2024 Frontera User Meeting

August 5–6, 2024

The 2024 Frontera User Meeting will be held in person August 5-6, 2024, at the Texas Advanced Computing Center in Austin, Texas. The Frontera User Meeting is a valuable opportunity to participate in the community of scientists, engineers, and technologists who use and operate this unique national resource.

New this year will be a half-day training related to Vista (learn more below), a new Arm-based system with a significant GPU component for simulation and machine learning. We will also feature science talks from Frontera users; roadmap sessions from project staff about Frontera and the Leadership-Class Computing Facility (LCCF); and feedback panels for users to provide input to the project.

Agenda

Please click on talk titles to access available presentation files.

Monday, August 5

8:00am	Registration Opens (ACB Lobby)
9:00am	Breakfast
9:30am	VISTA Training
12:00pm	Lunch
1:00pm	Welcome Remarks: Dan Stanzione Frontera Highlights
1:45pm	Yihao Zhou Carnegie Mellon University Gravitational wave emitted by massive blackholes: predictions from cosmological simulation Astrid
	Eduardo Gutierrez The Pennsylvania State University NumericalRelativity Simulations of Compact Binary Mergers
	Yanhui Yang University of California, Riverside Goku: A 10-Parameter Simulation Suite for Cosmological Emulation
	Himawan Winarto Princeton University Kinetic simulations of transport and turbulence in collisionless astrophysical plasmas
	Break
3:30pm	User Feedback Panel: TACC Team Dan Stanzione, John Cazes, Tommy Minyard, Virginia Trueheart
4:15pm	Federico Fraternale The University of Alabama in Huntsville, CSPAR Simulating the time-dependent global interaction of the solar wind and the LISM with kinetic H and He neutral atoms using MS-FLUKSS
	Rajat Chakraborty University of Illinois Urbana-Champaign Robust RNA Tail Detection Techniques for High-Density DNA Data Storage Using Solid-State Nanopores
	David Hardy University of Illinois at Urbana-Champaign NAMD Molecular Dynamics on Frontera and Beyond
	Seung Hyun Kim Gazzola Lab Modular High-throughput Computing Pipelines for Scalable Neuronal Electrophysiology Research
5:30pm	Day One Sessions End
5:30pm	Welcome Reception — TACC Lobby

Tuesday, August 6

8:30am	Breakfast
9:00am	Zuzanna Jedlinska University of Pennsylvania MATILDA.FT: A GPU accelerated software for coarse-grained molecular dynamics simulations
	Chendi Xie Clemson University Stabilizing Room-Temperature Superconductivity in Hydrides by Nonequilibrium Driving
	Irmak Taylan Karpuzcu University of Illinois Urbana-Champaign Particle-kinetic modeling of high speed shock boundary layer interactions and large separation bubbles
	Break
10:30am	Michael Gurnis California Institute of Technology Plate tectonics, great earthquakes, and rheology across timescales
	Arushi Saxena Clemson University Integration of Geophysical Constraints in Global Mantle Flow Models for Insights into Plate Tectonics
	Jason Fleming Seahorse Coastal Consulting HPC Powered Real Time Model Guidance with National Impact
	Machine Room Tour
12:00pm	Lunch
1:00pm	Richard Loft AreandDee LLC Early Benchmarks of the EarthWorks Model on the Grace/Hopper Architecture
	Jacob Badger University of Texas at Austin Advancing Subsurface Modeling with a Scalable Frequency-Domain Wave Solver
	Break
2:00pm	Future Directions Panel: TACC Team Dan Stanzione, John Cazes, Tommy Minyard, Lars Koesterke
3:00pm	Day Two Sessions End

Registration

Registration is now closed

Lodging

The user meeting will be held in the ACB building on the J.J. Pickle Research Campus (PRC) of The University of Texas at Austin at 10100 Burnet Road, Austin, TX, 78758.

Lodging recommendations can be found at: tacc.utexas.edu/about/hotels (hotels located on Domain Drive are close enough to walk).

Contact

Please direct any questions or comments to Natalie Henriques at natalie@tacc.utexas.edu.

Past Meetings

Learn more about past meetings by exploring agendas and presentations.

VISTA Training

Vista is a new cluster going into production in the second half of 2024. It is based on NVIDIA’s new integrated CPU and GPU technologies named Grace and Hopper. Vista consists of two different node types, Grace-Grace and Grace-Hopper. Both nodes types can be thought of as a tightly coupled dual-socket setup. Grace-Grace nodes host 2 NVIDIA ARM CPUs, and Grace-Hopper nodes host a Grace CPU and a Hopper GPU. This system will be the first of many based on a unified memory architecture that tightly integrates CPUs and GPUs.

The tutorial is comprised of a short overview and 2 main sections. We’ll introduce a tool developed at TACC that can automatically and very efficiently detect and offload SGEMM/DGEMM calls, while handling and avoiding data transfer between CPU and GPU. We will also demonstrate how to access containers on Vista and how to run HPC and AI applications.