Leadership Update at ICL
After years of contributions to ICL’s research and community, Mark Gates departs for a new position at AMD, and we welcome Heike Jagode and Ahmad Abdelfattah into interim leadership roles. After serving as Interim Director since 2024, Mark is joining AMD as Principal Software Developer, GPU Dense Linear Algebra. Mark came to ICL in 2011 as a post-doctoral researcher after completing his PhD at the University of Illinois, Urbana-Champaign, and over more than a decade became a valued colleague, mentor, and leader. We extend our sincere thanks to Mark for his many contributions, and we wish him every success in his new position.
Stepping into the role of Interim Director is Heike Jagode, Research Associate Professor at ICL and a fixture of the lab since 2008. Heike has led the ICL Performance Group since 2013 and has been a core contributor to our work in high-performance computing, with expertise in performance analysis, tuning, and the efficient use of advanced architectures. Supporting Heike as Interim Associate Director is Ahmad Abdelfattah, Research Assistant Professor at ICL and one of our research directors in the Linear Algebra group. Ahmad’s work spans high-performance computing, parallel numerical algorithms, and general-purpose GPU computing, and he currently serves as principal investigator of the MAGMA library.
SC25
The SC25 annual conference landed in St. Louis, Missouri from November 16–21, drawing more than 16,500 attendees and a record 559 exhibitors. The week kicked off with keynote speaker Thomas Koulopoulos offering a forward-looking view on the forces shaping our digital future and the opportunities and challenges of AI-driven progress. NVIDIA CEO Jensen Huang made a surprise appearance at the SC25 opening gala, sharing the company’s latest advances and even giving away DGX Spark AI systems to lucky attendees. From the exhbit floor, strength athlete Hafþór “The Mountain” Björnsson set a new data-transfer speed record on the show floor, moving hundreds of petabytes of data in a matter of seconds in a live demonstration of tomorrow’s storage performance.
A contingency of research staff and students represented ICL at the event and hosted the University of Tennessee booth in partnership with NICS and AI-Tenn. Throughout the week, ICL staff, students, and alumni contributed in technical sessions, student research, alumni networking, and community dialogues.
ICL Sticker Swag
Visitors to the University of Tennessee booth this year were greeted with a fun variety of sticker swag. In addition to classic ICL, TOP500, and Ginkgo stickers, we debuted new logos for four of ICL’s flagship initiatives: PAPI, PaRSEC, MAGMA, and SLATE.
SC25 Benchmark Updates
SC25 saw the release of the 66th edition of the TOP500 list as well as accompanying discussions of emerging benchmarks that reflect modern HPC workloads.
- The top performance slot remains held by El Capitan — an HPE Cray EX255a system at LLNL — with a sustained HPL throughput of 1.809 Exaflops and a leading HPCG result of 17.41 PFlops.
- The U.S. DOE systems Frontier (ORNL) and Aurora (Argonne) remain firmly in the Top 3, with Aurora posting 1.012 Exaflops on HPL.
- New this year, Europe’s first full-scale exascale system, JUPITER — housed at Forschungszentrum Jülich — officially reached 1.000 Exaflops, earning it the #4 spot and formally bringing Europe into the exascale club.
As modern HPC systems continue to evolve, supplementary benchmarks like HPCG and HPL-MxP provide a broader view of performance beyond traditional HPL. In the HPCG ranking released with the November 2025 results at SC25, El Capitan leads with 17.41 petaFLOP/s, followed by Fugaku and Frontier, reflecting sustained strength on conjugate-gradient workloads that stress memory, communication, and real-world application patterns. Meanwhile, HPL-MxP again shows El Capitan at the top with 16.68 exaFLOP/s, with other exascale systems Aurora and Frontier rounding out the podium.
ICL Student Spotlight
Congratulations to ICL student Tatiana Melnichenko, who presented work at SC25 which garnered recognition in multiple venues at SC25. Tatiana worked with a team from ORNL led by William Godoy to do a comprehensive study on the Mojo programming language vendor-neutral GPU capabilities for HPC science workloads. At the 12th Workshop on Accelerator Programming and Directives (WACCPD 2025), the team’s paper on this topic titled Mojo: MLIR-based Performance-Portable HPC Science Kernels on GPUs for the Python Ecosystem received a Best Paper Award. Tatiana also presented a poster on the same topic at SC25, where it was selected as a Best Poster Finalist in the ACM Student Research Competition.
Rabab Alomairy Recognized with Early Career Researcher Award
We also congratulate ICL alum Rabab Alomairy, who was honored as part of the SC25 Awards program with the IEEE Computer Society Technical Committee on High Performance Computing Early Career Researchers Award for Excellence in HPC. This recognition highlights her outstanding contributions to high-performance computing research early in her career and her impact on the broader HPC community.
RandNLA in HPC Birds-of-a-Feather
At SC25, ICL’s Max Melnichenko led a Birds-of-a-Feather session titled “Randomized Numerical Linear Algebra in HPC: Toward a Sustainable, Scalable Software Ecosystem.” The session brought together researchers and practitioners to talk about how RandNLA — Randomized Numerical Linear Algebra — is evolving from theoretical curiosity to practical, high-performance infrastructure across architectures, and what it will take to build sustainable software libraries and standards that support its broader adoption in HPC. For those interested in this emerging area, Max produced a handout which outlines RandNLA concepts discussed at the BoF.
