2025 Energy HPC Conference

Accelerating Downhole Distributed Acoustic Sensing Data Processing for Borehole Seismic Monitoring
Rosie Zhu (Rice University) and Jonathan Ajo-Franklin (Rice University)
 
Carbon-Neutral Recovery of Natural Gas from Shale: Sequestering CO2 While Enhancing Gas Production
Cristel Carolina Brindis Flores (Rice University), Walter Chapman (Rice University) and Philip Singer (Rice University)

Confidence and Degeneracy in Parameter Estimation of Biophysical Neuron Models
Anwar Khaddaj (Rice University), Saina Namazifard (Baylor College of Medicine), Matthias Heinkenschloss (Rice University) and Fabrizio Gabbiani (Baylor College of Medicine)

Efficient Monte Carlo Methods for Estimating System-Level Reliability of Power Grids
Ziran Wang (Rice University) and Leonardo Dueñas-Osorio (Rice University)

Energy HPC Orchestrator: A Cloud Native HPC Workflows Management Application
Kun Jiao (AWS), Cyril Lagrange (AWS), Dan Kahn (AWS), Shel Hussein (AWS), Max Liu (Shell) and Marwan Wirianto (Shell)
 
Enhancing Data Center Energy Efficiency with a Solar Thermal Boosted Waste Heat Recovery System
Kashif Liaqat (Rice University) and Laura Schaefer (Rice University)
 
Integration of Machine Learning, Geospatial Analysis, and Life Cycle Assessment to Determine the Well Pad-Specific Carbon Footprints of Natural Gas Extraction
Amir Sharafi (University of Nevada, Las Vegas) and Marie-Odile Fortier (University of Nevada, Las Vegas)

Learning to Route with Confidence Tokens
Yu Neng Chuang (Rice Univerisity), Guanchu Wang (Rice Univerisity) and Xia Hu (Rice Univerisity)

Leveraging HPC and Advanced Hydrological Modeling for Accurate Flood Forecasting
Girishchandra Yendargaye (C-DAC), Murugesh Prabhu (C-DAC), Upasana Dutta (C-DAC) and Yogesh Kumar Singh (C-DAC)
 
Measuring and Analyzing Application Performance At Exascale
Dragana Grbic (Rice University) and John Mellor-Crummey (Rice University)

One Rank at a Time: Cascading Error Dynamics in Sequential Learning
Mahtab Alizadeh Vandchali (Rice University), Fangshuo Liao (Rice University), Nikhil Chigali (Rice University) and Anastasios Kyrillidis (Rice University)
 
Optimizing US Industrial Heat and Power Systems with Geothermal Deployment in 2030
Chen Chen (Rice University) and Daniel Cohan (Rice University)

Pluvial Flood Emulation with Hydraulics-Informed Message Passing
Arnold Kazadi (Rice University), James Doss-Gollin (Rice University) and Arlei Silva (Rice University)

Repulsive Latent Score Distillation for Solving Inverse Problems
Nicolas Zilberstein (Rice University)

Reverse Time Migration on the STX Accelerator
Ryuichi Sai (Rice University), John Mellor-Crummey (Rice University), Timo Schlachter (Fraunhofer ITWM), Jens Krüger (Fraunhofer ITWM) and Mauricio Araya-Polo (TotalEnergies EP Research & Technology US)

Runtime and Energy Analysis of SpMV Hardware Execution Choice
Max Hawkins (Georgia Institute of Technology), Christian Engman (Georgia Institute of Technology) and Ivan Rocha (Georgia Institute of Technology)

Sensitivity-Driven Surrogate Model Refinement for Trajectory Optimization with Expensive Black-Box Functions
Jonathan Cangelosi (Rice University) and Matthias Heinkenschloss (Rice University)

Sweeping Heterogeneity with Smart MoPs: Mixture of Prompts for LLM Task Adaptation
Anastasios Kyrillidis (Rice University)

The Accessibility and Inaccessibility of Urban Public Charging Station
Hossein Gazmeh (Rice University), Xinwu Qian (Rice University), Mario Small (Columbia University), Qi Wang (Northeastern University) and Yuntao Guo (Tongji University)

Towards Flexible Demultiple with Deep Learning
Mario Ruben Fernandez (Fraunhofer ITWM), Norman Ettrich (Fraunhofer ITWM), Matthias Delescluse (École Normale Supérieure) and Janis Keuper (Offenburg University)

Unbalanced Research Effort: A Comparative Analysis of Energy Research Articles in Commercial and Residential Buildings
Sumeyra Danisman (Stony Brook University) and Elizabeth Hewitt (Stony Brook University)

Unsupervised Quantum Anomaly Detection: Quantum Computing for Detecting Critical Anomalous Events
Jason Ludmir (Rice University) and Tirthak Patel (Rice University)

Exhibit Hall | 8:30 am - 5:00 pm
Limited seating; 30 registrants - SOLD OUT

Speakers: Guillaume Barnier, Guillaume Thomas-Collignon, and Igor Terentyev (NVIDIA)

Schedule:

• 8:30 - 9:00 am: Check-in + Breakfast
• 9:00 - 10:00 am: Introduction, Theory Review (PDE + Numerical Scheme)
• 10:00 - 11:30 am: Initial Implementation + Profiler Report Introduction and Analysis
• 11:30 - 12:30 pm: Lunch
• 12:30 - 1:30 pm: Optimization #1 Using Shared Memory
• 1:30 - 2:30 pm: Optimization #2 Using Asynchronous Shared Memory Loads
• 2:30 - 3:30 pm: Optimization #3 Using TMA
• 3:30 - 4:00 pm: Break
• 4:00 - 5:00 pm: Theory Review on Adjoint System of Equations for FWI, Numerical Implementation, and Differences with Forward

Materials: It is highly recommended for attendees to bring their own laptop, but the speakers will still try to make the workshop understandable and adapted for people that do not have a computer. There will be power, but please charge in advance as some outlets may need to be shared.

Abstract: Elastic full waveform inversion (FWI) is becoming the industry's standard for subsurface model parameter estimation. However, this technique requires to simulate hundreds of thousands of wave propagations by numerically solving a system of partial differential equations (PDE). Consequently, implementing an efficient numerical scheme on GPUs is critical.

In this workshop, we propose to teach the attendees how to gradually build finite-difference (FD) propagators for elastic media (ISO and VTI) optimized for Nvidia's latest GPUs (Ampere, Hopper, and Blackwell).

We provide a brief theoretical review, and we describe the numerical scheme we implement, which is based on a staggered-grid approach for both time and space. We then gradually implement multiple versions of the forward propagator, starting from a baseline implementation that requires minimum GPU hardware knowledge, to our fastest version using asynchronous load to shared memory. At each step, we use our profiling tool - Nsight Compute (NCU) - to identify bottlenecks in our kernels and we show how to leverage Nvidia's new hardware features to mitigate these bottlenecks. Finally, we show how to derive and efficiently implement the adjoint propagator required for the elastic FWI gradient computation.