the second data according to a preset rule to obtain fourth data, where the third data isolates the first data from the second data; and completing data processing on the fourth data using a convolutional neural network.