Taylor Geospatial Institute Regional AI Learning System (TGI RAILS) Documentation

Introduction

TGI RAILS is a high-performance computing and data analysis system, funded by the National Science Foundation and the Taylor Geospatial Institute and managed and operated at the National Center for Supercomputing Applications at the University of Illinois. The RAILS system is available for use by any of the academic institutions that make up the Taylor Geospatial Institute. RAILS is also accessible via the Open Science Grid.

RAILS provides a highly capable GPU-focused compute environment for GPU and CPU workloads coupled with a VAST filesystem, providing high-speed access to large datasets. The system is designed to support a wide range of geospatial research and data analysis tasks, including machine learning, deep learning, and other AI-related workloads.

Who is eligible to use TGI RAILS?

Geospatial related researchers, faculty, and staff at each of the TGI member organizations are eligible to use RAILS for geospatial-related research. The system can also be used in courses of any discipline. Additionally up to 20% of RAILS capability is accessible via the Open Science Grid..

Getting Started with High-Performance Computing

There are no specific prerequisite courses or high-performance computing (HPC) experience required before using Delta. However, if you are unfamiliar with using an HPC cluster, it is highly recommended that you take NCSA’s short tutorial Using an HPC Cluster for Scientific Applications before continuing.

Browse NCSA’s HPC-Moodle for a full list of HPC training opportunities including self-paced tutorials and training events (in-person and virtual).

• TGI RAILS Status Updates
• Getting Help

User Guide

• System Architecture
  • Compute Nodes
  • Network
  • Storage (File Systems)
• Allocation and Account Administration
  • Who is eligible to use TGI RAILS?
  • How to get started on RAILS? Requesting an Allocation
  • Adding Users to Your Allocation
  • NCSA Account Managment
  • Allocation Policies
• Accessing The System
  • Connecting with SSH
  • Connecting with Open OnDemand
  • Troubleshooting and Connection Help
• Good Cluster Citizenship
  • Acceptable Use Policies
• Data Management
  • File Systems
  • Transferring Data
• Accessing and Transferring Files
  • Transferring a Few Small Files
  • Transferring Many or Large Files With Globus
  • Sharing Files with Collaborators
  • Access Controls
  • Quotas and Policies
• Running Jobs
  • Accounting
  • Performance tools
  • Sample Scripts
  • Interactive Sessions
  • Job Management
  • Job Status
  • Accessing the Compute Nodes
  • Scheduler
  • Partitions (Queues)
  • Job Policies
  • Monitoring a Node During a Job
  • Refunds
• Programming Environment (Building Software)
  • Modules
  • Compiling
  • Serial Programs
  • MPI Programs
  • Python
  • OpenMP Programs
  • Hybrid MPI/OpenMP Programs
  • OpenACC Programs
  • CUDA
• Installed Software
  • modules/lmod
  • Python
  • R
• Containers
  • Apptainer (formerly Singularity)
  • NVIDIA NGC Containers
  • Container list (as of March, 2022)
  • AMD Infinity Hub containers for MI100
  • Other Containers
• Open OnDemand
  • Shell Access
  • Managing files
  • Job Management
  • Interactive apps
• Debugging and Performance Analysis
  • AMDuProf Guide
  • NVIDIA Nsight Systems
  • Linux Perf performance counting
  • Linux/Unix gprof
  • Linux/Unix strace
  • mpiP MPI profiling
  • Debugging MPI (OpenMPI) codes
  • Debugging Open OnDemand Problems
• Acknowledging TGI Rails