公开(公告)号：US20240427629A1

公开(公告)日：2024-12-26

申请号：US18826499

申请日：2024-09-06

Applicant: Huawei Technologies Co., Ltd.

Inventor： Tao Wang

IPC: G06F9/48 ,  G06F13/38 ,  G06F15/78

Abstract: An artificial intelligence (AI) control apparatus includes an AI controller that configures N acceleration units in an AI accelerator as at least a first-type acceleration unit and a second-type acceleration unit, where a security level of the first-type acceleration unit is higher than a security level of the second-type acceleration unit. The AI controller receives a first-type task delivered by a processor in a first execution environment, and distributes the first-type task to the first-type acceleration unit. The AI controller receives a second-type task delivered by the processor in a second execution environment, and distributes the second-type task to the second-type acceleration unit. A security level of the first-type task is higher than a security level of the second-type task. Tasks of different security levels can be simultaneously run on one AI acceleration chip in order to provide hardware-level isolation for the tasks of different security levels.

公开(公告)号：US20230267099A1

公开(公告)日：2023-08-24

申请号：US18309403

申请日：2023-04-28

Applicant: Huawei Technologies Co., Ltd.

Inventor： Bintian Wang ,  Feng Han ,  Tao Wang

IPC: G06F16/176 ,  G06F11/34

CPC classification number: G06F16/176 ,  G06F11/3409

Abstract: A data management method includes obtaining a hot data set that includes hot data statistics information of at least one terminal device; performing data affinity classification on the hot data set to obtain a hot data subset corresponding to each of the at least one terminal device; and storing, into a first terminal, hot data in a first hot data subset other than hot data that has been stored in the first terminal, where the first hot data subset is a hot data subset corresponding to the first terminal.

公开(公告)号：US11294675B2

公开(公告)日：2022-04-05

申请号：US16677816

申请日：2019-11-08

Applicant: Huawei Technologies Co., Ltd.

Inventor： Jun Yao ,  Yasuhiko Nakashima ,  Tao Wang ,  Wei Zhang ,  Zuqi Liu ,  Shuzhan Bi

IPC: G06F9/312 ,  G06F9/30 ,  G06F12/0862 ,  G06F13/16

Abstract: A method for accessing a memory of a multi-core system, a related apparatus, a system, and a storage medium involve obtaining data from a system memory according to a prefetch instruction, and sending a message to a core that carries the to-be-accessed data. Each segment of data is stored in an intra-core cache based on the prefetch instruction.

公开(公告)号：US10740657B2

公开(公告)日：2020-08-11

申请号：US15964045

申请日：2018-04-26

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Jun Yao ,  Tao Wang ,  Yu Wang

IPC: G06K9/62 ,  G06T1/60 ,  G06K9/00 ,  G06T1/20 ,  G06K9/46

Abstract: An image processing device and a method performed by the device are provided. The image processing device includes processors forming a partitioning unit, processors forming a partitioning unit, and memories shared by the partitioning unit and the classification unit. The partitioning unit obtains an image from a first memory, partitions the image into one or more areas, each area includes one or more objects to be classified, and saves information of the partitioning of the image in a second memory. The classification unit obtains the image from the first memory, obtains information of partitioning of the image from the second memory and classifies the objects in each area of the image to obtain a classification result of the image. The image processing device may further include a scheduling unit for controlling optimizing performance of the partitioning unit and the classification unit.

公开(公告)号：US10372429B2

公开(公告)日：2019-08-06

申请号：US15988225

申请日：2018-05-24

Applicant: HUAWEI TECHNOLOGIES CO.,LTD.

Inventor： Tao Wang ,  Fenglong Song ,  Jun Yao

IPC: G06F8/41 ,  G06F9/448

Abstract: A method for generating an accelerator program is disclosed, to help increase utilization of an accelerator and increase program development efficiency. In some feasible implementations of the present invention, the method includes: obtaining an accelerator program description that is based on a state machine, where the accelerator program description includes multiple state machines separately configured to implement an application program, and the multiple state machines form a pipeline according to a data dependency in a directed acyclic graph DAG corresponding to the application program; and performing state machine splicing on the state machines in the accelerator program description by using an accelerator compilation tool, to generate an accelerator program.

公开(公告)号：US20180267784A1

公开(公告)日：2018-09-20

申请号：US15988225

申请日：2018-05-24

Applicant: HUAWEI TECHNOLOGIES CO.,LTD.

Inventor： Tao Wang ,  Fenglong Song ,  Jun Yao

IPC: G06F8/41 ,  G06F9/448

CPC classification number: G06F8/433 ,  G06F8/42 ,  G06F8/4452 ,  G06F9/4498

Abstract: A method for generating an accelerator program is disclosed, to help increase utilization of an accelerator and increase program development efficiency. In some feasible implementations of the present invention, the method includes: obtaining an accelerator program description that is based on a state machine, where the accelerator program description includes multiple state machines separately configured to implement an application program, and the multiple state machines form a pipeline according to a data dependency in a directed acyclic graph DAG corresponding to the application program; and performing state machine splicing on the state machines in the accelerator program description by using an accelerator compilation tool, to generate an accelerator program.