GPU Virtual Resource Segmentation for Multi-Task Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing resource allocation methods for GPUs are inefficient, as they can only allocate computing resources and encoding-decoding resources to the same task at the same time, leading to low resource utilization and reduced multi-task execution efficiency.
Innovation Solution
The proposed method involves acquiring a target resource type corresponding to a target task, determining the specific resource type from the available resource types on a GPU, and allocating the corresponding virtual processor resources to the target task, allowing multiple tasks to use the same GPU in parallel.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If computing resources and encoding-decoding resources of one physical GPU are allocated to the same task at the same time, then task execution is simplified, but resource utilization rate decreases
Solution Approach 1:
The patent segments the physical GPU resources into multiple virtual processor resources with different resource types (computing resources and encoding-decoding resources). This segmentation allows different virtual processor resources to be allocated to different tasks simultaneously, resolving the contradiction by enabling multi-task parallel execution while maintaining resource type specificity.
Solution Approach 2:
The patent creates virtual processor resources that can be universally allocated to different task types. The same physical GPU can serve multiple functions by virtualizing its resources, allowing both computing tasks and encoding-decoding tasks to execute concurrently on the same hardware platform.
2Device complexity
If computing resources and encoding-decoding resources are shared by the same task, then resource management is simplified, but multi-task execution efficiency decreases
Solution Approach 1:
The patent introduces virtual processor resources as an intermediary layer between the physical GPU and tasks. This intermediary enables complex resource allocation (different resource types to different tasks) while maintaining simplified management through virtualization abstraction, thus improving multi-task execution efficiency without proportionally increasing management complexity.
Solution Approach 2:
The patent adds a virtualization dimension to the resource allocation model. By introducing virtual processor resources as an intermediate layer, it enables multi-dimensional resource allocation where different resource types can be assigned to different tasks simultaneously, thereby improving multi-task execution efficiency.
3Measurement precision
If virtual processor resources are segmented into different resource types, then resource allocation precision improves, but device complexity increases
Solution Approach 1:
The patent segments virtual processor resources into different resource types (computing and encoding-decoding) to achieve precise allocation matching task requirements. This segmentation improves resource allocation precision by ensuring tasks receive the specific resource type they need, while the virtualization layer manages the complexity of this segmented structure.
Data Source
AI summary
The present disclosure provides a resource allocation method and apparatus, a readable medium and an electronic device. The method includes: acquiring a target resource type corresponding to a target task; acquiring a plurality of resource types of a target processor, wherein each resource type corresponds to one or more virtual processor resources of the target processor; determining a specified resource type that is identical to the target resource type from the plurality of resource types; determining a target processor resource from one or more virtual processor resources corresponding to the specified resource type; and allocating the target processor resource to the target task.