iHARP at UMBC held a High Performance Computing (HPC) bootcamp in November 2024, below are the resources that were developed. Please note some access to materials and high performance resources are restricted to UMBC members only.
Introduction
The video and document (access restricted to UMBC) introduces basic BASH commands for navigating the Linux terminal, including file navigation and manipulation. It then walks users through the process of accessing Ada, the university’s High-Performance Computing (HPC) cluster, via CLI tools. This section provides instructions for logging in, navigating the cluster environment, and accessing resources.
The resources covers job management with SLURM, teaching users how to write, submit, and track SLURM jobs on Ada. This includes creating job scripts, submitting tasks to the cluster, tracking job progress, and optimizing resource usage to ensure that computational workloads run smoothly and efficiently on the HPC cluster.
Instructional Introduction Video on UMBC’s High Performance Computing Ada Cluster
The bootcamp recording includes a voice over walk-through of UMBC’s Ada cluster.
Video Chapters
00:00 UMBC HPC Bootcamp 2024
00:25 Let’s log in to ADA
01:10 High Performance Computing
01:15 What is HPC?
02:10 Purpose of HPC?
03:04 Nodes in ADA
03:16 Login Node
04:13 Worker Nodes
05:37 ADA Storage System
05:55 Home directory vs Working directory
07:16 Home Directory
08:16 Shared Group Storage
09:13 HPCF Research Storage
11:22 Scratch Space
12:49 List of Commands
13:04 ls
14:23 chmod
14:44 mkdir
15:11 cd
15:41 vim
17:55 grep
18:56 cat
19:34 pwd
21:46 Running the Bash Script
22:43 mv
23:53 cp
25:10 Environment Setup
25:28 Bash Environment
27:33 Why we need modules?
28:51 Solution
30:08 Managing Modules
32:38 Python Environment
34:32 Venv Environment
37:37 Conda Environment
42:00 SLURM
43:21 Requesting Resources in ADA
44:36 Requesting Memory
45:21 Requesting GPUs
46:21 Requesting Time
47:48 Batch Jobs
49:05 Running MNIST script through sbatch
50:14 Installing packages
50:53 How to create a shell file to request resources
57:59 Jupyter Notebooks
1:06:21 Monitoring SLURM Jobs
1:06:27 Monitoring the status of the running jobs
1:06:57 SQUEUE Command
1:07:27 Monitoring Jobs
1:08:39 Case Study: Using patch -CNN in High -Performance Computing…
1:09:29 Overview
1:10:21 Problem Statement
1:11:05 Motivation
1:11:53 Model Explanation
1:13:13 Research Objectives
1:13:59 Methodology Overview
1:14:31 Role of HPC
1:15:23 Challenges
1:16:13 Results
1:16:49 Conclusion
1:17:45 Methodology Walkthrough
1:20:40 Cancelling jobs
UMBC High Performance Computing (HPC) Bootcamp Walk-through Documentation