Vignesh Adhinarayanan is a Research Scientist at AMD Research and Advanced Development (AMD RAD) in Austin, TX. His work focuses on the architecture and systems design of large-scale AI inference, including high-bandwidth memory systems, GPU interconnection networks, and parallel execution strategies for large language models.
His research develops analytical models and system-level techniques to understand and improve the performance of modern AI workloads, with particular emphasis on distributed inference, KV-cache scaling, and the efficiency limits of mixture-of-experts architectures. By connecting model design with hardware and network behavior, his work aims to advance high-performance and energy-efficient computing architectures that enable the next generation of AI and scientific computing systems.
Select Recent Publications
2026
Vignesh Adhinarayanan and Nuwan Jayasena. "The qs Inequality: Quantifying the Double Penalty of Mixture-of-Experts at Inference." arXiv preprint arXiv:2603.08960. 2026. [PDF] [arXiv] [TL;DR]
2025
Vignesh Adhinarayanan, Bradford M. Beckmann, Wantong Li, Mohammad Seyedzadeh, Sergey Blagodurov, Derrick Aguren, and Hayden Hyungdong Lee. "Folded Banks: 3D-Stacked HBM Design for Fine-Grained Random-Access Bandwidth." In Proceedings of the 52nd Annual International Symposium on Computer Architecture, pp. 1819-1833. 2025. [PDF] [ACM]
Vignesh Adhinarayanan, and Wu-chun Feng. "Looking Back to Look Forward: 15 Years of the Green500." Computer 58, no. 1 (2025): 76-86. [PDF] [IEEE] [Computer]
2023
Gabriel H. Loh, Michael J. Schulte, Mike Ignatowski, Vignesh Adhinarayanan, Shaizeen Aga, Derrick Aguren, Varun Agrawal et al. "A research retrospective on amd's exascale computing journey." In Proceedings of the 50th Annual International Symposium on Computer Architecture, pp. 1-14. 2023. [PDF] [ACM]
2020
Vignesh Adhinarayanan. "Models and Techniques for Green High-Performance Computing." Virginia Tech, Blacksburg, VA, USA, 2020. [PDF] [VTWorks]
2016
Vignesh Adhinarayanan, Indrani Paul, Joseph L. Greathouse, Wei Huang, Ashutosh Pattnaik, and Wu-chun Feng. "Measuring and modeling on-chip interconnect power on real hardware." In 2016 IEEE International Symposium on Workload Characterization (IISWC), pp. 1-11. IEEE, 2016. Best Paper Award [PDF] [IEEE] [Computer]