VLSI-SoC 2026

34th IFIP/IEEE International Conference
on Very Large Scale Integration SoC
Limassol (Cyprus), 11-14 October, St. Raphael Resort.

KEYNOTES


KEYNOTE SPEAKERS


Prof. Giovanni De Micheli

Prof. Giovanni De Micheli

Affiliation: EPFL, Switzerland

Email: giovanni.demicheli@epfl.ch

Website: si2.epfl.ch/~demichel/

Bio

Giovanni De Micheli is a research scientist in electronics and computer science. He is credited for the invention of the Network on Chip design automation paradigm and for the creation of algorithms and design tools for Electronic Design Automation (EDA). He is Professor and Director of the Integrated Systems Laboratory and Scientific Director of the EcoCloud center at EPFL Lausanne, Switzerland. Previously, he was Professor of Electrical Engineering at Stanford University. He was Director of the Electrical Engineering Institute at EPFL from 2008 to 2019 and program leader of the Swiss Federal Nano-Tera.ch program. He holds a Nuclear Engineer degree (Politecnico di Milano, 1979), a M.S. and a Ph.D. degree in Electrical Engineering and Computer Science (University of California at Berkeley, 1980 and 1983).

Keynote Title

Sustainable and Secure Computing: the Last Frontier

Abstract

This keynote explores the convergence of sustainability and security in modern computing, highlighting key challenges and emerging approaches for building energy-efficient and resilient systems.


Prof. Valeria Bertacco

Prof. Valeria Bertacco

Affiliation: University of Michigan, USA

Email: valeria@umich.edu

Website: web.eecs.umich.edu/~valeria/

Bio

Valeria Bertacco is the Mary Lou Dorf Collegiate Professor of Computer Science and Engineering, and Arthur F. Thurnau Professor of Engineering at the University of Michigan. Her research interests are in the area of computer system design. Throughout her career, she has contributed novel solutions in design validation and reliability, hardware-security assurance, and the design of specialized architectures for graph algorithms and machine learning. From 2018 to 2023, Prof. Bertacco was the Director of the Applications Driving Architectures (ADA) Research Center, funded by a consortium of semiconductor companies and the Defense Advanced Research Projects Agency (DARPA). The Center engaged the work of 10 US academic institutions and over 130 Ph.D. student researchers. Currently, she leads the MAVERIC collaborative, the University of Michigan's initiative to advance semiconductor research and education. At Michigan, she also serves as the Vice Provost for Engaged Learning, supporting all international partnerships and co-curricular engagements. In the international space, her initiatives have included promoting collaborations with African institutions, through initiatives like Africa Week, UMAPS and AURA. Domestically, the primary partners reside within the State of Michigan, and especially within the City of Detroit. In her Vice Provost role, Prof. Bertacco facilitates the work of nine central units, whose goals range from promoting environmental sustainability to supporting journalists, to promoting the arts in research universities, and to increasing the participation of gender minorities in STEM fields. She received her M.S. and Ph.D. degrees in Electrical Engineering from Stanford University. Prof. Bertacco joined the University of Michigan in 2003, after working as a research engineer for Systems Science Inc., and Synopsys from 1997 to 2001.

Keynote Title

Envisioning SoC Design with an Army of Agentic Minions.

Abstract


Prof. Freddy Gabbay

Prof. Freddy Gabbay

Affiliation: Hebrew University of Jerusalem, Israel

Email: freddy.gabbay@mail.huji.ac.il

Website: freddygabbay.huji.ac.il

Bio

Freddy Gabbay is a Professor and head of the VLSI Systems Lab at the Institute of Applied Physics at the Hebrew University of Jerusalem. His research interests span VLSI design, microelectronics, computer architecture, machine learning, and domain-specific accelerators. Throughout his career, Prof. Gabbay has contributed to the design and development of advanced semiconductor systems, combining architectural innovation with practical implementation in high-performance computing and communication devices. Prof. Gabbay's professional experience bridges academia and industry, with over two decades of leadership in both domains. He began his career in 1998 as a researcher at Intel's Microprocessor Research Lab. He then joined Mellanox Technologies, where he played a key role in the architecture and ASIC design of high-performance switching products. In 2003, he moved to Freescale Semiconductor as a senior design manager, leading the development of baseband ASIC solutions. He later returned to Mellanox Technologies (now part of NVIDIA) in 2012, where he served as Vice President of Chip Design, overseeing large-scale silicon development programs. In addition to his industrial leadership, Prof. Gabbay is actively engaged in the academic and professional community. He serves as an Associate Editor for IEEE Computer Architecture Letters and as Chairman of the IEEE Computer Society Israel Chapter. He holds 19 patents and is a Senior Member of IEEE. Prof. Gabbay received his B.Sc., M.Sc., and Ph.D. degrees in Electrical Engineering from the Technion - Israel Institute of Technology.

