Here’s what nine early adopters from six institutions and seven domains have to say about ACES!�
ACES | u.tamu.edu/aces | NSF Award #2112356
Ruisi Cai (UT-Austin) uses ACES to process long context sequences in Large Language Models (LLMs).��“LLMs command substantial computational power and agile memory management.”
Her team’s paper titled, “Learning to Compress Long Contexts by Dropping-In Convolutions,” was accepted by the International Conference on �Machine Learning (ICML24).
ACES | u.tamu.edu/aces | NSF Award #2112356
Aocheng Li (Purdue) uses ACES for data-driven archaeological site reconstruction.
“I love ACES’ elegant and light-weight web interface for file manipulation and job creation/submission. Using the composability features, I combine virtual network computing and TensorBoard servers to launch jobs and monitor training output with just a few clicks - all within one browser session. The HPRC staff are extremely helpful. Using ACES has been an enjoyable experience.”
ACES | u.tamu.edu/aces | NSF Award #2112356
Freddie Witherden (Texas A&M Department of Ocean Engineering) uses ACES to perform high-order accurate fluid flow calculations of �bluff bodies.
“ACES’ range of hardware, including CPUs, NVIDIA GPUs, and Intel PVC GPUs, is perfect for the development, testing and evaluation of performance-portable coding paradigms. Additionally, the large-memory nodes have proved invaluable for enabling us to perform preprocessing work for simulations on leadership-class computing resources.”
ACES | u.tamu.edu/aces | NSF Award #2112356
Rubem Mondaini (University of Houston) uses ACES to study quantum many-body problems in Condensed Matter Physics with the goal of understanding how Coulomb repulsion between electrons can affect quantum matter topology.
"ACES’ abundant supply of the latest CPUs (Sapphire Rapids), large memory and fast interconnect make it possible to reach physical system sizes unforeseen until now. This unique combination of assets makes all the difference with investigations in the quantum world.”
ACES | u.tamu.edu/aces | NSF Award #2112356
Chen-Chun Chen (Ohio State University NOWLAB) primarily uses the Intel GPUs and XeLink nodes on ACES.
“Using TensorFlow and Horovod, I’ve been running OSU Micro Benchmarks (OMB) to extend the MVAPICH library to support Intel PVC GPUs,” he said, and added, “I receive invaluable assistance from the HPRC helpdesk, and my experiments on ACES have been �consistently smooth.”
ACES | u.tamu.edu/aces | NSF Award #2112356
Wonmuk Hwang (Texas A&M Department of Biomedical Engineering) performs molecular dynamics simulations of biomolecules.
Dr. Hwang uses ACES to investigate the mechanical response of T-cell receptors which defend against pathogens, like influenza and the SARS CoV-2 virus.
“ACES’ NVIDIA H100s are great for carrying out multiple simulations. The HPRC staff are always helpful when troubleshooting aspects of this novel testbed.”
ACES | u.tamu.edu/aces | NSF Award #2112356
Junyuan Hong (UT-Austin) cited ACES in his latest research which presents a new method for private prompt tuning of LLMs, like ChatGPT.
The solution is called Differentially-Private Offsite Prompt Tuning (DP-OPT) which employs a discrete client-side prompt that can be applied to desired cloud models without significantly compromising performance.
ACES | u.tamu.edu/aces | NSF Award #2112356
Hanning Chen (Texas Advanced Computing Center) used ACES to conduct a Molecular Dynamics (MD) simulation of Satellite Tobacco Mosaic Virus with more than 28 million atoms.
“ACES is a powerful tool for MD simulations of large biological systems that reveal functions contributed by millions of atoms, or more. Our benchmark test with NAMD3 and a 64-node run revealed a performance of 4.8 ns/day, with an impressive 80 percent scaling factor when increasing the number of nodes from 1 to 64.
The HPRC team’s knowledge of this novel platform helps researchers progress quicker.”
ACES | u.tamu.edu/aces | NSF Award #2112356
Ishan Vatsaraj (Johns Hopkins University) uses ACES for research that was accepted by the 2024 European Society of Cardiology Congress in London.
Vatsaraj and his team are developing an artificial intelligence model for comprehensive electrocardiogram (ECG)-based diagnoses called ECGWiz. With the capacity to learn from only five vs. 120 ECG recordings required by conventional deep learning methods, its robust framework can more quickly evolve and adapt to new diagnostic challenges. Ongoing improvements, particularly through dataset expansion, underscore ECGWiz’ potential to enhance cardiovascular diagnostics for more precise and efficient patient care.
