Why Use GPUs?

Training and evaluating deep neural networks is a computationally intensive task. For modestly sized problems and datasets it may be possible to train your network on the CPU in your local machine, but it could take anywhere from 15 minutes to several hours depending on the number of epochs, the size of the neural networks, and other factors. A faster alternative is to train on a GPU (Graphics Processing Unit), which is a type of processor that supports greater parallelism.

Cloud Service Provider

If you do not already have a computer with a built-in NVIDIA GPU, we suggest you use a cloud service provider. In the next few sections, we’ll go over the step you need to take to get a neural network running on the cloud using both AWS and GCE. You can choose whichever service you prefer; you can read about the capabilities of Google Cloud Platform and Amazon Elastic Compute Cloud at the provided links.

We recommend starting with Google Cloud and moving to AWS should you need more time or find that you prefer its interface.

Note: Please skip this section if you are planning to use your own GPU or CPU (or otherwise not planning to use a cloud platform).

Outline

In this lesson, we'll go through the typical steps to setup, use, and shut down a GPU instance.

1. Setting up a Google Cloud Account
2. Defining and requesting a GPU instance
3. Launching an Instance
4. Logging into an Instance
5. Shutting Down an Instance

This last step is important to ensure that you do not accrue additional charges.