Keynote Title

Do We Really Need All the Bits? Value-Driven Approximate Computing for AI Accelerators

Abstract

Generative AI is reshaping computing, enabling breakthroughs across language, vision, and autonomous systems. However, these models come at a steep cost: they are increasingly limited by compute, memory bandwidth, and energy, pushing modern hardware systems toward fundamental scalability barriers. This keynote presents a value-driven perspective on approximate computing, leveraging the inherent resiliency of deep neural networks to rethink how AI systems are designed. Instead of treating all data uniformly, we prioritize computation and data movement based on their contribution to model accuracy. This enables aggressive optimizations through quantization, decomposition, dynamic approximation, and compression. We will showcase a set of hardware–software co-designed techniques that significantly reduce memory traffic, improve arithmetic intensity, and enhance overall system efficiency. Through case studies, we demonstrate substantial gains in performance, energy efficiency, and memory utilization, while maintaining high model accuracy. Finally, we discuss how value-driven approximate computing reshapes the design space of next-generation AI accelerators, highlighting the central role of co-design in overcoming the memory and energy bottlenecks of large-scale generative AI systems.

INDUSTRIAL KEYNOTES


Victor Grimblatt

Victor Grimblatt

Affiliation: Synopsys, Vitacura, Chile

Email: victor.grimblatt@synopsys.com

Website: scholar.google.com/citations?user=lj8W9R4AAAAJ

Bio

Victor Grimblatt has an engineering diploma in microelectronics from Institut Nationale Polytechniquede Grenoble (INPG - France) and an electronic engineering diploma from Universidad Tecnica Federico Santa Maria (Chile). He got his PhD on Electronics in 2021 from University of Bordeaux. He is currently R&D Executive Director and General Manager of Synopsys Chile. He has published several papers in IoT, EDA, Smart Agriculture, Climate Change, and embedded systems development. From 2012 to 2024 he was chair of the IEEE Chilean join chapter EDS/CASS/SSCS. He has been part of several conferences TCP (ISCAS, ICECS, LASCAS, VLSI SoC) and Steering Committees. He was member of the IEEE CASS Board of Governors for the period 2021 - 2023. He founded the Electronics for Agrifood SIG at CASS and chaired it until 2024. He was Chair of LASCAS Steering Committee from 2018 to2022. He was CASS representative at the IEEE Climate Change TAB. He is member of the IEEE CASS Board of Governors for the period 2025 - 2027.He was President of the Chilean Electronic and Electrical Industry Association (AIE) from 2017 to 2021.From 2006 to 2008 he was member of the “Chilean Offshoring Committee” organized by the Minister of Economy of Chile. In 2010 he was awarded as “Innovator of the Year in Services Export”. In 2022 he was awarded as “IEEE/AIE Best Engineer” in Chile. In 2023 he was awarded as IEEE R9 Outstanding Engineer”. In 2024 he was awarded as “2024 CASS Meritorious Service Award”. Victor's research areas are EDA (Electronic Design Automation), Climate Change, and IoT for Smart Agriculture.

Keynote Title

Low Power IC Design for a Sustainable World

Abstract

In an era where sustainability is essential, designing low-power integrated circuits (ICs) plays a crucial role in reducing energy consumption and mitigating environmental impact. This talk examines innovative strategies in low-power IC design to meet the increasing demand for energy-efficient electronics, from IoT devices to large-scale computing systems. By utilizing advanced techniques such as subthreshold operation, power gating, clock gating, and dynamic voltage scaling, we outline a framework for optimizing performance while minimizing power dissipation. The discussion emphasizes the interplay between circuit-level innovations, such as the strategic use of clock gating to reduce dynamic power in idle states, process technology advancements, and system-level integration, highlighting their collective contribution to sustainability. Through simulations and case studies, we demonstrate how these approaches can significantly lower the carbon footprint of electronic systems, paving the way for greener technology solutions that align with global sustainability goals.


