METHOD, APPARATUS, AND COMPUTER PROGRAM PRODUCT FOR PROCESSING COMPUTING TASK

• US 20190324809A1
• Filed: 04/12/2019
• Published: 10/24/2019
• Est. Priority Date: 04/20/2018
• Status: Active Grant

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Inventors: Zhao, Junping, Peng, Layne Lin, Ying, Zhi, Wang, Kun
• Granted Patent: US 11,249,811 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 2209/5011 Pool
• CPC Class Codes: G06F 9/3877 using a slave processor, e.g. coprocessor peripheral processor G06F13/12; vector processor G06F15/8053
• CPC Class Codes: G06F 9/4881 Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
• CPC Class Codes: G06F 9/5005 to service a request
• CPC Class Codes: G06F 9/505 considering the load
• CPC Class Codes: G06N 3/04 Architecture, e.g. interconnection topology
• CPC Class Codes: G06N 3/08 Learning methods

Abstract

Implementations of the present disclosure relate to a method, apparatus and computer program product for processing a computing task. The method comprises: obtaining status information of multiple computing resources; in response to receiving a neural network model-based computing task, determining configuration information of multiple layers associated with the neural network model; obtaining parameter data associated with at least one part of the multiple layers on the basis of the configuration information; and based on the status information and the parameter data, selecting from the multiple computing resources a group of computing resources for processing the computing task. According to the example implementations of the present disclosure, multiple computing resources may be utilized sufficiently, and it may be guaranteed that a load balance may be stricken between the multiple computing resources.

Claims

1. A method for processing a computing task, comprising:

obtaining status information of multiple computing resources;

in response to receiving a neural network model-based computing task, determining configuration information of multiple layers associated with the neural network model;

obtaining parameter data associated with at least one part of the multiple layers on the basis of the configuration information; and

based on the status information and the parameter data, selecting from the multiple computing resources a group of computing resources for processing the computing task.

2. The method of claim 1, wherein selecting a group of computing resources on the basis of the status information comprises:

regarding a first layer among the at least one part of the multiple layers,

determining a first size of parameters associated with the first layer on the basis of the parameter data;

based on the status information, selecting a first computing resource matching the first size from the multiple computing resources to process the parameters associated with the first layer.

3. The method of claim 2, wherein selecting a group of computing resources on the basis of the status information comprises:

determining corresponding sizes of parameters associated with the at least one part of the multiple layers on the basis of the parameter data;

ranking the at least one part of the multiple layers on the basis of the corresponding sizes; and

selecting the first layer on the basis of the rankings.

4. The method of claim 2, further comprising:

determining a first resource allocation required for processing the parameters associated with the first layer; and

updating the status information on the basis of the first resource allocation.

5. The method of claim 4, further comprising:

regarding a second layer among the at least one part of the multiple layers,

determining a second size of parameters associated with the second layer on the basis of the parameter data; and

based on the updated status information, selecting a second computing resource matching the second size from the multiple computing resources to process the parameters associated with the second layer.

6. The method of claim 5, further comprising:

determining a number of computing resources required for processing the at least one part of the multiple layers; and

in response to determining that the selected group of computing resources reach the number, selecting a computing resource from the selected group of computing resources to process parameters associated with a third layer among the at least one part of the multiple layers.

7. The method of claim 1, wherein determining status information of the multiple computing resources comprises:

monitoring resource information of the multiple computing resources; and

determining status information of the multiple computing resources on the basis of the resource information.

8. The method of claim 7, wherein the resource information of the multiple computing resources comprises at least any one of the following indicators:

processing capacity information, memory resource information and bandwidth resource information of the multiple computing resources.

9. The method of claim 8, wherein determining the status information of the multiple computing resources on the basis of the resource information further comprises:

regarding a given computing resource among the multiple computing resources,

determining an importance of a corresponding indicator among the indicators for the given computing resource on the basis of the computing task; and

determining status information of the given computing resource on the basis of the corresponding indicator and the importance of the corresponding indicator.

10. The method of claim 1, wherein the multiple computing resources are multiple graphics processing units.

11. An apparatus for processing a computing task, comprising:

at least one processor;

a volatile memory; and

a memory coupled to the at least one processor and having instructions stored thereon, the instructions, when executed by the at least one processor, causing the apparatus to perform acts comprising;

obtaining status information of multiple computing resources;

in response to receiving a neural network model-based computing task, determining configuration information of multiple layers associated with the neural network model;

obtaining parameter data associated with at least one part of the multiple layers on the basis of the configuration information; and

based on the status information and the parameter data, selecting from the multiple computing resources a group of computing resources for processing the computing task.

12. The apparatus of claim 11, wherein selecting a group of computing resources on the basis of the status information comprises:

regarding a first layer among the at least one part of the multiple layers,

determining a first size of parameters associated with the first layer on the basis of the parameter data;

based on the status information, selecting a first computing resource matching the first size from the multiple computing resources to process the parameters associated with the first layer.

13. The apparatus of claim 12, wherein selecting a group of computing resources on the basis of the status information comprises:

determining corresponding sizes of parameters associated with the at least one part of the multiple layers on the basis of the parameter data;

ranking the at least one part of the multiple layers on the basis of the corresponding sizes; and

selecting the first layer on the basis of the rankings.

14. The apparatus of claim 12, wherein the acts further comprise:

determining a first resource allocation required for processing the parameters associated with the first layer; and

updating the status information on the basis of the first resource allocation.

15. The apparatus of claim 14, wherein the acts further comprise:

regarding a second layer among the at least one part of the multiple layers,

determining a second size of parameters associated with the second layer on the basis of the parameter data; and

based on the updated status information, selecting a second computing resource matching the second size from the multiple computing resources to process the parameters associated with the second layer.

16. The apparatus of claim 15, wherein the acts further comprise:

determining a number of computing resources required for processing the at least one part of the multiple layers; and

in response to determining that the selected group of computing resources reach the number, selecting a computing resource from the selected group of computing resources to process parameters associated with a third layer among the at least one part of the multiple layers.

17. The apparatus of claim 11, wherein determining status information of the multiple computing resources comprises:

monitoring resource information of the multiple computing resources; and

determining status information of the multiple computing resources on the basis of the resource information.

18. The apparatus of claim 17, wherein the resource information of the multiple computing resources comprises at least any one of the following indicators:

processing capacity information, memory resource information and bandwidth resource information of the multiple computing resources; and

wherein determining the status information of the multiple computing resources on the basis of the resource information further comprises;

regarding a given computing resource among the multiple computing resources,

determining an importance of a corresponding indicator among the indicators for the given computing resource on the basis of the computing task; and

determining status information of the given computing resource on the basis of the corresponding indicator and the importance of the corresponding indicator.

19. The apparatus of claim 11, wherein the multiple computing resources are multiple graphics processing units.

20. A computer program product, tangibly stored on a computer readable medium and comprising machine executable instructions which, when executed, cause a machine to implement a method comprising:

obtaining status information of multiple computing resources;

in response to receiving a neural network model-based computing task, determining configuration information of multiple layers associated with the neural network model;

obtaining parameter data associated with at least one part of the multiple layers on the basis of the configuration information; and

based on the status information and the parameter data, selecting from the multiple computing resources a group of computing resources for processing the computing task.