Create VM Instance
This guide assumes you already have an AWS account setup.
- In the AWS console, Go to Compute -> EC2
- Click Launch an Instance
- Choose Ubuntu as the Operating System
- Select a GPU instance type. I recommend g5.xlarge, because it’s very fast, but you can choose g4dn.xlarge for a cheaper option.
- Select your Key pair for SSH login. If you don’t have one yet, you should create one with the default settings, and download the private key to a secure location on your computer.
- Create a security group, or select an existing one. If you are creating a new one, you can simply check all 3 options (allow SSH, allow HTTPS, and allow HTTP). If you are selecting an existing policy, make sure port 7860 is exposed.
- Set your storage to at least 50 GB of gp3 storage (I recommend making it 100 GB to allow space to install more models and extensions).
- Click Launch Instance
Connect to your instance
When your instance is started, click Connect. You can use either EC2 Instance Connect or SSH client to connect to your instance.
If you use EC2 Instance Connect, you will launch a command prompt right in the browser. This is the easiest way.
If you use SSH client, you can follow the provided directions, using the Key Pair private key you downloaded earlier.
Setup your instance
Using the command line, we need to install GPU drivers and python libraries before we can run Automatic1111. It is recommended to run all these commands from your home directory.
Install GPU Driver
wget https://raw.githubusercontent.com/GoogleCloudPlatform/compute-gpu-installation/main/linux/install_gpu_driver.py
sudo python3 install_gpu_driver.py
Install Cuda (Nvidia GPU Software)
wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-keyring_1.0-1_all.deb
sudo dpkg -i cuda-keyring_1.0-1_all.deb
sudo apt-get update
sudo apt-get -y install cuda
There is no need to restart services, you can cancel any prompts that appear.
Install python libraries
sudo apt-get update
sudo apt -y install python3-venv python3-pip
sudo apt-get install -y python3-opencv
sudo apt install --no-install-recommends -y google-perftools
There is no need to restart services, you can cancel any prompts that appear.
Download the Automatic1111 repo
git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
Running Automatic1111
stable-diffusion-webui/webui.sh --listen --xformers --api --enable-insecure-extension-access --skip-torch-cuda-test
It will take a few minutes to start the first time.
To get to the Automatic1111 app, you will need to find the Public IPv4 Address of the server. You will find this on the EC2 instance summary page.
Go to your browser, and enter your IP Address with the 7860 port: http://YOUR_IP:7860 and you should see the Automatic1111 UI!
Automatic Startup
If you want to Automatic1111 to startup automatically when your server starts, you can add it to your cron file:
sudo apt-get install nano cron
crontab -e
Add the following to the end of the file that opens in nano:
@reboot ~/stable-diffusion-webui/webui.sh --listen --xformers --api --enable-insecure-extension-access --skip-torch-cuda-test
Save the file, and now Automatic1111 should start automatically the next time you start your instance.
Starting and Stopping the VM
You can start and stop your instance via the instance summary or instance list pages. When your server is stopped, you only pay for the persistent hard drive, while the actual VM incurs costs only when it is running. So don’t forget to stop your instance when you are finished using it.
Spot Provisioning
Spot provisioning is an option that gives you cheaper pricing on instances. But you can only run a spot server when AWS has enough resources available, and it gives AWS the right to shut down your server if they need resources for something else.
Spot provisioning in Google Cloud worked great for me during off-peak hours, but I didn’t have as good of an experience with AWS. So in AWS I tend to run normal provisioning.
Here are some tips to avoid getting shutdown when running spot provisioning:
Run your server during non-business hours. For example, if you create your instance in your region, run your server at night or on weekends.
Create instances in multiple regions. Create another instance in a region on the other side of the globe. That way you can run that instance during the day (when it is night in your server’s region). Note that the requests will take longer in a distant region.
The Automatic1111 API
If you want to build your own UI, or run bulk generating of images, the API is very helpful. Because we added the --api parameter when we started Automatic1111, the api will be enabled.
You can access the API at: HOST:7860/sdapi/v1/txt2img