As a charitable service-based non-profit organization (NPO) educating and connecting the High Performance Computing (HPC) user community to state of art technology for the purpose of optimizing business processes and workforce advancement. Our technology focus includes AI/Machine Learning, Data Science, Cloud Computing, and Visualization utilized in applications ranging from Energy, Life Sciences, Manufacturing and Engineering, Financial Services, Academia, and Government.
The Society of HPC Professionals past lunch and learn event
Lunch & Learn – January 2021
QEarth on Alveo cards: Accelerating machine learning inference to achieve real time earth-modeling
28 January 2021
Live stream due to COVID-19
Watch video and/or download presentation
About the Event
The goal of this presentation is to highlight the integration of Xilinx hardware and software acceleration with innovative machine learning driven products and services to bring material value to a broad population of oil and gas operators. To achieve this goal Xilinx and Quantico Energy Solutions undertook cooperative research and development efforts to accelerate the inferencing capabilities of the QEarth product using Xilinx Alveo U250 cards and Vitis libraries. QEarth utilizes multiple data inputs with an emphasis on well logs and 3D seismic volumes to create training data sets for the prediction of physical properties across the 3D space. The process ends with multiple models ensembled to produce a highly accurate, industry leading prediction. The model ensembles are very compute intensive and can run for days. Xilinx Alveo U250 cards allow data scientists to access FPGAs from within a datacenter or the cloud. FPGAs are used in multiple applications for real time inferencing in self-driving cars, 5G networks and even the space station. Alveo cards provide class leading processing speeds and acceleration that rival the fastest GPU’s on the market with a fraction of the acquisition cost and a quarter of the power requirements of a GPU. QEarth, while fully integrated and delivered via AWS cloud services, underwent full integration with Vitis libraries to achieve performance levels that made real time earth-modeling a reality. The entire QEarth workflow brings contextually relevant AI-driven earth-modeling capabilities for asset operators while Xilinx acceleration creates strategic and operational opportunities via the space of real time intelligence.
The development teams were able to take a MATLAB built model and port this model via Vitis BLAS, HPC libraries and C functions into the FPGA platform. The C functions used can store the NN parameters in on-chip memory. By iterating through the input data, we can run multiple NN’s in an ensemble.
About the Speakers
Orlando De Jesus
Senior Technical Advisor, Quantico
Orlando De Jesus is a Senior Technical Advisor for Quantico Energy Solutions with 30+ years of professional experience in developing, implementing, designing and redesigning engineering solutions. Demonstrated track record on SCADA systems, high temperature electronics, neural networks development, experimental validation, Multiphysics simulations, control algorithm development, object oriented software development, interface between different applications, embedded software development. Orlando holds a Ph.D. in Electrical and Computer Engineering from Oklahoma State University and Project Management Specialist from Universidad Simon Bolivar, Caracas Venezuela.
Director of Operations, Quantico
Nathan Chang is the Director of Operations at Quantico Energy Solutions, an energy services company dedicated to applying machine learning to subsurface G&G and D&C challenges. Nathan is a graduate of the University of Houston and has over 15 years of professional experience. Seven of those years he spent with Fugro NV in various strategic roles with the majority of his time leading Global Innovation and R&D.
Senior Engineer, Xilinx
Lisa Liu is a Senior Engineer in Xilinx. She has been working with FPGA platforms for more than 10 years. Her recent work is focused on library development including but not limited to BLAS, SPARSE and HPC for Xilinx FPGAs and Alveo cards.
REGISTRATION IS CLOSED
FREE for Members and U of H Student Members
$15 for Non-Members
This is a good time to consider joining SHPCP since members get free lunch, ability to live stream events, and copies of presentations and videos (when approved by presenter)
Click here to join now and attend for free
Due to COVID-19, this month’s Lunch & Learn is a live stream event only.
For members, the link is on the Member Resources page when you log in.
For non-members, the link will be provided after you register