HPC Interactive Environment (HIE) User Guide

1. Introduction

1.1. The HPC Interactive Environment

There are many HPC workloads which may greatly benefit from rapid response and high availability. This primarily includes interactive sessions for remote visualization, application development, and live debugging. The HPC Interactive Environment (HIE) is a mechanism on HPCMP allocated systems to deliver accelerated resource availability for such tasks. This guide provides information on how to use HIE.

Primary Use Cases of HIE
  • Remote visualization
  • Application development for General Purpose Graphics Processing Units (GPGPUs)
  • Application development for other non-standard processors

Other Interactive Use Cases Available
  • Active/Live debugging
  • Complete pre-and post-processing
  • Building applications

1.2. Appropriate Use of HIE

HIE provides elevated priority of workloads. There are a limited number of nodes and resources available in HIE. This mechanism is not intended to for elevating regular batch job priority. Using HIE in this way is considered abuse. Any considered abuse of HIE will be heavily scrutinized and may lead to usage limits or revoking availability to the user.

1.3. Requesting Assistance

The HPC Help Desk is available to assist users with unclassified problems, issues, or questions. Analysts are on duty 8:00 a.m. - 8:00 p.m. Eastern, Monday - Friday (excluding Federal holidays).

For more detailed contact information, please see our Contact Page.

2. HIE Configuration

2.1. Number of Resources Available

Jobs submitted to HIE have elevated priority on the systems, allowing accelerated access to resources. Resources available via HIE vary from system to system. The table below shows three main points of information:

  • Types of resources available to HIE on each system
  • Number of dedicated nodes and types allocated to HIE
  • Total limits per user
Available HIE Resources
System Node Types Number of Nodes Max Nodes Per Job Max Running Jobs Per User
Blueback All 24 None 3
Carpenter Compute 2 1 1
GPU
Jean GPU 24 Unlimited 10
BigMEM (lm) 6
AIML 36
Compute 496
Narwhal* All 24 None 3
Nautilus All 24 24 3
Raider All 16 1 2
Ruth Compute 632 Unlimited 2
GPU 24
BigMem 24
AIML 16
HiPerf 64
Warhawk Compute 8 1 2
GPU 6
BigMEM 1
Wheat Compute All resources are available 1 1
BigMEM (lm)
GPU
MLA

*Transition to Slurm in progress. Expected completion in FY26

2.2. How to Access

HIE is configured on all HPC systems, and by default is open to all users who can submit jobs on the system. The user need only submit their job to the HIE queue.

3. Submitting Jobs to HIE

We encourage you to read the respective system's User Guide for information regarding general submission of batch and interactive jobs on the system. Provided below are general outlines and targeted changes for adapting your job to target HIE; they do not comprehensively explain the job submission process. The main goal of most HIE jobs is to gain interactive access to graphics compute nodes. Each system may have specific requestable node attributes that can further refine your node selection, including bigmem or non-generic GPU types.

3.1. PBS Professional

All jobs of PBS are submitted through the qsub command. % qsub -I -q HIE -lselect=1:ngpus=1 -A Project_ID -l walltime=00:10:00

3.2. Slurm

All jobs of Slurm are submitted through the srun or salloc command. % srun --account= Project_ID --qos=hie --constraint=viz --pty /bin/bash -i