Skip to content

AMD Accelerator Cloud

The AMD Accelerator Cloud (AAC) is a private hosted platform that offers users access to the latest AMD hardware resources, tools, and ready-to-use application software. This platform facilitates the rapid and cost-effective development of solutions using AMD GPUs.

On this website, you will find comprehensive documentation on how to use the AMD Accelerator Cloud.

AAC sign in

AAC dashboard

Users

The users section outlines the process of gaining access to AAC, changing or resetting a password, viewing the teams and queues assigned, and checking the list of workloads launched by the user.

User account page

User Files

The user files section describes how users can upload files if required for their workload, as well as how to download or delete these files as needed.

User files page

Applications

The applications section describes how users can create applications on AAC. This section can be referred to if the user has the Developer role. Without the developer role, applications cannot be created.

Applications page available for the user

Workloads

The workloads section describes how to run, monitor, cancel, or rerun a workload. It also explains how to check the logs and metrics associated with a workload. Additionally, it provides instructions on how to run a workload using your own Docker image.

User workloads page

Bare metal access

This section provides a step-by-step guide for users on how to connect to the different clusters available for bare metal access, using SSH and telnet clients like MobaXterm and PuTTY. This is only relevant for users who have direct SSH access, for those who have been granted this type of access and shared their public keys with the AAC team.

It also includes guides on how to use Podman, run Megaron, NanoGPT, PyTorch multinode, RCCL tests, maintain long SSH sessions with Tmux and Screen, and set up a Conda environment.

Note: this is not related to the AAC website because most of the applications configured there provide capabilities to access to their running workloads in different manners (SSH on the fly or web apps like Jupyter notebooks)

FAQs

This section includes a list of frequently asked questions (FAQs) that can serve as a useful reference for users. It provides quick answers to common queries and can help clarify any doubts or issues that may arise during usage.