TAMU ACES

2FA/MFA

ACES is an innovative computing platform that provides a holistic solution for a wide range of users across diverse research communities, accommodating varying levels of computational adoption.

It leverages Liqid’s composable infrastructure framework through PCIe hybrid Gen4 and Gen5 architectures on Intel’s Sapphire Rapids processors. This enables a robust accelerator testbed featuring Intel Ponte Vecchio GPUs, Intel FPGAs, NVIDIA H100 GPUs, NEC Vector Engines, NextSilicon co-processors, and Graphcore IPUs.

These accelerators are integrated with Intel Optane memory and DDN Lustre storage, all interconnected via NVIDIA Mellanox NDR 400 Gbps InfiniBand, enabling high-throughput, low-latency data movement.

The platform supports the convergence of AI and machine learning with traditional simulation and modeling techniques. As edge computing and instrument-driven data collection continue to expand, ACES addresses the growing need to verify, process, store, analyze, and query massive volumes of unstructured data in real time.

Ask about ACES

File Transfer

Supported Methods Data Transfer Node URL
GLOBUS | RECOMMENDED ACCESS TAMU ACES DTN https://app.globus.org/dashboard
SCP/SFTP ACCESS TAMU ACES DTN
FTP ACCESS TAMU ACES DTN
RSYNC ACCESS TAMU ACES DTN
RCLONE ACCESS TAMU ACES DTN
PORTAL ACCESS TAMU ACES DTN https://portal.hprc.tamu.edu

Storage

File System

Directory Path Quota Purge Backup Notes
$HOME `/home/username` 10 GB ~10,000 files 6 months after account deactivation Nightly Small scripts, config files, not for general use
$SCRATCH `/scratch/user/username` 1TB ~250,000 files Not scheduled, but purged when quotas are exceed. None Primary working directory for jobs, not for long-term storage.
$PROJECT `scratch/group/projectid` 5TB ~500,000 files N/a No Shared storage for group members

Jobs

Queue specifications

Metrics updated 2026-05-02

Name Purpose CPUs GPUs RAM Jobs
30 days
Wait Time
30-day trend
Wall Time
30-day trend
cpu General CPU-only jobs Intel Sapphire Rapid: up to 96 cores/node, 64 nodes (6,144 cores max) 0 ~488 GB per node 7,728
cpu wait time: average 3.8 hours, range 0.4 to 13.2 hours over 30 days 3.8
cpu wall time: average 4.5 hours, range 1.2 to 6.8 hours over 30 days 4.5
gpu NVIDIA GPU workloads (AI/ML, CUDA, parallel GPU jobs) Intel Sapphire Rapid: 96/node H100 (up to 8 per node) High-memory GPU nodes (~256–512+ GB per node, varies) 1,745
gpu wait time: average 14.5 hours, range 3.4 to 51 hours over 30 days 14.5
gpu wall time: average 6.9 hours, range 2.5 to 11.7 hours over 30 days 6.9
gpu-debug Short GPU testing/debugging Intel Sapphire Rapid: 96 cores (1 node max) A30 (2 per node) ~488 GB per node
pvc Intel GPU Max (PVC) job Intel Sapphire Rapid: up to 3,072 cores across 32 nodes Up to 32 Intel PVC GPUs ~488 GB per node 395
pvc wait time: average 1.1 hours, range 0 to 15.1 hours over 30 days 1.1
pvc wall time: average 3.4 hours, range 0.3 to 15.3 hours over 30 days 3.4
bittware FPGA-based workloads and hardware acceleration Intel Sapphire Rapid: up to 96 cores across 2 nodes 0 (2 FPGA devices) ~488 GB per node 1 0 bittware wait time: 0 hours
bittware wall time: average 0.6 hours, range 0 to 0.6 hours over 30 days 0.6
memverge Memory-intensive workloads and large dataset processing Intel Sapphire Rapid: 96 cores (1 node) 0 ~488 GB per node
nextsilicon Experimental NextSilicon accelerator workloads (restricted access) Intel Sapphire Rapid: 96 cores (1 node) 0 (NextSilicon coprocessor) ~488 GB per node

Datasets

Name Description
pytorch-computer-vision-datasets

A collection of standard computer vision datasets formatted for PyTorch, supporting tasks like image classification and object detection. On ACES, these are used to benchmark GPU performance and test distributed deep learning workflows across accelerators.

pytorch-language-modelling-datasets

Text-based datasets for training NLP and language models in PyTorch. In ACES, they support benchmarking of large-scale, memory-intensive workloads and evaluating performance of transformer-based models across hardware.

tensorflow-computer-vision-datasets

Computer vision datasets optimized for TensorFlow, covering tasks such as classification and segmentation. Within ACES, they enable framework comparisons and validation of TensorFlow pipelines on heterogeneous accelerators.

tensorflow-language-modelling-datasets

NLP datasets prepared for TensorFlow, used for language modeling, translation, and text analysis. On ACES, they help evaluate distributed training performance and accelerator efficiency for sequential data workloads.

videollama_dataset

A multimodal dataset combining video and text for tasks like video understanding and captioning. In ACES, it is used to test high-throughput, multi-accelerator workflows and benchmark complex AI pipelines.