Device data processing using shared memory pool

Patent metadata

Field	Value
Publication number	US-12561247-B2
Application number	US-202418771430-A
Country	US
Kind code	B2
Filing date	Jul 12, 2024
Priority date	Jan 14, 2022
Publication date	Feb 24, 2026
Grant date	Feb 24, 2026

Abstract

A computing device includes a central processing unit, at least one heterogeneous device, and a shared memory pool. The central processing unit is configured to: divide the shared memory pool into a plurality of memory spaces, store, in a first memory space in the plurality of memory spaces, first to-be-processed data that is provided by a client and that is associated with a service, and notify a first heterogeneous device of an address of the first to-be-processed data in the first memory space and a first operation that needs to be performed by the first heterogeneous device. The first heterogeneous device performs the first operation on the first to-be-processed data, and stores obtained first data in a second memory space.

First claim

What is claimed is:

1 . A computing device comprising:
a shared memory pool comprising shared memories; a central processing unit configured to:
divide the shared memory pool into memory spaces comprising a first memory space, a second memory space, a third memory space, and a fourth memory space;
store, in the first memory space, first to-be-processed data that is from a client and that is associated with a service;
send a first notification comprising a first address of the first to-be-processed data in the first memory space and comprising a first operation to be performed on the first to-be-processed data;
store, in the third memory space, second to-be-processed data that is from the client and that is associated with the service;
send a second notification comprising a second address of the second to-be-processed data in the third memory space, a second operation to be performed on the second to-be-processed data, and a first indication that the fourth memory space is a second storage location of second data generated through the second operation; and
send a third notification comprising a third operation to be performed on first data and the second data and comprising a second indication to send third data generated through the third operation to an external computing device; and
at least one heterogeneous device comprising:
a first heterogeneous device configured to:
receive the first notification from the central processing unit;
perform the first operation on the first to-be-processed data in the first memory space to obtain the first data; and
store the first data in the second memory space;

a second heterogeneous device configured to:
receive the second notification from the central processing unit;
perform the second operation on the second to-be-processed data in the third memory space to obtain the second data; and
store the second data in the fourth memory space; and

a third heterogeneous device configured to:
connect to the external computing device through a network;
receive the third notification from the central processing unit;
perform the third operation on the first data in the second memory space and the second data in the fourth memory space to obtain the third data; and
send the third data to the external computing device.

2 . The computing device of claim 1 , wherein the first notification further comprises an indication that the second memory space is a first storage location of the first data.

3 . The computing device of claim 1 , wherein the first heterogeneous device and the second heterogeneous device are graphics processing units (GPUs).

4 . The computing device of claim 1 , wherein the third heterogeneous device is a graphics processing unit (GPU), a neural processing unit (NPU), or a video processing unit (VPU).

5 . The computing device of claim 1 , wherein the third heterogeneous device is a network adapter.

6 . A method comprising:
dividing, by a central processing unit of a computing device, a shared memory pool of the computing device into memory spaces comprising a first memory space, a second memory space, a third memory space, and a fourth memory space; storing, by the central processing unit in the first memory space, first to-be-processed data that is from a client and that is associated with a service; sending, by the central processing unit, a first notification comprising a first address of the first to-be-processed data in the first memory space and comprising a first operation to be performed on the first to-be-processed data; receiving, by a first heterogeneous device of the computing device and from the central processing unit, the first notification; performing, by the first heterogeneous device, the first operation on the first to-be-processed data in the first memory space to obtain first data; storing, by the first heterogeneous device, the first data in the second memory space; storing, by the central processing unit in the third memory space, second to-be-processed data that is from the client and that is associated with the service; sending, by the central processing unit and to a second heterogeneous device of the computing device, a second notification comprising a second address of the second to-be-processed data in the third memory space, a second operation to be performed on the second to-be-processed data, and a first indication that the fourth memory space is a second storage location of second data generated through the second operation; receiving, by the second heterogeneous device and from the central processing unit, the second notification; performing, by the second heterogeneous device, the second operation on the second to-be-processed data in the third memory space to obtain the second data; storing the second data in the fourth memory space; providing, by the central processing unit, a third address of the second memory space and a fourth address of the fourth memory space for a third heterogeneous device of the computing device; sending, by the central processing unit and to the third heterogeneous device, a third notification comprising a third operation to be performed on the first data and the second data and comprising a second indication to send third data generated through the third operation to an external computing device; receiving, by the third heterogeneous device and from the central processing unit, the third notification; performing, by the third heterogeneous device, the third operation on the first data in the second memory space and the second data in the fourth memory space to obtain the third data; and sending, by the third heterogeneous device, the third data to the external computing device.

7 . The method of claim 6 , wherein the first notification further comprises an indication that the second memory space is a first storage location of the first data.

8 . The method of claim 6 , wherein the first heterogeneous device and the second heterogeneous device are graphics processing units (GPUs).

9 . The method of claim 6 , wherein the third heterogeneous device is a graphics processing unit (GPU), a neural processing unit (NPU), or a video processing unit (VPU).

10 . The method of claim 6 , wherein the third heterogeneous device is a network adapter.

11 . A non-transitory computer-readable storage medium storing instructions that, when executed by one or more processors, cause a computing device to:
divide, by a central processing unit of the computing device, a shared memory pool into memory spaces comprising a first memory space, a second memory space, a third memory space, and a fourth memory space; store, by the central processing unit in the first memory space, first to-be-processed data that is from a client and that is associated with a service; send, by the central processing unit, a first notification comprising an address of the first to-be-processed data in the first memory space and comprising a first operation to be performed on the first to-be-processed data; receive, by a first heterogeneous device of the computing device and from the central processing unit, the first notification; perform, by the first heterogeneous device, the first operation on the first to-be-processed data in the first memory space to obtain first data; store, by the first heterogeneous device, the first data in the second memory space; store, by the central processing unit in the third memory space, second to-be-processed data that is from the client and that is associated with the service; send, by the central processing unit and to a second heterogeneous device of the computing device, a second notification comprising a second addr

Assignees

• Huawei Cloud Computing Tech Co Ltd

Inventors

• Liu Xiao
• Yu Zhou

Classifications

• G06F2212/621

  Coherency control relating to peripheral accessing, e.g. from DMA or I/O device · CPC title

• G06F15/167

  using a common memory, e.g. mailbox · CPC title

• G06F9/5044

  considering hardware capabilities · CPC title

• G06F9/5016 Primary

  the resource being the memory · CPC title

• G06F9/54

  Interprogram communication · CPC title