Odysseas Economides

Odysseas Economides

Affiliation: HardwareX Engineering, Cyprus

Email: odysseas@robo.com.cy

Website: www.uclancyprus.ac.cy/academic/odysseas-economides/

Bio

Odysseas Economides is a Cypriot electrical engineer and hardware entrepreneur focused on bridging the gap between research and industry through embedded systems, intelligent electronics, and robotics. He is the founder and Managing Director of Robo Educational & Research Robotics and HardwareX Engineering, where he leads the development of low-power electronics, IoT sensing platforms, edge AI systems, and custom embedded products from rapid prototyping through to production-ready deployment. His work spans education, research, and industry, with a strong emphasis on translating innovative ideas into practical, manufacturable hardware solutions. A SEMICON Europa “20 Under 30” awardee and NASA Space Apps People's Choice winner, Odysseas is also active in Cyprus' innovation ecosystem through mentoring, workshops, and STEM initiatives that empower teams to turn ideas into real, working technology.

Keynote Title

Bridging the gap between research and industry through embedded systems

Abstract

This keynote explores how embedded systems enable the translation of research innovations into real-world industrial applications, highlighting challenges and strategies for effective technology transfer.


Prof. Wolfgang Ecker

Prof. Wolfgang Ecker

Affiliation: Technical University of Munich, Germany

Email: wolfgang.ecker@infineon.com

Website: www.ce.cit.tum.de/en/eda/persons/wolfgang-ecker/

Bio

Wolfgang Ecker is Distinguished Engineer at Infineon and Professor at Technical University of Munich. His research and innovation focus lie on digital system modeling, digital design automation, SoC architectures, embedded AI and AI for design automation. Wolfgang Ecker published over 250 papers, received seven publication awards and has been granted with the German EDA achievement award. He is member of Acatech, the German Academy of Science and Engineering and has been member of the AI commission of inquiry of the German Government.

Keynote Title

What is the IP Reuse Trap for Digital Hardware, and how can it be escaped?

Abstract

Despite its theoretical high productivity impact, IP reuse (intellectual property reuse) in semiconductor design—the practice of using pre-designed digital hardware blocks (like processor cores, interface components, …) in new System-on-a-Chip (SoC) projects—frequently fails to work as expected in industry due to several causes:
- High integration complexity, including reading and understanding (often incomplete) documentation, physical dependencies between IP and environment, and finding the right configuration
- The IP only mostly fits, but changes in the IP are needed due to lacking features, adaptation to meet different PPA constraints, and to use the IP in different technologies.
- IP development is costly and takes more time for development than non-reusable components.
Reasons are manifold:
- Interfaces even to widely used buses like AHB allow many variations, resulting in a very high space of options.
- RTL synthesis capabilities are limited; QoS—quality of source—is still the dominating principle. Different PPA constraints require re-architecting and re-implementing to satisfy them.
- Also, different power-down, security, and safety constraints require modifying IPs.
- Lastly, a verified IP must be verified and, more importantly, validated in its environment: Does the IP provide the intended functionality to the overall system?
The talk addresses these challenges from semiconductor industry's point of view. It first introduces these and some other challenges. An elaboration of methods dealing with the challenges, such as platform-based design, IP generation as well as hardware generation languages, and automated interface adoption, follows. It closes with the discussion of whether we are reusing the right things and in the right way.

PANEL SESSIONS


Title

AI as the Chip Designer: Evolution or Illusion?

Abstract

Artificial intelligence is entering System-on-Chip design at every level, from automating repetitive engineering tasks to supporting architectural exploration, optimization, and verification. This panel will discuss whether AI can deliver meaningful productivity gains and open new design possibilities, or whether it also risks weakening engineering expertise, trust, and accountability. Bringing together perspectives from academia, startups, and industry, the discussion will ask whether AI marks a true paradigm shift in SoC engineering or remains a powerful tool whose impact depends on human oversight, responsibility, and judgment.

Panelists

Prof. Christos Sotiriou, University of Thessaly Prof. Christos Sotiriou, University of Thessaly
Dr. Sandro Belfanti, Chipmind Dr. Sandro Belfanti, Chipmind
Dr. Valerio Tenace, PrimisAI Dr. Valerio Tenace, PrimisAI
Dr. Frank K. Gurkaynak, ETH Dr. Frank K. Gurkaynak, ETH

TUTORIALS


Tutorial Title

TO BE ANNOUNCED

Speaker

TBA

Topic

TO BE ANNOUNCED

TECHNICAL PROGRAM


Below is the detailed schedule for the VLSI-SoC 2026 Conference Program.


