Special Sessions

Monday | October 28, 2024 | 10:00 - 12:00

In this dedicated session, we delve into five pivotal talks concerning hardware security. The session encompasses:

1. Investigation into attacks on quantum random access memory
2. Enhancing security in the NAND Flash memory  supply chain
3. The utilization of permissioned blockchains for supply chain management
4. Introduction of a novel  method for detecting hardware Trojans in manufactured microelectronics
5. Harnessing random self-reducibility  to combat physical attacks on microelectronics

Monday | October 28, 2024 | 10:00 - 12:00

At the core of the transformative Autonomy Economy lies the crucial technology of Autonomous  Machine Computing (AMC). This essential technology stack enables a wide array of autonomous machine form factors,  including intelligent vehicles, autonomous drones, delivery robots, home service robots, agricultural robots, industrial  robots, and many others yet to be conceptualized. AMC encompasses various technical domains such as sensing  technologies, computing technologies, communication technologies, autonomous machine algorithms, and aspects of  reliability and security. Currently, AMC is in a continuous state of evolution and definition. 

Monday | October 28, 2024 | 10:00 - 12:00

Fully Homomorphic Encryption (FHE) has drawn significant attention from various communities recently as it can process computing over encrypted data. On the other side, however, the practical deploying of FHE in actual applications like machine learning is greatly hindered by its sophisticated operations and extraordinary-large computational complexity. Various strategies have been proposed by the research community to facilitate FHE's smooth deploying in practical applications, and more recently, building efficient accelerators for FHE on different hardware platforms like FPGA/GPU.

Monday | October 28, 2024 | 13:30 - 15:30

We are excited to propose a special session on "AI Accelerators" for the upcoming ICCAD conference. This session,  developed in collaboration with experts from both industries, including leaders like Nvidia and academia, is designed to  comprehensively explore the full-stack technologies and methodologies at the forefront of AI acceleration. Organized into four  insightful talks, it will cover three pivotal aspects of AI accelerator technology: First, we will examine the role of large language  models (LLMs) in enhancing electronic design automation (EDA) tools specifically developed for AI accelerator chips. This  discussion aims to demonstrate how LLMs can revolutionize design methodologies and optimization processes. Second, the  session will highlight critical enabling technologies including Photonics and Emerging Memories. These technologies are crucial  for the development of advanced AI accelerators, providing the necessary infrastructure to support next-generation  computational needs. Third, we will delve into cutting-edge architectures with a focus on Computer-in-Memory (CIM) and Spiking  Neural Networks (SNN). These architectures represent innovative approaches to hardware design, promising significant  improvements in performance and efficiency. It promises to be highly relevant and welcomed by ICCAD participants, appealing  to a broad spectrum of the conference's audience, from researchers and engineers to industry leaders and innovators. Our  collaborative approach, integrating insights from both the industrial and academic sectors, ensures a rich dialogue and a  comprehensive exploration of the state-of-the-art in AI accelerator technology.

Monday | October 28, 2024 | 13:30 - 15:30

Artificial intelligence (AI) and Machine Learning (ML) methods are profoundly changing design automation (DA). All areas of DA are  exploring AI-based approaches to improve the quality result. From system-level design to logic synthesis, placing, and routing, and all physical  layout methods, AI promises to identify optimal solutions while significantly reducing time-to-completion of the search. In this special session, we  specifically discuss the use of AI-enabled methods in High-Level Synthesis (HLS). While generative AI and large language models (LLMs) have the  potential of being revolutionary for HLS, thanks to their ability to generate register transfer-level (RTL) descriptions starting from natural language,  they need to be carefully evaluated when considering quality and verifiability of the resulting design, and the size of the design they can currently  deal with. Additionally, they are not the only potentially disruptive solution, as design space exploration and classical optimization methods in HLS  might greatly benefit from ML. This special session discusses the opportunities and challenges of augmenting HLS with AI, including LLMs,  prediction methods, and novel optimization approaches, and their impacts on the overall agile hardware design automation flow. 

Monday | October 28, 2024 | 16:00 - 18:00

Artificial intelligence (AI) techniques have demonstrated remark able effectiveness in electronic design automation (EDA) and agile IC design. For such data-driven technology, the access to high quality, diverse, and representative circuit data is essential for both ML model development and evaluation. However, the lack of circuit data remains a long-standing and primary technical bottleneck. First, the lack of open datasets raises a high barrier to the devel opment of AI for EDA solutions. For ML model training, the label collection process can be highly time-consuming and resource demanding. Also, limited open-source circuit designs may not pro vide sufficient diversity in training, limiting the ML model perfor mance. Second, the lack of open benchmarks makes fair compar isons among different AI for EDA solutions highly challenging. It is difficult for truly outstanding AI solutions to stand out. 