ICL Alumni Dinner
ICL’s annual alumni dinner at Carmine’s Steak House in St. Louis brought together 27 alumni, current members, and friends of ICL for an enjoyable evening of food, conversation, and community.
Jack Dongarra’s “Responsibly Reckless Algorithms” Talk at ADIA Lab Symposium 2025
Jack Dongarra’s “High-Performance Computing and Responsibly Reckless Algorithms” talk explores how the HPC landscape is evolving in the era of AI and heterogeneous architectures. The edition of this talk delivered at the AI for Climate Science – ADIA Lab Symposium 2025 held in October 2025 in Abu Dhabi, is now available to watch online.
ICL Office Celebrates Homecoming 2025
ICL office staff transformed the main office into a Big Orange celebration zone for UT’s 2025 Homecoming office decor contest.
Jack Dongarra Honored with 2025 NEC C&C Prize
We are proud to share that Jack Dongarra has been named a recipient of the 2025 NEC Computer and Communications (C&C) Prize, an award that recognizes outstanding contributions to research and development in the integration of computing and communications technologies and their impact on society. This year’s C&C Prize honors two distinct groups: one recognizes the team responsible for the invention, practical implementation, and global spread of the QR Code — a technology used worldwide in logistics, payments, mobile authentication, and more — and the other celebrates Jack’s decades of pioneering work in high-performance scientific computing, including foundational numerical libraries and performance benchmarking infrastructure that have shaped the field. The formal award ceremony took place November 26 in Tokyo.
Congratulations
Congratulations to Our Latest PhD Graduates
We are delighted to congratulate two ICLers who officially received their PhDs last week, a testament to ICL’s long-standing tradition of mentoring and fostering doctoral researchers whose work advances high-performance computing. Daniel and Max now join a distinguished legacy of ICL graduates making an impact in research and beyond.
Dr. Daniel Barry completed his PhD under the supervision of Heike Jagode and in collaboration with Anthony Danalis. His dissertation, “Automated Classification and Verification of Performance Counters,” focused on methods for systematically organizing and validating hardware performance counter data, enabling more effective performance analysis and optimization of scientific applications on HPC systems. Having officially graduated, Daniel will be staying on with ICL as a Research Scientist. We look forward to his continued impact on our projects and collaborations.
Dr. Max Melnichenko completed his PhD under the guidance of Jack Dongarra, with a dissertation titled “Responsible Recklessness: Why High Performance Computing Must Embrace Randomized Numerical Linear Algebra“. Max also acknowledges the valuable mentorship of Piotr Luszczek, who served as an informal advisor throughout his graduate work. Max has announced that he will begin a postdoctoral appointment at the International Computer Science Institute (ICSI) and the UC Berkeley Department of Statistics starting January 2026, where he will work under the supervision of Dr. Michael Mahoney. We extend our warmest congratulations to Max as he embarks on this exciting next step.
Adoption Day
Celebrating Harrison Flanigan! On November 18th, Dayle Flanigan officially welcomed her son, Harrison, through adoption.
Recent Releases
Ginkgo 1.11.0
The Ginkgo team is proud to announce the new Ginkgo minor release 1.11.0.
Ginkgo is a high-performance numerical linear algebra library for many-core systems, with a focus on solution of sparse linear systems. It is implemented using modern C++, with GPU kernels implemented for NVIDIA, AMD and Intel GPUs.
The highlights of the new release include:
- New interface for sparse matrix-matrix multiplication and addition. The new interface can yield speedups, in the case of repeated computations where the matrix values may change, but the matrix structure (i.e. position of non-zero entries) doesn’t.
- Performance improvements for the distributed SpMV and pipelined CG solver.
- Better support for ARM platforms and support for ROCm 7 and CUDA 13.
The release contains many more additions, improvements, and bug fixes. A full list of the release changes can be found at: https://github.com/ginkgo-project/ginkgo/releases/tag/v1.11.0
Interview
Waldhys Rodoli
Tell us a little about yourself. Where are you from and what was your professional background before joining ICL?
I am originally from the Dominican Republic. My family moved to the States, specifically Miami, when I was in elementary; Miami is home for me. Prior to joining ICL, I worked in a few industries, e.g., legal, local government, and higher education. My time in higher education was mainly focused on financial and budget management, and human resources.
How did you first learn about ICL and what is your role in the group?
I had no idea ICL existed until I saw the job posting. I am glad I did because this center does some amazing stuff, even if I cannot understand most of the projects; they go over my head. I am the business manager at ICL. I get to work directly with both Tonya and Dayle, who are amazing, and I work closely with the director of ICL. We handle the administrative side of ICL.
What do you enjoy most about working at ICL?
I have really enjoyed the people at ICL, both past and present. At ICL35, I got to meet many of the alumni and hear great stories (professional and personal) that shine a good light on ICL. ICL has a good work atmosphere that feels like family. I also enjoy what I do. I love managing sponsored projects and leading the A-Team (administrative group).
If you could have any job in the world, what would it be and why?
I am already doing it. I am a pastor at a small bilingual Baptist church.
What are some of your interests/hobbies outside of work?
I do not have many hobbies. I like to read fiction and non-fiction, such as history and theological books. Although it is not a hobby, I do enjoy eating. I feel that food brings people together and it has the power to create profound relationships.
What is something that might surprise people to know about you?
I think people would be surprised to know that I dabble on the piano. While I am not an expert, I experiment with harmony and simple melodies; it’s therapeutic for me.