Academic keynote Industrial keynote Panel / reception Poster display Special session Regular session Break / admin
Sunday, 11 October 2026 – Tutorials Day
08:00–09:00Registration
09:00–12:00Morning TutorialPlenary · 3 hrs · TBA
12:00–13:00Lunch Break
13:00–16:00Afternoon TutorialPlenary · 3 hrs · TBA
Monday, 12 October 2026 – Conference Day 1
08:00–08:30Registration
08:30–09:00Welcome Ceremony
09:00–10:00 Academic Keynote 1 – Sustainable and Secure Computing: the Last Frontier Prof. Giovanni De Micheli · EPFL, Switzerland · Plenary · 60 min
10:00–10:20Coffee Break + PhD & Student Forum20 min
10:20–11:50 RS1 – AI/ML Hardware Architectures I Track 1 · 4 papers + 2 poster presentations · 90 min
  • 141 CIM-Blocks: A Synthesizable Multi-Precision Compute-in-Memory CGRA with Bit-Serial Data Reuse
  • 18 At-the-Roofline Sparse Tensor Contractions on Vector Processors for Transformer Inference
  • 24 Optimizing ML Workload Partitioning between CPUs and CIM Accelerators for Heterogeneous Computing
  • 71 Hardware-Aware Vision Transformer Deployment on a Convolution-Centric AI SoC

  • Poster presentations
  • T153 S4oP: Operator-level Pruning of Structured State Space Models for Resource-Constrained Devices
  • T234 Physically-Aware Preemptive Virtual Channels for Deadlock-Free AXI Networks-on-Chip
RS6 – System Verification, Test & Dependability Tracks 11+12 · 4 papers + 1 poster presentation · 85 min
  • 86 Hardware Support for Statistical Methods Applied to Hardware-Software Verification and Debugging
  • 94 Counterfactual Exploit Validation on CHERI-enabled RISC-V Using Virtual Prototypes
  • 65 RISCar: RISC-V In a Simulated Car for DNN Training and Deployment in AD Systems
  • 76 Energy-Aware Fast and Accurate Design Space Exploration of Near-Memory Computing

  • Poster presentation
  • T1131 AI-based Automated HDL Validation Using Abstract Syntax Tree and Signal Trace Analysis
11:50–12:30 Industrial Keynote 1 – What is the IP Reuse Trap for Digital Hardware, and How Can It Be Escaped? Prof. Wolfgang Ecker · Infineon / TU Munich, Germany · Plenary · 40 min
12:30–13:30Lunch Break60 min
13:30–15:00 RS3 – Digital Design & EDA I Track 4 · Synthesis & Optimization · 4 papers + 2 poster presentations · 90 min
  • 149 Automated RTL Complexity Estimation with Synthesis-Validated Optimization and Programmatic Code Transformation
  • 90 SPFD-Based Resynthesis for Dual-Output LUT Networks
  • 110 No Tree Required: Predicting Post-CTS Clock Timing from Placement Features Alone
  • 133 Allocating a Unified Domain Platform Following Market Analysis Using ProdDSE

  • Poster presentations
  • T4116 A Self-Recoverable Nonvolatile Magnetic Latch with High-Speed SEU-Tolerant Backup Module
  • T472 A Scalable End-to-End Framework for Multi-Objective Design Space Exploration: Application to AI Accelerators
RS7 – Embedded Systems & Low-Power Design Tracks 5+7 · 4 papers + 2 poster presentations · 90 min
  • 1 A Silicon-Validated High-Frequency RISC-V SoC for Extreme-Temperature Applications
  • 147 Interpretable Fuzzy Control for Lightweight Cache Replacement and Prefetch Throttling in RISC-V SoCs
  • 21 An event-Driven ASK Demodulator Based on Slope Measurements for low-Power Applications
  • 146 Separate Read-Write Differential 11T SRAM-based In-Memory Computing Architecture

  • Poster presentations
  • T538 R5-Link: Enhancing RISC-V Multi-Core Efficiency with Hardware Message Passing Channels in a 2D Mesh Network
  • T6118 Surrogate-assisted DTCO for 1T1C FeMFET Bitcells: A Comparative Study on Model Choice and Data Sampling
15:00–15:20Coffee Break + PhD & Student Forum20 min
15:20–16:40 Special Session 180 min · TBA Special Session 280 min · TBA
17:30–19:00 Welcome Reception Welcome Cocktail is the first social gathering between all conference delegates and it will take place at the Venue Hotel. It will be a relaxing evening during which delegates will have the opportunity to talk to colleagues and peers, while enjoying local drinks and ample canapés, with a view of the calming waters of the Mediterranean Sea.
Tuesday, 13 October 2026 – Conference Day 2 · Social Event from 17:00
09:00–10:00 Academic Keynote 2 – Envisioning SoC Design with an Army of Agentic Minions Prof. Valeria Bertacco · University of Michigan, USA · Plenary · 60 min
10:00–10:20Coffee Break20 min
10:20–11:50 RS2 – AI/ML Architectures & Computing Paradigms Tracks 1+3 · 4 papers + 2 poster presentations · 90 min
  • 30 Chip-Agnostic Hardware-Aware Training for ADC-Efficient BNNs in SOT-MRAM Crossbars
  • 104 PreDSE: Predictive Design Space Exploration for FPGA CNN Dataflow Accelerators
  • 37 A Noise-based Obfuscation Technique for Safe and Secure In-Memory AI inference
  • 23 New Obfuscation Strategies for Approximate Adders

