Guide to selecting the right serverless GPU solution based on business requirements

Traditional GPU deployments suffer from low resource utilization, high fixed costs, and limited ability to scale. Function Compute's serverless GPU lets you select GPU models and computing resources that match your actual workload. This topic helps you match your inference workload to the right Function Compute GPU configuration.

How to choose

The following flowchart shows the decision process for selecting a serverless GPU solution.

What's next

After selecting a scenario, see the corresponding guide for configuration details:

• Quasi-real-time inference scenarios

• Real-time inference scenarios

• Offline asynchronous task scenarios