Welcome Remarks

Speaker: Dionisios Pnevmatikatos

Title: Reconfigurable technologies in HPC and data-centers. Challenges and opportunities

Abstract: Reconfigurable technology has been successfully showcased in several computationally intensive applications that exploit the underlying adaptability to extract performance. When applying this technology to more general environments (HPC and/or data-centers), the necessary tradeoffs and performance tuning are more challenging. In this talk I will describe our the current state of play in the field and our activities and progress towards the two settings. For the HPC environment we are building a set of open source libraries for typical HPC kernels, starting from basic ones (BLAS L1), and gradually moving towards the more involved, interesting and difficult ones: (e.g. BLAS L2 & L2, SpMv, Jacobi, LU decomposition). For deploying accelerators at the data center we are building a substrate able to flexibly support mutli-tenancy while ensuring isolation of the data accessed by hte accelerators. Scalabiity in the number of accelerators and support of high-performance memory systems (with multiple channels) is supported via a properly dimensioned NoC.

Bio: Dionisios Pnevmatikatos (BSc U. Crete ’89, PhD UW-Madison 95) is Professor at the School of ECE of the National Technical University of Athens and a member of the Computing Systems Lab (CSLab). Before that he was a Professor of ECE at the Technical University of Crete (2000-2019) and a research associate with FORTH-ICS (1997-2019).His research interests are in the broader area of Computer Architecture, where he investigates the design and implementation of high-performance and cost-effective accelerated, heterogeneous parallel/rack-scale systems.He is the coordinator of the AERO ongoing project and he has participated and coordinated many EU-funded and several national-funded projects in the past. He has also been in the Program Committee and served as a Program (co)Chair in many prestigious conferences in his field.

Speaker: Vasileios Karakostas

Title: Optimizing the memory access path across the computing stack

Abstract: The performance gap between accessing data and processing them has been a long-standing problem. Accessing data consists of two steps: (i) performing address translation to identify where the data lives, and (ii) retrieving the data itself.  However, two critical trends in modern computing systems prevent closing or even widen this performance gap even further. First, the memory resources become larger to satisfy the immense demand of modern applications for processing increasingly large datasets, stressing address translation. Second, the introduction of persistent memory constitutes data retrieval much faster compared to traditional devices, revealing new sources of overheads. In this talk, we will discuss the challenges and opportunities that these trends introduce, and will present some concepts and approaches for improving memory accesses across the computing stack.

Bio: Vasileios Karakostas is an Assistant Professor at the Department of Informatics and Telecommunications of the National and Kapodistrian University of Athens. He is a member of the Computer Architecture Lab. Before joining the University of Athens, he was a postdoctoral researcher at the Computing Systems Lab of the School of Electrical and Computer Engineering of the National Technical University of Athens. He received his PhD degree in Computer Architecture from Universitat Politècnica de Catalunya and Barcelona Supercomputing Center in 2016, and his Diploma in Electrical and Computer Engineering from the National Technical University of Athens. His research interests lie in the areas of computer architecture, memory systems, hardware/OS interaction, operating systems, resource management, and parallel systems. His research results have been published and awarded at international computer architecture and systems conferences. He has participated in several European research projects, and has served as reviewer in numerous conferences and journals. He is a member of IEEE, ACM, and HiPEAC.