ACES | u.tamu.edu/aces | NSF Award #2112356
Five research teams have�NSF ACES allocations via the �National Artificial Intelligence Research Resource (NAIRR) Pilot�
The NAIRR Pilot, launched in January, 2024, aims to connect U.S. researchers and educators to computational, data, and training resources needed to advance AI research. Dozens of public and private entities are collaborating on this effort. The Pilot is a preparatory step toward an eventual full�NAIRR implementation.
ACES | u.tamu.edu/aces | NSF Award #2112356
While digitization of patient health records facilitates autonomous medical diagnoses, hidden biases can yield unfavorable outcomes. Using counterfactual data generation techniques, we will evaluate clinical decision models to ensure they treat patients from different demographic groups with varied behavioral patterns fairly without the need for additional human oversight. Innovation such as this will be especially useful in rural communities that suffer from a critical shortage of healthcare specialists. �
This research team has 230,400 SUs on ACES via the NAIRR (National Artificial Intelligence Research�Resource) pilot.
Evaluating and Mitigating Biases in Clinical Decision Models�Wei Wang, University of California at Los Angeles
ACES | u.tamu.edu/aces | NSF Award #2112356
Improving the Reliability of Large Multimodal Models for�Radiology with Confidence Estimates��Pranav Rajpurkar (Harvard Medical School)
Dr. Rajpurkar’s research addresses the critical issue of automation bias that can lead to diagnostic errors when using AI-assisted X-ray and CT-scan report-generation.
This research team has 240,000 SUs on ACES via the NAIRR (National Artificial Intelligence Research Resource) pilot.
ACES | u.tamu.edu/aces | NSF Award #2112356
Computational Foundations for Tractable Deep Generative Models
Guy Van den Broeck (University of California)
Probabilistic circuits provide an alternate, more capable and trustworthy architecture for the future of generative AI. Their pilot project will build a stronger foundation for gen-AI workflows, by scaling-up applications that employ tractable model learning, such as large language models, natural image distributions, and offline reinforcement learning models.
This research team has 256,320 SUs on ACES via the NAIRR (National Artificial Intelligence Research Resource) pilot.
ACES | u.tamu.edu/aces | NSF Award #2112356
Integrating Logic Reasoning with Vision Language Models
Kant, Krishna (Temple University)
Dr. Kant’s research employs novel strategies to improve AI-assisted video monitoring. One of his studies measures student engagement in virtual classrooms. To overcome limitations of traditional computer vision, Kant integrates automated logic reasoning with video language models (VLM) to achieve greater compatibility and consistency across a range of VLMs that can then be fine-tuned for specific events.
This research team has 48,000 SUs on ACES via the NAIRR (National Artificial Intelligence Research Resource) pilot.
ACES | u.tamu.edu/aces | NSF Award #2112356
Artificial Intelligence and Intelligent Systems�David Bamman (U-California at Berkeley)
Large language models (LLMs) are increasingly used to extract information from language, but they tend to excel at common tasks, vs. complex social constructs. Dr. Bamman’s team examines LLM performance with three complex measurement tasks (income inequality, representation of race in narratives, and relationships between characters), while exploring the degree to which LLMs capture language dialects and style to discover new knowledge about culture and society.
This research team has 600,000 SUs on ACES via the NAIRR Pilot (National Artificial Intelligence Research Resource).
ACES | u.tamu.edu/aces | NSF Award #2112356
PEARC24 Papers
Eleven Texas A&M HPRC and ACES’ affiliate papers were presented at PEARC24; 10 are included in the conference proceedings and ACM Digital Library. One is featured in a special edition of the Journal of Computational Science Education (JOCSE). HPRC brought home two best paper awards, and the prestigious Phil Andrews Award.
ACES | u.tamu.edu/aces | NSF Award #2112356
Since 2017, HPRC has offered computing and cybersecurity summer camps for secondary students that draw a diverse cohort from many states. Participants were asked what led them to be engaged in the first place. Four themes emerged: real-life applications, curiosity, the prospect of collaboration and desire to problem solve. We then followed-up after three months to ask how camp influenced their lives since attending. Many were inspired to dive even deeper into training that will prepare them for future cybersecurity careers!
Sandra B. Nite, Trenton J. Gray, Seonhu Lee, and Sheri Stebenne. 2024.
Engaging Secondary Students in Computing and Cybersecurity. In Practice
and Experience in Advanced Research Computing (PEARC ’24), July 21–25,
2024, Providence, RI, USA. ACM, New York, NY, USA, 5 pages. https://doi.
org/10.1145/3626203.3670624
Engaging Secondary Students in Computing and Cybersecurity
By Sandra Nite, PhD, et al.�Texas A&M University High Performance Research Computing (HPRC)�
Best Short Paper Workforce Development�PEARC24!
