Guide to running InvokeAI on AWS

I needed to get Invoke AI running on an AWS server for a new project I am involved with. Here are my notes on how I did that.

In this article you will find out:

• Which EC2 instance to use
• How to prepare it for working with an image AI application like InvokeAI
• How to install and run InvokeAI
• Configuring a GPU monitor and sending the data to CloudWatch
• Configuring an Nginx Reverse Proxy to works with InvokeAI
• Getting an SSL certificate
• Setting up an Alarm in CoudWatch to turn the machine off if you’re not using it

I am going to assume some familiarity with the madness that is AWS and Security Groups and all the rest.

Let’s get started!

Which EC2 instance to use

Create Instance

1. Instance Type: Choose the g4dn.2xlarge for optimal performance with GPU support. You can get away with a plain g4dn.large but you won’t be able to run models like SD XL.
2. Storage: Allocate a 200GB hard disk to ensure enough space for models and other data.