TAMU Launch is a Texas A&M HPRC Dell Linux cluster available through ACCESS for general research computing and data-enabled workflows. It supports batch-scheduled CPU and GPU jobs, interactive portal-based work, shared storage, and common research software through modules.
Launch includes 45 compute nodes: 35 CPU nodes and 10 GPU nodes. CPU and GPU compute nodes use AMD EPYC Genoa processors with 192 cores per node; GPU nodes also provide two NVIDIA A30 GPUs. The cluster uses HDR100 InfiniBand and shared NFS storage.
Jobs
Launch uses Slurm for batch-scheduled work. Resource-intensive workloads should be submitted to compute nodes rather than run directly in an interactive shell.
Most users will use standard Slurm commands such as sbatch, squeue, scancel, and seff. The HPRC Drona Workflow Engine is also available through the portal for users who prefer a graphical workflow for generating and submitting jobs.
GPU jobs should request the GPU partition and GPU resources. For example:
#SBATCH --partition=gpu #SBATCH --gres=gpu:1To request an A30 GPU specifically:
#SBATCH --partition=gpu #SBATCH --gres=gpu:a30:1For detailed Slurm examples, queue information, Drona instructions, and job-script templates, see the official Launch Batch System documentation.
Queue specifications
| Name | Purpose | CPUs | GPUs | RAM | Jobs
30 days
|
Wait Time
30-day trend
|
Wall Time
30-day trend
|
|---|---|---|---|---|---|---|---|
| CPU compute node | General CPU jobs | 2 × AMD EPYC Genoa 9654; 96 cores/socket; 192 cores/node | None | 384 GB DDR5-4800 | — | — | — |
| GPU compute node | GPU-accelerated jobs | 2 × AMD EPYC Genoa 9654; 96 cores/socket; 192 cores/node | 2 × NVIDIA A30 | 768 GB DDR5-4800 | — | — | — |
| Login node, not a job queue | 32 AMD EPYC Genoa 9124 | 384 GB | — | — | — |
Storage
File System
| Directory | Path | Quota | Purge | Backup | Notes |
|---|---|---|---|---|---|
| Home | /home/<USERID> | 10 GB / 10,000 files | Deleted 6 months after account deactivation | Backed up nightly | Small files, scripts, editing |
| Scratch | /scratch/user/<USERID> | 1 TB / 250,000 files | Temporary; remove unused files | Not backed up | Temporary work space |
| Project | /scratch/group/<PROJECTID> | 5 TB / 500,000 files | Removed 90 days after allocation expiration | Shared project space |
External Storage
Immediately upon logging in to Launch, the following message about the status of your disk space use greets you (format may vary):
Your current disk quotas are:
Disk Disk Usage Limit File Usage Limit
/home 2.49G 10G 113 10000
/scratch 1.25G 1T 40 250000
Type 'showquota' to view these quotas again.File Transfer
Launch users can upload and download files through the HPRC Portal file browser. This is convenient for small files and browser-based workflows, but HPRC does not recommend portal upload/download for files larger than 2 GB.
HPRC also documents transfer tools such as rsync, Globus, rclone, and gdown. Because direct SSH login is currently not available for Launch, users who need command-line transfer workflows should follow the current HPRC Launch and File Transfer documentation rather than assuming that direct local-to-Launch SSH-based transfer commands are available.
| Supported Methods | Data Transfer Node | URL |
|---|---|---|
| HPRC PORTAL FILE BROWSER | RECOMMENDED | https://portal-launch.hprc.tamu.edu | |
| RSYNC | https://hprc.tamu.edu/kb/Helpful-Pages/File-Transfer/#rsync | |
| GLOBUS | RECOMMENDED | ACCESS TAMU LAUNCH DTN | https://hprc.tamu.edu/kb/Software/Globus/ |
Login to TAMU Launch
Direct SSH login is currently not available for Launch. Authorized ACCESS users should log in through the Launch OnDemand portal:
https://portal-launch.hprc.tamu.edu/The Launch portal authenticates ACCESS users through the web portal. After login, users can access shell sessions, files, job tools, and interactive applications through the portal interface.