ACES | u.tamu.edu/aces | NSF Award #2112356
Exploring the Viability of Composable Architectures to Overcome Memory Limitations in�High Performance Computing Workflows
By Wesley A. Brashear, et al.�Texas A&M University High Performance Research Computing
Data- and AI-intensive computational workloads require copious memory which often results in bottlenecks once system resources are exceeded. With ACES’ menu of composable attributes, we identified a cost-effective way to increase memory capacity while sacrificing very �little performance.
PEARC ’24, July 21–25, 2024, Providence, RI, USA
© 2024 Copyright held by the owner/author(s).
ACM ISBN 979-8-4007-0419-2/24/07
https://doi.org/10.1145/3626203.367062019
ACES | u.tamu.edu/aces | NSF Award #2112356
We tested acceleration and scaling properties of ACES’ NVIDIA H-100 and Intel PVC GPUs over Liqid composable fabric with Kokkos and LAMMPS molecular dynamics software. Disparate computational and communication patterns emerged that affected performance. Among other outcomes, we believe Intel PVC performance would improve if they adopted an “all-on-the-GPU” strategy that Kokko provides.
PEARC ’24, July 21–25, 2024, Providence, RI, USA
© 2024 Copyright held by the owner/author(s).
ACM ISBN 979-8-4007-0419-2/24/07
https://doi.org/10.1145/3626203.3670631
Performance of Molecular Dynamics Acceleration Strategies on
Composable Cyberinfrastructure�
By Richard Lawrence, et al.
Texas A&M University High Performance Research Computing
ACES | u.tamu.edu/aces | NSF Award #2112356
By optimizing a convolutional neural network model to run on ACES’ Graphcore Intelligence Processing Units (IPUs), we conclude that IPUs offer a viable accelerator alternative to GPUs for machine learning applications in the fields of materials science and battery research. We also saw significantly improved performance of the Bow IPU over its predecessor, the Colossus IPU.
PEARC ’24, July 21–25, 2024, Providence, RI, USA
© 2024 Copyright held by the owner/author(s).
ACM ISBN 979-8-4007-0419-2/24/07
https://doi.org/10.1145/3626203.3670631
Insight Gained from Migrating a Machine Learning Model to Graphcore
Intelligence Processing Units
By Hieu Le, et al.�Texas A&M University High Performance Research Computing
ACES | u.tamu.edu/aces | NSF Award #2112356
This study investigates the importance of accelerator memory bandwidth over computational power in the performance of scalable high-order numerical simulations. Using ACES’ NVIDIA H100 and Intel PVC GPUs with the PyFR open-source fluid flow solver, we demonstrate how matrix multiplication characteristics influence the demand for memory bandwidth.
PEARC ’24, July 21–25, 2024, Providence, RI, USA
© 2024 Copyright held by the owner/author(s).
ACM ISBN 979-8-4007-0419-2/24/07
https://doi.org/10.1145/3626203.3670540
Impact of Memory Bandwidth on the Performance of Accelerators�
By Sambit Mishra, et al.�Texas A&M University High Performance Research Computing
ACES | u.tamu.edu/aces | NSF Award #2112356
Containers lower the barrier to entry with advanced cyberinfrastructure (CI) by offering a pre-assembled software stack and intuitive user interface. In this paper, we share lessons learned with a variety of container options by a team who manages a diverse CI portfolio. We have chronicled the varied ways they’re used, plus our experience with uptake, outreach and training.
PEARC ’24, July 21–25, 2024, Providence, RI, USA
© 2024 Copyright held by the owner/author(s).
ACM ISBN 979-8-4007-0419-2/24/07
https://doi.org/10.1145/3626203.3670550
Container Adoption in Campus High Performance Computing�at Texas A&M University
By Richard Lawrence, et al.�Texas A&M University High Performance Research Computing
ACES | u.tamu.edu/aces | NSF Award #2112356
Insights from computing transform work, education, scientific inquiry, industrial practice, economies, and our quality of life. But under-resourced communities have a steeper barrier to entry. Leveraging past progress with HPRC’s Building Research Collaborations at Community Colleges (BRICCs) program, “Pathways” continues to foster a regional effort to increase uptake and adoption of advanced computing and networking.
PEARC ’24, July 21–25, 2024, Providence, RI, USA
© 2024 Copyright held by the owner/author(s).
ACM ISBN 979-8-4007-0419-2/24/07
https://doi.org/10.1145/3626203.3670535
BRICCs: Building Pathways to Research Cyberinfrastructure at
Under-Resourced Institutions
By Dhruva Chakravorty, PhD, et al.�Texas A&M University High Performance Research Computing
Phil Andrews Award & Best Full Paper Workforce Development�PEARC24
ACES | u.tamu.edu/aces | NSF Award #2112356
Our student “fellows” are well-prepared for a range of advanced CI careers in both public and private sectors. Whether they work at the help desk, support researchers who use CI, develop applications, or test novel architecture, they help fill the national CI workforce pipeline upon graduation. With this paper, we share our time-tested framework for student success.
