Scenario

With the widespread application of AI in scenarios such as deep learning, video processing, scientific computing, and graphic visualization, the market demand for GPU instances is growing. However, the problem of excess computing power and rising costs caused by ultra-large-scale elastic computing starting out with a single physical GPU is becoming increasingly prominent. The emergence of lightweight GPUs has broken the limitations of the traditional pass-through mode. A lightweight GPU can provide finer-grained services than a physical GPU, allowing customers to conduct business more flexibly at lower costs.

Procedures

The deployment instruction for the solution - Use Lightweight GPU Instances to Deploy Cloud Games is following:

1. Check the alibaba cloud account permission, should include these services ECS, and VPC.

2. Click the Create Stack button/link, follow the wizard to specify parameters and etc.

3. Once the stack created, query ECS information through the stack output page, log in to ecS instance via SSH console, execute commands (cd /root/tmp && ./start_server.sh), and you can turn on the service side.

Please check the best practice for detailed information.

Resources

Architecture Diagram