David Eberius, Ph.D. Email and Phone Number
I am a computer scientist specializing in High-Performance Computing (HPC) research. My research focus within HPC is performance analysis/modeling and tools development. I have extensive experience with performance analysis and optimization of MPI and GPU-based systems. I am looking to expand my research into the AI field, particularly applying my HPC performance analysis and optimization experience into inference and training pipelines. I also created the banner for my LinkedIn profile using Midjourney to create AI art of a cyberpunk-style desk setup.My Ph.D. dissertation research focused on analyzing performance of distributed applications using low-level metrics, both within the Open MPI runtime, and through NVIDIA's performance counters.The majority of my research projects have required programming skills in C/C++ and Python, using technologies such as MPI, CUDA, HIP, OpenSHMEM, and OpenMP. Some of my research highlights include:- Demonstrated the viability of resizable HPC jobs through simulation indicating significant potential improvements to run time, wait time, turnaround time, and system utilization- Created an extended roofline model for targeting GPU-based systems- Worked to help the ExaBiome ECP project get up and running on the pre-Frontier systems at ORNL- Published a paper on using low-level GPU metrics to provide a more accurate metric for system load on GPUs. Our hierarchical metric improves the correlation of measured performance and application workload by up to 20.61%, and reduces the residual load imbalance by up to 4× in a proxy application.- Completed my Ph.D. dissertation- Performed several research studies within the Open MPI project throughout my time at the Innovative Computing Laboratory at the University of Tennessee. My personal research was focused on performance analysis and profiling within the Open MPI runtime, but I assisted in multithreaded MPI research as well.I am interested in computer science research and computer programming, specifically within the fields of HPC and AI. I aspire to work as a Research Scientist/Engineer and eventually become a professor of Computer Science at the University level. I am also interested in computer graphics, specifically in the areas of animation and ray tracing.
-
Senior Middleware Development EngineerIntel Corporation Nov 2022 - PresentI develop software in support of runtime systems for communication libraries such as OpenSHMEM. -
Postdoctoral Research AssociateOak Ridge National Laboratory Oct 2020 - Oct 2022Oak Ridge, Tennessee, United StatesProvide assistance with optimization and analysis of large-scale scientific applications on High-Performance Computing systems.My primary project is assisting the ExaBiome ECP project in preparing the code to run on the Frontier system. My work includes porting codes from a CUDA implementation to HIP and assisting in trying to achieve comparable performance to the existing CUDA code. This project uses UPC++ for the distributed programming using a PGAS model.My secondary… Show more Provide assistance with optimization and analysis of large-scale scientific applications on High-Performance Computing systems.My primary project is assisting the ExaBiome ECP project in preparing the code to run on the Frontier system. My work includes porting codes from a CUDA implementation to HIP and assisting in trying to achieve comparable performance to the existing CUDA code. This project uses UPC++ for the distributed programming using a PGAS model.My secondary project is working on an extended roofline model for GPU-based systems. This extended model allows for more precise upper bounds to performance using extensions that are specifically targeted to GPUs. This work was published as a part of the P3HPC Workshop at SC22.I also finished publication of my work on GPU load balancing from my time at LLNL during this postdoc and presented my research at SC21. Show less -
Graduate Research AssistantInnovative Computing Laboratory (Icl) Aug 2014 - Jul 2020Knoxville, Tennessee, United StatesI worked on my Ph.D. dissertation work in performance analysis of distributed HPC applications through low-level metrics. This work operated in two primary directions: providing a tool for MPI library performance analysis, and creating a new methodology for load assessment on GPUs.For my MPI performance analysis work, I added Software-based Performance Counters (SPCs) to the Open MPI runtime, which allowed for reading of internal MPI library metrics through the MPI_T interface and… Show more I worked on my Ph.D. dissertation work in performance analysis of distributed HPC applications through low-level metrics. This work operated in two primary directions: providing a tool for MPI library performance analysis, and creating a new methodology for load assessment on GPUs.For my MPI performance analysis work, I added Software-based Performance Counters (SPCs) to the Open MPI runtime, which allowed for reading of internal MPI library metrics through the MPI_T interface and through a custom interface which worked with PAPI's software-defined events component. The initial version of this approach was published at EuroMPI '17.I worked with another student at my lab on performance analysis of multi-threaded MPI performance through two different publications in 2019, one at CLUSTER and the other at SC19.I also worked on projects such as the Performance Application Programming Interface (PAPI) and Parallel Runtime Scheduling and Execution Controller (PaRSEC). Show less -
Summer Research InternLawrence Livermore National Laboratory May 2018 - May 2020Livermore, California, United StatesI performed research on load balancing in GPU-based applications on High-Performance Computing systems. This work was included in my Ph.D. dissertation.The GPU load balancing work used several low-level GPU metrics provided through nvprof to get a more accurate view of the actual load on a GPU, which facilitates more effective load balancing. This work improves the correlation of measured performance and application workload by up to 20.61%, and reduces the residual load imbalance by… Show more I performed research on load balancing in GPU-based applications on High-Performance Computing systems. This work was included in my Ph.D. dissertation.The GPU load balancing work used several low-level GPU metrics provided through nvprof to get a more accurate view of the actual load on a GPU, which facilitates more effective load balancing. This work improves the correlation of measured performance and application workload by up to 20.61%, and reduces the residual load imbalance by up to 4× in a proxy application. I continued this work at ORNL during my postdoc and published it at SC21. Show less -