PEARC ’24, July 21–25, 2024, Providence, RI, USA
© 2024 Copyright held by the owner/author(s).
ACM ISBN 979-8-4007-0419-2/24/07
https://doi.org/10.1145/3626203.3670544
Cultivating Cyberinfrastructure (CI) Careers through Student Engagement at Texas A&M University�High Performance Research Computing
By Wesley A. Brashear, et al. (TAMU-HPRC)
ACES | u.tamu.edu/aces | NSF Award #2112356
Paper accepted by SEHET’24 - the Seventh Workshop on Strategies for Enhancing HPC Education and Training half-day workshop by �Nitin Sukhija (Slippery Rock University), et al.
Monday, July 22, 2024
9:00 a.m. - 12:30 p.m. ET
Room 557
Assessing the Impact of a CyberTraining Project: Expanding the Metrics By Sandra Nite, PhD, et al.�Texas A&M University High Performance Research Computing (HPRC)�
ACES | u.tamu.edu/aces | NSF Award #2112356
Providing Accessible Software Environments Across Science Gateways and HPC by Shaowen Wang, et al.�(University of Illinois at Urbana-Champaign and Massachusetts Institute of Technology)
While High-Performance Computing (HPC) resources accelerate the process of research discovery, they’re out of reach for many due to a lack of access or skills needed to use them. Dr. Wang and his colleagues have made it easier for everyone to access software environments across science gateways and HPC through the development of CyberGIS-Compute - a middleware toolkit that democratizes access for all.
Alexander Michels, Mit Kotak, Anand Padmanabhan, John Speaks, and Shaowen Wang. 2024. Providing Accessible Software Environments Across Science Gateways and HPC. In Practice and Experience in Advanced Research Computing (PEARC ’24), July 21–25, 2024, Providence, RI, USA. ACM, New York, NY, USA, 4 pages. https://doi.org/10.1145/3626203.367061
ACES | u.tamu.edu/aces | NSF Award #2112356
NetPointLib: Library for Large-Scale Spatial Network Point Data Fusion and Analysis by Shaowen Wang, et al.�(University of Illinois at Urbana-Champaign and Virginia Tech)
Figure 2: a) Chicago road network with crime data, b) synthetic data, c) network local auto k function hotspots, d) most significant network scan statistic hotspot.
Yunfan Kang, Fangzheng Lyu, and Shaowen Wang. 2024. NetPointLib: Library for Large-Scale Spatial Network Point Data Fusion and Analysis. In Practice and Experience in Advanced Research Computing (PEARC ’24), July 21–25, 2024, Providence, RI, USA. ACM, New York, NY, USA, 4 pages. https://doi.org/10.1145/3626203.3670615
The integration of high-performance computing (HPC) with NetPointLib is pivotal for managing and analyzing spatial data in social sciences, as demonstrated in the case study of Chicago’s crime data. This research was enabled by ACES.
ACES | u.tamu.edu/aces | NSF Award #2112356
Texas A&M HPRC was involved with 18 PEARC24 workshops, tutorials, paper presentations, and BoFs
ACES | u.tamu.edu/aces | NSF Award #2112356
HPRC Holiday Fun!
ACES | u.tamu.edu/aces | NSF Award #2112356
New Faculty Orientation
ACES | u.tamu.edu/aces | NSF Award #2112356
Receiving the Phil Andrews Award at PEARC24 in Providence, RI
Paper:
�BRICCs: Building Pathways to Research Cyberinfrastructure at
Under-Resourced Institutions��By Dhruva Chakravorty, et al.
ACES | u.tamu.edu/aces | NSF Award #2112356
Texas A&M HPRC and Friends in the News
ACES | u.tamu.edu/aces | NSF Award #2112356
San Jacinto Community College and Intuitive Machines work with Texas A&M’s BRICCs �initiative to build the space economy workforce.
ACES | u.tamu.edu/aces | NSF Award #2112356
Texas A&M University HPRC hosted the “SWEETER Winter Computing Festival” (SouthWest Expertise in Expanding, Training, Education, and Research) at the Omni Hotel in Corpus Christi, Texas, December 11-14, 2023.
ACES | u.tamu.edu/aces | NSF Award #2112356
The South African Student Cluster Competition Team visited Texas A&M HPRC in February following a visit to the Texas Advanced Computing Center and Dell, and prior to a NASA field trip. This team has consistently won or placed in the top three at the international competition in Germany every year since 2017.
ACES | u.tamu.edu/aces | NSF Award #2112356