Si2 sponsored an insightful panel discussion at The First IEEE International Workshop on LLM-Aided Design (LAD’24) on June 28, held at the IBM Almaden Research Center with over 200 attendees. The conference organizers praised the panel for giving the mostly academic audience a reality check on scaling and testing needed to move Generative AI from theory to production for the semiconductor industry. Si2 was invited to sponsor a panel again next year.
The panel discussed their positions on:
- Generative AI for semiconductor from architecture to fab: A pipedream? Or ready for prime time?
- How is Gen AI useful to us in semiconductor design?
- What is a successful AI-centric mindset to approach Design Automation?
- What are the challenges of using this technology in production, not just in a PoC?
- What are the implications of using LLM from a security perspective?
- What is the direction of large models vs limited, tuned models – how does this effect the methodology flow?
A White Paper from the AI/ML in EDA SIG is planned, summarizing the panel discussion and the next areas of focus in Generative AI where standardization or common frameworks are needed to support industry/academic cooperation.
Industry and University Panelists
Akhilesh Kumar is a Senior Principal R&D Engineer at Ansys leading the AI/ML and Generative AI initiatives for EDA solutions. Additionally, he is the co-chair of Si2 AI/ML SIG. Dr Kumar received his PhD and MS in Electrical and Computer Engineering from the University of Waterloo, Canada, and Bachelor of Technology in Electrical Engineering from the Indian Institute of Technology, Roorkee, India.
Ioannis Savidis received his Ph.D. degree in electrical and computer engineering from the University of Rochester, Rochester, NY. He joined the Department of Electrical and Computer Engineering, Drexel University, Philadelphia, PA, USA, in 2013, where he is currently an Associate Professor and directs the Integrated Circuits and Electronics (ICE) Design and Analysis Laboratory. Dr. Savidis is a senior member of the Institute of Electrical and Electronic Engineers. He is a recipient of the 2018 National Science Foundation Early Faculty (CAREER) Award, and a recipient of two best paper awards and one best paper award nomination.
Mike Kazda is a Senior Technical Staff Member and AI Architect for Design Productivity at IBM Infrastructure. Mike has worked in design automation for over thirty years. He is an innovator of several construction tools and methodologies used in-house at IBM, which have taped-out IBM’s Z and POWER microprocessors, as well as past ASIC designs.
Srinivas Bodapati is a Principal Engineer at Intel Corporation. Srinivas is chair of the Si2 AI/ML SIG, driving new initiatives for Generative AI in EDA. Srinivas is responsible for pre-silicon parametric yield analysis, anomaly detection and rare event prediction algorithms using advanced statistical and machine learning techniques.
Sid Dhodhi works on the functional verification of GPU designs at Nvidia, overseeing multiple verification teams. Aside from contributing to numerous GPU tape outs over the years he has worked on integrating AI/ML methodologies to tackle conventional DV challenges. Sid is currently spearheading the deployment of ChipNemo at Nvidia, aiming to leverage LLMs to boost productivity of engineers. He holds a master’s degree in electrical engineering from USC and graduated from Stanford GSB’s LEAD program with a focus on product design and thinking.
Ivan Kissiov is a principal engineer at Siemens EDA, where he develops ML-based products for the fabs. Ivan has also developed EDA products for Synopsys and two EDA startups acquired by Mentor Graphics and Cadence. His expertise includes analog circuit modeling and simulation, VLSI placement, process modeling, deep learning, and high-performance computing.
Ilhami Torunoglu is R&D Director at Siemens EDA, where his group develops ML-based products for the fabs and chip designers. IIhami has also developed EDA products in one EDA (first GPU based EDA tool) to D2S and in two other startups developed 3D image sensors (sold to Microsoft) and consumer electronics. His expertise includes chip and sensor design, layout optimization, RET enhancements for fabs, deep learning, and high-performance computing especially on GPUs.
Moderator
Leigh Anne Clevenger is Vice President, Technology, at Silicon Integration Initiative (Si2), a community of suppliers, users, and academics who innovate on trusted standards to add value and share the solutions. Leigh Anne has established and driven efforts to identify and solve industry needs in applying AI/ML to EDA as the facilitator for member-led collaborative projects at Si2. Leigh Anne has published and presented research on data science, and her industry experience includes design tool development and design methodology definition.
