Year 2022
Alabandi, Ghadeer
Nominated by Dr. Martin Burtscher
“Ghadeer is advancing the state of the art of parallelizing important graph algorithms. For example, she has published a new approach to extract several times more parallelism out of graph coloring than the prior work, leading to the fastest graph-coloring GPU implementation available to date. Similarly, she has published a high-speed algorithm for balancing social networks that is orders of magnitude faster than the related work, making it possible, for the first time, to process large real-world inputs. Both of these works have appeared in strong conferences. She also created the first parallel heuristics for improving the coloring of a graph, which is under submission at a journal. She is currently working on a multi-GPU maximal independent set algorithm that minimizes data transfers as well as on a new approach to GPU-based graph pattern mining that works on large patterns on which pre-existing techniques run out of memory.
Ghadeer Alabandi, Jelena Tesic, Lucas Rusnak, and Martin Burtscher. Discovering and Balancing Fundamental Cycles in Large Signed Graphs. Proceedings of the 2021 ACM/IEEE International Conference for High-Performance Computing, Networking, Storage and Analysis, Article 68, 1-17. November 2021. (27% acceptance rate)
Ghadeer Alabandi, Evan Powers, and Martin Burtscher. Increasing the Parallelism of Graph Coloring via Shortcutting. Proceedings of the 2020 ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming, pp. 262-275. February 2020. (23% acceptance rate)”
Azami, Noushin
Nominated by Dr. Martin Burtscher
“Noushin is advancing the state of the art of parallel graph codes. For example, she is the first person who managed to accelerate the fastest graph analytics GPU codes by using a compressed in-memory representation of the graph. Whenever the program accesses the graph, the information needs to first be decompressed. Noushin designed and implemented a massively-parallel decompression algorithm that has such a low latency that it is able to deliver a speedup over accessing the uncompressed graph. This research is currently under submission. Noushin also helped design a new class of benchmark suite (for testing program verification tools). The suite automatically generates meaningful buggy and bug-free variations of important code patterns found in irregular algorithms. Because the behavior of such codes is data dependent, the suite also automatically synthesizes inputs for running the codes and eliciting different behaviors. The suite comprises over 1000 codes and an even larger number of inputs for each code, yielding several million possible experiments that can be executed. This work has been accepted for publication at ISPASS’22.
Yiqian Liu, Noushin Azami, and Martin Burtscher. The Indigo Program-Verification Microbenchmark Suite of Irregular Parallel Code Patterns. Proceedings of the 2022 IEEE International Symposium on Performance Analysis of Systems and Software. May 2022.
One more paper currently under submission.”
Aziz, Samantha
Nominated by Dr. Oleg Komogortsev
"I nominate Samatha Aziz for her outstanding work in eye-tracking signal quality as related to signal evaluated on multiple devices and submitting two papers on this topic."
Bhandari, Banooqa Keshav
Nominated by Dr. Ziliang Zong
"Keshav Bhandari has been working on a research project in the past 6 months on Learning Omnidirectional Flow in 360◦ Video via Siamese Representation. He has submitted a paper to ECCV (currently under review), which is one of the top conferences in computer vision. He is now working on another research project and plans to submit the 2nd paper in May to NeurlPS, another top conference in AI."
Burtchell, Brandon
Nominated by Dr. Chen Xiao
“Brandon worked on research in Smart & Connected Communities in the NSF REU program in 2021 and a paper has been written and is ready to submit.
Also, he worked very hard on his application to the NSF Graduate Research Fellowship program.”
Everman, Bradford
Nominated by Dr. Ziliang Zong
"Bradford Everman has been working on a research project in Evaluating and Reducing Cloud Waste and Cost - A Data-Driven Case Study from Azure Workloads. His work has
been recently published at the Journal of Sustainable Computing. He is working on another Green AI project, which will be submitted to the Green Computing Conference in June."
Fallin, William Alex
Nominated by Dr. Martin Burtscher
“Alex has created a very promising rectilinear Steiner-Tree heuristic (used in VLSI design for chip layout) that is not only effective but also fast and easy to parallelize. The heuristic is based on the optimal solution for 3-pin nets, which it intelligently applies to 3-pin subnets of larger nets, where a net is a set of locations on the chip that must be electrically connected. The GPU implementation is two orders of magnitude faster and only requires 0.38% more wire length on average than the most widely used approach. Moreover, Alex’s code supports nets with tens of thousands of pins whereas most preexisting implementations only support up to 256 pins. This work has been accepted for publication. Independently, Alex has also pioneered a brand new kind of source-code optimization that does not affect the running time of CUDA programs but significantly reduces the energy consumption (by up to over 10%) by minimizing bit flips on the memory bus. We have just finished writing up this work and will submit it to a conference in a few days.
