|
|
Preparing to Run on Endeavour
The Endeavour supercomputer comprises two nodes, Endeavour3 and Endeavour4. Each of the two systems includes 896 cores and six terabytes (TB) of global shared memory. See Endeavour Configuration Details for more hardware information. Endeavour 3...
04 Nov, 2026
Views: 4331
NAS Lustre Filesystems
All NAS systems share several filesystems intended to provide working space for compute jobs. These filesystems, called "nobackup" (/nobackuppX), provide more than 90 petabytes (PB) of disk space and serve many thousands of cores. The /nobackup...
03 Nov, 2026
Views: 38025
Job Accounting
Usage on HECC compute systems is charged using a standard billing unit (SBU). SBUs are charged to each PBS job running on the compute nodes according to the following formula: SBUs charged = wall-clock hours used * number of MAUs * SBU rate ...
16 Oct, 2026
Views: 18469
Reserving a Dedicated Compute Node
The reservable front end feature allows you to reserve dedicated access to an Electra compute node for up to 90 days. You can use the node the same way you would use a PFE—to submit jobs and perform other tasks such as post-processing,...
16 Oct, 2026
Views: 20422
PBS Job Queue Structure
Running Jobs on Athena To run a PBS job on the Athena compute nodes, submit the job from one of the Athena front-end systems (ATHFEs) to one of the queues managed by the PBS server pbs06a: devel, normal, long, and wide. Note: You cannot submit...
14 Oct, 2026
Views: 55599
Requesting Cabeus GPU Resources
The Cabeus cluster comprises two types of NVIDIA GPU nodes, including nodes with x86_64-based CPUs and nodes with aarch64-based CPUs. Note: You must have a Standard Billing Unit (SBU) allocation specifically for GPUs in order to use the GPU...
13 Oct, 2026
Views: 19643
Cabeus GPU Nodes
Cabeus, named after a lunar crater in the south polar region of the Moon, is a cluster system that supports HPC and machine learning/artificial intelligence (ML/AI) applications that can benefit from the GPU technology. GPU Hardware Summary The...
13 Oct, 2026
Views: 49888
CUDA Toolkit and CUDA Driver
CUDA stands for Compute Unified Device Architecture. It is a parallel computing platform and programming model created by NVIDIA in 2006 for accelerating computing on General-Purpose Graphics Processing Units (GPGPUs). This article provides a...
13 Oct, 2026
Views: 13542
Preparing to Run on Cabeus Grace Hopper Nodes
Cabeus includes 538 GPU nodes based on the NVIDIA GH200 Grace Hopper Superchip. To help you prepare for running jobs on the Grace Hopper GPU nodes, this short user guide includes information on the general configuration of the nodes, compiling...
13 Oct, 2026
Views: 16397
Software on NAS x86_64 Systems
Software programs on NAS systems are managed as modules or packages. Among the NAS systems, Cabeus' aarch64-based GH200 GPU nodes running the TOSS5 operating system have a relatively small collection of sofware packages; the x86_64-based CPU-only...
25 Sep, 2026
Views: 28313
Enabling User-Defined Software Stack with Apptainer
Apptainer, previously known as Singularity, is an open-source project under the Linux Foundation since November 2021. Apptainer containers provide an application virtualization layer enabling mobility of compute via both application and...
14 Sep, 2026
Views: 249
Running an Apptainer Container Image on NAS Systems
If you have an existing Apptainer image that is ready to run, you should be able to run it on any platform where Apptainer is installed. This article provides information to help you run an Apptainer container image on NAS systems. Note: A...
14 Sep, 2026
Views: 396
Performance Analysis of Your GPU CUDA Kernels with Nsight Compute CLI
Nsight Compute is a tool that collects metrics via hardware counters and software instrumentations for deep-dive profiling and guided performance analysis of the CUDA kernels. GPU architectures supported by Nsight Compute started with the NVIDIA...
14 Sep, 2026
Views: 14395
Basics on NVIDIA GPU Hardware Architecture
This article provides basic hardware architecture information to help you get familiar with the NVIDIA A100 and H100 graphics proessing units (GPUs) used in the NAS GPU nodes (mil_a100 and gh200). Refer to the official guides from NVIDIA listed in...
14 Sep, 2026
Views: 64833
Simplifying GPU Programming with Unified Memory
This article introduces the unified memory feature available in all of the NAS GPU nodes and describes how to take advantage of this feature to simplify GPU programming. Introduction In a heterogeneous compute system containing a CPU host and...
14 Sep, 2026
Views: 21345
Performance Analysis of Your GPU Applications with Nsight Systems
NVIDIA Nsight Systems is a low-overhead statistical sampling profiler with tracing features to check overall performance of your GPU applications. The Feature Table in Get Started with Nsight Systems provides a glimpse of its capabilities,...
14 Sep, 2026
Views: 13075
Compiling CUDA Applications for Different Generations of NVIDIA GPUs at NAS
The NAS Cabeus cluster currently includes NVIDIA GPUs with the Ampere (mil_a100) and Hopper (gh200) microarchitectures. This article provides some information on the compute capability of the various microarchitectures and how to compile...
14 Sep, 2026
Views: 11147
Getting Started with Podman at NAS
Podman is an experimental offering on NAS systems with the x86_64 architecture. This article helps you get started, provides a list of commands, and outlines the limitations of this offering. For more information about Podman, see the Podman...
14 Sep, 2026
Views: 3170
Using TMPDIR As a Temporary Filesystem for GPU Jobs
For single-node I/O that uses memory-based temporary filesystem, we recommend using a job-based temporary directory, $TMPDIR, instead of the node-wide /tmp directory. This recommendation is particularly important for jobs running on the NAS GPU...
14 Sep, 2026
Views: 2964
Preparing to Run on Cabeus A100 GPU Nodes
The Cabeus cluster comprises multiple NVIDIA GPU nodes, including: 128 nodes with Milan CPU and NVIDIA A100 GPUs (model=mil_a100) 350 nodes with NVIDIA GH200 Superchip (model=gh200) Among them, the gh200 nodes are served by their own front...
14 Sep, 2026
Views: 7992
How to Run a Graphics Application in a PBS Batch Job
In order to run a graphics application in a PBS batch job, you must provide a value for the DISPLAY environment variable—even if you do not have or need an actual display for the job. If the batch job cannot find this value, it will quit and return...
14 Sep, 2026
Views: 14575
Basic Tasks
Once you have completed preliminary tasks and read the Environment Overview, you can get started with using the supercomputers. The basic tasks below cover a typical user workflow cycle. Even these basic tasks require numerous setup steps...
14 Sep, 2026
Views: 54665
Common Reasons for Being Unable to Submit Jobs
Listed below are several common reasons why a job submission might not be successful. Compute Model not Recognized If you get the following message after submitting a PBS job: qsub: Unrecognized model xxx It is possible that you are...
14 Sep, 2026
Views: 39335
Job Accounting Utilities
You can use the following job accounting utilities to obtain resource usage and charging information for your project group IDs (GIDs): Year-to-date summary: acct_ytd Detailed account queries: acct_query TIP: Use the myNAS portal to easily...
14 Sep, 2026
Views: 20702
Portable Batch System (PBS): Overview
All NAS facility supercomputers use the Portable Batch System (PBS) from Altair for batch job submission, job monitoring, and job management. Note that different systems may use different versions of PBS, so the available features may vary slightly...
14 Sep, 2026
Views: 47143
|