Summer InternUnited States Army Researcy Laboratory May 2016 - Jul 2016Aberdeen Proving Ground, MarylandI worked on an individual project with Mr. Song Park as my mentor. My project focused on solving an Army relevant path-finding problem using modern Artificial Intelligence techniques. The idea was to represent the problem as a Markov Decision Problem and then use a Monte Carlo Tree Search (MCTS) algorithm to determine a solution. I decided to implement my MCTS algorithm in parallel using OpenMP to prove that the algorithm could be parallelized efficiently. The future work would be to… Show more I worked on an individual project with Mr. Song Park as my mentor. My project focused on solving an Army relevant path-finding problem using modern Artificial Intelligence techniques. The idea was to represent the problem as a Markov Decision Problem and then use a Monte Carlo Tree Search (MCTS) algorithm to determine a solution. I decided to implement my MCTS algorithm in parallel using OpenMP to prove that the algorithm could be parallelized efficiently. The future work would be to implement this solution on a larger scale using MPI and a large distributed memory computer such as the Army's Excalibur supercomputer. Show less
-
Summer InternOak Ridge National Laboratory May 2015 - Jul 2015Oak Ridge, TnI worked on an image registration framework as a part of the Geographic Information Science and Technology (GIST) group at Oak Ridge National Laboratory (ORNL). My role in this project was primarily focused on accelerating a sensor model based orthorectification process using NVIDIA's CUDA technology and OpenMP. I also worked on implementing a profiling system for this project in order to better understand the performance bottlenecks of the image registration program to target areas for… Show more I worked on an image registration framework as a part of the Geographic Information Science and Technology (GIST) group at Oak Ridge National Laboratory (ORNL). My role in this project was primarily focused on accelerating a sensor model based orthorectification process using NVIDIA's CUDA technology and OpenMP. I also worked on implementing a profiling system for this project in order to better understand the performance bottlenecks of the image registration program to target areas for further optimization. This work ended up having a several times improvement in performance over the initial implementation. Show less
-
Resident AssistantSalisbury University Aug 2011 - May 2014I helped manage and create a welcoming and educational environment in Residence Halls on campus through event programming, educational initiatives, University policy enforcement, conflict analysis and dispute resolution, and mentorship. We were expected to regularly provide events for our residents to participate in that were either educational or social in nature. These events helped residents make friends and thrive in the campus environment. This position required extensive… Show more I helped manage and create a welcoming and educational environment in Residence Halls on campus through event programming, educational initiatives, University policy enforcement, conflict analysis and dispute resolution, and mentorship. We were expected to regularly provide events for our residents to participate in that were either educational or social in nature. These events helped residents make friends and thrive in the campus environment. This position required extensive training in interpersonal skills, conflict resolution, handling high-stress situations, community management, and many other skills. We were expected to live with our residents and get to know all of them personally. Show less -
Summer InternNational Science Foundation Jun 2013 - Aug 2013Salisbury, MarylandI worked with Dr. Arthur Lembo Jr. as my research mentor, and another student researcher, David Knipprath, to reduce the calculation time of performing terrain analysis on raster data sets to be used within Geography Information Science (GIS) software packages. The end result was a speed increase of multiple orders of magnitude over a traditional CPU implementation with actual speedups depending on the size of the data set (larger data sets provide better results). Unfortunately, Amdahl's Law… Show more I worked with Dr. Arthur Lembo Jr. as my research mentor, and another student researcher, David Knipprath, to reduce the calculation time of performing terrain analysis on raster data sets to be used within Geography Information Science (GIS) software packages. The end result was a speed increase of multiple orders of magnitude over a traditional CPU implementation with actual speedups depending on the size of the data set (larger data sets provide better results). Unfortunately, Amdahl's Law renders this speed improvement largely negligable because the overal operation is almost 99% I/O bound grabbing data from disc and moving it into memory and then to GPU memory. This work serves as an excellent proof of concept that GPGPUs can be used to great effect in speeding up raster terrain analysis operations. Continued work on this project focuses on performing a large number of calculations on data while it is already in memory to counteract the I/O bottleneck. Show less -
InternUs Army May 2011 - Jan 2012I worked on a project pertaining to analysis of subterranean blast events in order to be able to provide better understanding of threats through crater analysis.
David Eberius, Ph.D. Education Details
Frequently Asked Questions about David Eberius, Ph.D.
What company does David Eberius, Ph.D. work for?
David Eberius, Ph.D. works for Intel Corporation
What is David Eberius, Ph.D.'s role at the current company?
David Eberius, Ph.D.'s current role is Senior Middleware Development Engineer at Intel | HPC | Performance Analysis/Modeling | MPI | GPUs | Parallel Programming.
What schools did David Eberius, Ph.D. attend?
David Eberius, Ph.D. attended University Of Tennessee-Knoxville, Salisbury University.
Who are David Eberius, Ph.D.'s colleagues?
David Eberius, Ph.D.'s colleagues are Chakradhar Mella, Arthi Danabal, Ruchama Aharoni, Trisha Sparacino, Chao-Kai (Ck) Liang, Yan Hern Tan, Reshmi Dani.
Free Chrome Extension
Find emails, phones & company data instantly
Download 750 million emails and 100 million phone numbers
Access emails and phone numbers of over 750 million business users. Instantly download verified profiles using 20+ filters, including location, job title, company, function, and industry.
Start your free trial