GPU Virtualization
Last updated:2025-05-12 17:48:10
Overview
Tencent Cloud TI Platform provides GPU virtualization functionality, which can allocate the computing power of the same GPU card to different training tasks and inference services for use, enhancing resource allocation flexibility and usage efficiency.
Supported GPU Models
The following GPU types are supported by the platform's GPU virtualization functionality:
T4... (Other types will be launched on the international site successively. Stay tuned.)
Feature Usage Instructions
Prerequisites
You need to create a resource group in advance on the Resource Group Management page and add your CVM to the resource group in the form of a node.
Usage Method
When creating a development machine instance, task-based modeling training task, or online service, after selecting your resource group, if it includes GPU types that support virtualization, you can select a GPU quantity between 0.1 and 1 when configuring resources.
Notes
Since GPU types on the market are constantly increasing and related drivers are continuously updated, the platform's GPU virtualization functionality for GPU types is gradually expanding. The platform's newly supported GPUs for virtualization may not be able to use the virtualization feature in existing resource groups. Please pay attention to the platform's related prompts when using.
If nodes with GPU types not yet supported for virtualization by the platform are added when creating a resource group, the virtualization functionality of the above-mentioned GPUs in the resource group still cannot be used after the platform subsequently supports virtualization for that model.
In the above situation, you also cannot directly move the above nodes to other resource groups; however, you can remove the nodes from the resource group and re-add them to a newly created resource group, so that you can use the virtualization feature.
Was this page helpful?
You can also Contact Sales or Submit a Ticket for help.
Yes
No
Feedback