This Graphcloud Quick Start describes how to run a simple PopART application from the Graphcore tutorials repository and is one of a set of quick starts for Graphcloud.
There are two ways you can run applications:
directly on the system using the setup described in this document
in a Poplar Docker container that has already been setup. Refer to Using IPUs from Docker for more information.
1.1. Graphcloud Quick Starts
The full set of Graphcloud Quick Starts leads you from learning about Graphcloud and how to access it, to running simple applications to monitoring the running programs and hardware.
For brevity, we refer to the Poplar Graph Programming Framework as “Poplar” and specifically to the “Poplar SDK” when we refer to the SDK.
1.1.1. About working on Graphcloud
- About Graphcloud Quick Start: An overview of Graphcloud.
Describes how the hardware is configured, the storage that is available and what software and datasets you can access. Also, lists what you need to be familiar with before you use Graphcloud.
- Graphcloud Access Quick Start: How to access Graphcloud.
Describes how you create the necessary SSH keys to access Graphcloud and how to log into Graphcloud.
- Graphcloud Setup Quick Start: General setup steps.
Performing these setup steps makes working on Graphcloud easier. Includes how to use the features of the Poplar SDK that is installed on Graphcloud.
1.1.2. Running simple applications
These applications are from the Graphcore tutorials repository.
- Graphcloud Jupyter Notebook Quick Start
Run an application in a Jupyter Notebook.
- Graphcloud Poplar Graph Programming Framework Quick Start
Run an application directly in the Poplar Graph Programming Framework.
- Graphcloud PyTorch Quick Start
Run a PyTorch application.
- Graphcloud PopART Quick Start
Run a PopART application.
- Graphcloud TensorFlow 1 Quick Start
Run a TensorFlow 1 application.
- Graphcloud TensorFlow 2 Quick Start
Run a TensorFlow 2 application.