Launch GPUs through the Shadeform Console

1. Creating an account

Get started using Shadeform by going to Shadeform.ai and clicking “Sign up for free”. Once you sign up, you’ll be prompted to enter your first name, last name, and organization info.

2. Finding the right GPU

Once you have created an account and filled out the first time login form, you can go to the Marketplace to look at all GPUs available to launch. The drop downs in the top right corner allows you to filter by GPU type, number of GPUs in an instance, and the interconnect type.

3. Launch your machine

After finding the GPU instance that you need, click on “Launch” to open the launch sidebar. On the sidebar, you can see additional information about the instance, select the operating system, name the instance, and confirm the launch.

4. Add funds to your Shadeform Wallet

If this is the first time launching an instance, you’ll be prompted to load up your Shadeform Wallet displayed in the top right corner. By following the flow, you can add balance to your wallet via our Stripe integration. Additionally, you can configure Auto Top Up to automatically charge your payment method when your balance falls below the configured threshold.

5. Access your machine

Once your Shadeform Wallet is funded, and the machine is launched, the machine will take some time to spin up. You can see the status of your machine on the “Running Instances” tab on the Shadeform sidebar. The machine may remain in the pending state for up to 10 minutes depending on the provider. Once the machine moves to the “Active” state, you can click on the machine to open up the info sidebar to get your SSH string to access the machine.

In order to use the SSH string, you must first download your Shadeform private key. You can do so by clicking the “Download Private Key” button on the “Running Instances” page. After downloading the key, you must change the permissions of the private key by running “chmod 600 private_key.pem”. After changing the private key permissions, you can now use the Shadeform SSH string to access the machine.

chmod 600 private_key.pem
ssh -i private_key.pem shadeform@<ssh_string>

6. Delete your machine

Using the Shadeform API

Shadeform provides an API that allows you to query for available GPUs, launch instances, get instance information, and programmatically access instances.

In this guide, we’ll discuss a common usage pattern for programmatically finding available GPUs and launching them.

1. Generating an API Key

Once you have created a Shadeform account, you can generate an API Key by going to the API Keys page. Click on “Generate” to create a new API key.

2. Add funds to your Shadeform Wallet

To ensure that instances can be launched, you must first add funds to your Shadeform Wallet on the billing page.

3. Get all cloud providers and region availability and price

You can get a list of all available machines by calling the Types endpoint. This will return a list of all available machines, their price, and availability in each region.

curl --request GET \
--url https://api.shadeform.ai/v1/instances/types \
--header 'X-API-KEY: <x-api-key>'

4. Select your machine

Parse through the output and find your desired, available GPU instance.

5. Call the Create endpoint with the information returned from the availability API

curl --request POST \
--url https://api.shadeform.ai/v1/instances/create \
--header 'Content-Type: application/json' \
--header 'X-API-KEY: <x-api-key>' \
--data '{
    "cloud": "aws",
    "name": "cool-gpu-server",
    "region": "us-east-1",
    "shade_cloud": true,
    "shade_instance_type": "A100x8"
}'

6. Get your instance status

You can call this API in a loop until the status turns from pending to active. Once the instance turns to active, you can then use the ssh_user and IP to access the machine programmatically. Your private key for the instance can be downloaded from the Shadeform UI on the Running Instances page.

 curl --request GET \
--url https://api.shadeform.ai/v1/instances \
--header 'X-API-KEY: <x-api-key>'

7. Delete your instance

Once you are done with the instance, you can delete the instance by calling the Delete endpoint. The instance_id in the delete endpoint URL is the value returned as “id” by the Instances endpoint.

curl --request GET \
--url https://api.shadeform.ai/v1/instances/<instance_id>/delete \
--header 'X-API-KEY: <x-api-key>'