  • Poster presentations
  • T366 Reducing the Footprint of Approximate Ternary Neural Networks via Per-Neuron Input Permutations
  • T5131 Deployment of a Safety-Critical, Distilled YOLOV8m Vision System on a Dual-Core, 1MB, BLE-Connected Programmable Logic Controller
RS5 – Hardware Security & Testing Tracks 8+12 · 4 papers + 1 poster presentation · 85 min
  • 56 A Differential Power Analysis Attack Exploiting Early Propagation Effect in Dual-Rail Pre-Charge Logic Circuits
  • 98 Self-Interpretable Hardware Trojan Detection on Gate-Level Netlists with Sufficient and Necessary GNN Explanations
  • 109 SW-PUF: An ML-Resistant Strong PUF Architecture Using Weak-PUF Entropy and Ascon-Hash Obfuscation
  • 105 JTAG You're It: A Scalable JTAG Network in Chiplet Arrays via Fabric-Reuse

  • Poster presentation
  • T8111 Side-channel aware design of FeFET based memory for cryptographic SBox implementation
11:50–12:30 Industrial Keynote 2 – Low Power IC Design for a Sustainable World Victor Grimblatt · Synopsys, Chile · Plenary · 40 min
12:30–13:30Lunch Break60 min
13:30–14:15 Panel – AI as the Chip Designer: Evolution or Illusion? Plenary · 45 min · Frank K. Gurkaynak (ETH Zürich) · Sandro Belfanti (Chipmind)
14:15–15:35 Special Session 380 min · TBA Special Session 480 min · TBA
15:35–15:55 Coffee Break + Poster Display 20 min · All 10 posters displayed simultaneously · Conference day ends 15:55
16:15–19:00 Tour and Conference Dinner We will depart from the venue hotel in air-conditioned busses with licensed tour guides for a walking tour (weather permitting) across all the historical sites and landmarks within the center of Limassol. We will then head to the conference dinner venue at a local tavern, where participants will enjoy an array of authentic Cypriot dishes complimented with local drinks, desserts and traditional entertainment.
Wednesday, 14 October 2026 – Conference Day 3
09:00–10:00 Academic Keynote 3 – Do We Really Need All the Bits? Value-Driven Approximate Computing for AI Accelerators Prof. Freddy Gabbay · Hebrew University of Jerusalem, Israel · Plenary · 60 min
10:00–10:20 Coffee Break + Poster Display 20 min · All 10 posters displayed simultaneously
10:20–11:40 RS4 – Digital Design & EDA II Tracks 3+4 · Timing, Async & FPGA · 4 papers · 80 min
  • 81 A Robust Asymmetric Delay Cell for High-Performance 4-Phase Bundled-Data Circuits
  • 44 AutoSDC: Correct-by-Construction Timing Constraint Generation via Hybrid Neuro-Structural Training, Multivariate Quality Prediction, and Recursive Signoff-Driven Refinement
  • 129 Exploring the Impact of 2D Convolutional Layer Hyperparameters in FPGA Implementation
  • 79 Efficient Scalable Approximate Multipliers via Significance-Driven Partial Product Removal
RS8 – Communications, Sensing & Emerging Technologies Tracks 2+6+9 · 4 papers · 80 min
  • 40 Community-Based ILP for Application Mapping and Deadlock-Free Routing on Large NoCs
  • 15 White Rabbit–Enabled Deterministic Triggering for Bi-Static ISAC in 6G
  • 91 Selective frame processing for accelerating visual SLAM
  • 54 Impact of Programming Pattern Strategy on ReRAM Relaxation and Retention Stability
11:40–12:20 Industrial Keynote 3 – Bridging the Gap Between Research and Industry Through Embedded Systems Odysseas Economides · HardwareX Engineering, Cyprus · Plenary · 40 min
12:20–13:20Lunch Break60 min
13:20–14:40 Special Session 580 min · TBA Special Session 680 min · TBA
14:40–15:00Coffee Break20 min
15:00–16:00 Closing Ceremony & Awards Conference ends 16:00

Program Information

Below you can view or download the Program Flyer and Detailed Program files.


Program Flyer
Detailed Program

SOCIAL PROGRAM


Info will be posted. (TBA)