Alex Fallin, Aarti Kothari, Jiayuan He, Christopher Yanez, Keshav Pingali, Rajit Manohar, and Martin Burtscher. “A Simple, Fast, and GPU-friendly Steiner-Tree Heuristic.” Proceedings of the 23rd IEEE International Workshop on Parallel and Distributed Scientific and Engineering Computing. May 2022.
One paper to be submitted in a few days.”
Ford, Blake
Nominated by Dr. Ziliang Zong
"Ford Blake has been working on research in software migration from the x86 architecture to the ARM architecture. He has recently published two research papers at the Journal of Sustainable Computing and the IEEE International Conference on Networking, Architecture, and Storage (NAS)."
Katrychuk, Dmytro
Nominated by Dr. Oleg Komogortsev
"I nominate Dima Katrychuk for his outstanding work on extraction of oculomotor plant characteristics and submitting a paper on this topic."
Liu, Yiqian
Nominated by Dr. Martin Burtscher
“Yiqian has pioneered a new class of benchmark suite (for testing program verification tools). The suite automatically generates meaningful buggy and bug-free variations of important code patterns found in irregular algorithms. Because the behavior of such codes is data dependent, the suite also automatically synthesizes inputs for running the codes and eliciting different behaviors. The resulting suite comprises over 1000 codes and an even larger number of inputs for each code, yielding several million possible experiments that can be executed, which is orders of magnitude beyond any other benchmark suite. This work has been accepted for publication at ISPASS’22. Yiqian is now further advancing the state of the art by designing a second benchmark suite based on important graph analytics algorithms. The suite automatically applies different parallelization styles to each algorithm, including combinations thereof where meaningful. For both of the two algorithms she has tried so far, her approach yields hundreds of different parallel GPU and CPU implementations. No related work exists that describes more than ten versions. Hence, I believe her work will not only usher in a new era of benchmark suite design but also allow researchers, for the first time, to study which parallelization styles work well together and what devices benefit from which combination the most.
Yiqian Liu, Noushin Azami, and Martin Burtscher. The Indigo Program-Verification Microbenchmark Suite of Irregular Parallel Code Patterns. Proceedings of the 2022 IEEE International Symposium on Performance Analysis of Systems and Software. May 2022.”
Lohr, Dillon
Nominated by Dr. Oleg Komogortsev
"I nominate Dillon Lohr for his outstanding work on eye movement biometrics and publishing a paper at TBIOMs."
Melnyk, Kateryna
Nominated by Dr. Oleg Komogortsev
"I nominate Kate Melnyk for her outstanding work on the discovery and documentation of entropy and chaos features as related to the eye tracking signal and working on a paper on this topic."
Raju, Mehedi
Nominated by Dr. Oleg Komogortsev
"I nominate Mehedi Raju on his outstanding work on eye tracking signal filtering and understanding and working on a paper on this."
Ramadan, Tarek
Nominated by Dr. Tanzima Islam
"Tarek is working on developing new representation learning techniques for performance analytics in HPC that drastically improves the accuracy of predictive performance models. Based on his research, he has already published a paper last year in a prestigious conference (~23% acceptance rate), submitted a journal, and will submit another paper in June. He is an MS student and will defend his thesis next week."
Roquebert Martinez, Ian
Nominated by Dr. Anne Ngu
“Ian started working on my research project this semester and as compared to other students, he is much more rigorous and thorough in his approach. He is more reliable in the delivery and we plan to publish the work that he did as a paper in May.”
Wesley, Kayla
Nominated by Dr. Martin Burtscher
“For her master’s thesis, Kayla has parallelized the classic LZ77 data compression algorithm using a brand new approach. Both her encoder and her decoder scale to massively parallel devices while being essentially work efficient. Nobody before her has been able to achieve this with the LZ77 encoder. We hope to be able to publish this work soon in a peer-reviewed venue.
First person to implement a scalable massively-parallel LZ77 encoder for GPUs.”