Computing Services Provided by EMC Hub

GPU-based Virtual Host Services

Usage

Users can select GPU models on EMC Hub and create their own virtual hosts utilizing the Docker images of Pytorch and TensorFlow base environments provided by EMC Hub. After obtaining the SSH link, users can log in and use the virtual hosts, typically for AI training, research, and testing purposes.

Steps

• Select the appropriate GPU model and quantity.

• Create a virtual host using the base environment image provided by EMC Hub.

• Obtain the SSH link and log in to begin usage.

GPU-based Container Cloud Services

Usage

Users can select GPU models and quantities on EMC Hub and specify their own Docker images to create computing plans. Docker images can be pulled by computing nodes from specified DockerHub paths or uploaded to a dedicated server. The computing cloud service will screen computing network nodes based on the user's computing plan and ensure that the user's Docker image successfully runs the desired number of GPU container instances. This service is typically suitable for AI inference computing needs.

Steps

• Select the appropriate GPU model and quantity.

• Specify the path for your Docker image or upload it to a dedicated server.

• Create a computing plan, and EMC Hub will automatically screen computing nodes and run the instances.

• Monitoring services will oversee the computing plan in real-time to ensure that the number of instances meets expectations.