Getting Started¶

Access¶

Access is provisioned by Lucid Computing — reach out to andriy@lucidcomputing.ai.
Once approved, you will receive a username.

Logging in¶

Access to the VCF Cluster is provided via Secure Shell (SSH) login. Most Unix-like operating systems provide an SSH client by default that can be accessed by typing the ssh command in a terminal window.

To login, open a terminal and type the following command, where should be replaced with the username you were provided with:

$ ssh -i ~/.ssh/id_ed25519 <username>@vcf.lucidcomputing.ai
$ <username>@vcf-login-0:~$

Upon logging in, you will be connected to a login node (e.g., vcf-login-0).

If you are having trouble logging in check out our section on SSH.

Login nodes are not for computing

Login nodes are shared among many users and must not be used for computationally intensive tasks. Instead, submit those tasks to the scheduler, which will dispatch them to the compute nodes.

Getting compute¶

Get an interactive session on a node with a single GPU and two CPU cores (the default ratio is 2 CPUs per GPU):

$ <username>@vcf-login-0:~$ srun --gpus=1 --pty /bin/bash
$ <username>@vcf-node-0:~$

Run nvidia-smi to confirm which GPU(s) have been assigned to your session:

$ <username>@vcf-node-0:~$ nvidia-smi

How-To Guides

After gaining access to the cluster, you can familiarize yourself with it using our How-To Guides. They include information on connecting to the cluster, running jobs, general cluster usage, and more.