Our special session (SS) delves into this long-lasting circuit data availability challenge by presenting four open-source works from different perspectives. Topics in our SS include 1) an RTL-to-GDS digital dataset and benchmark, 2) an analog circuit synthesis bench mark, 3) both dataset and benchmark about LLM-aided IC design, and 4) AI-assisted circuit data generation techniques and corre sponding datasets. They cover not only both digital and analog design, but also emerging techniques such as LLM-aided design and AI-generated circuit datasets. By releasing four latest explo rations to the public, we hope to contribute to more democratized and reproducible AI for EDA techniques. Specifically, open datasets will allow every researcher to train their AI solutions without an EDA license, and open benchmarks with a leaderboard will facilitate fair comparisons and encourage replications of AI solutions.

Monday | October 28, 2024 | 16:00 - 18:00

This special session is dedicated to the exploration and challenge of applying quantum technologies in practical settings, with a particu lar focus on quantum optimization. The rapid evolution of quantum computing has predominantly been within academic realms, but the imperative now is to extend these advancements to industrial applications, thereby unveiling significant economic benefits. Our session aims to narrow the gap between academic theories and real world applications, broadening the understanding and utilization of quantum technologies for sustainable growth. For this purpose, we have invited esteemed experts from JPMorgan, QuEra, LBNL and FinQ Tech, who are innovators in their respective fields. This session will present four in-depth talks: LBNL will research on the specialized compiler for the quantum optimization tasks. JP Morgan will explore the applications of quantum optimization in the financial sector, demonstrating how quantum technologies can revolutionize traditional financial models. QuEra will delve into a graph learning based strategies for predicting the performance of maximum independent set program, a critical aspect of determine if there is quantum advantage by using the quantum optimization for the task. Lastly, FinQ Tech will discuss the optimization of the quantum optimization, highlighting its potential to solve complex optimization problems more effectively than classical algorithms and potentially adopted by energy sector. By fostering an interdis ciplinary dialogue, this session aims to accelerate the progress of quantum technologies, ensuring they can be foundational to various industry sectors. The collaboration between academia and industry is crucial in overcoming current technological challenges and un locking novel solutions to global issues. This initiative promises to catalyze the practical adoption of quantum technologies, shaping them as a pivotal force in modern scientific and economic land scapes. 

Tuesday | October 29, 2024 | 13:00 - 14:30

The emergence of innovative non-volatile memory (NVM) technology has significantly changed the design of  intermittently powered embedded systems. NVM offers the opportunity to provide persistent data structures that  allow the system to maintain its state and data despite power shortages. However, NVM also presents new  challenges, such as wear and tear. At the same time, key memory parameters like read/write access times vary  greatly, which can lead to new trade-offs for customization in embedded systems design and for novel embedded  architectures in general. When NVM memories are deployed in an embedded system, this affects the entire  computing stack, from programming and operating systems to embedded systems architectures and micro architectures. This special session aims to demonstrate the new opportunities in embedded architectures with NVM  memories and how, in particular, embedded machine-learning applications may benefit. 

The team behind this ICCAD special session proposal offers a wide range of expertise, from software techniques to  embedded architectures and NVM technologies. It comprises researchers from six institutions in Taiwan, Germany,  the Netherlands, and the USA, offering diverse research perspectives from both academia and industry. 

Tuesday | October 29, 2024 | 13:00 - 14:30

Advances in artificial intelligence (AI)are transforming science and technologies. However, the exploding computational demands of  the most powerful AI algorithms lead to serious energy pitfalls. This trend is further accelerated by the simultaneous move towards  trillions of intelligent edge devices. Several efforts are underway to address this challenge which includes in-memory computing,  neuromorphic computing, silicon photonics architecture, etc. In this special session, we intend to showcase five eminent research talks  covering circuit-level innovation, system-level optimization, network-level creativity, and algorithmic codesign for next-generation  edge-intelligence systems. A major goal of this session is to bring communities together embarking a pathway towards decentralizing  AI in edge infrastructure with secured and energy-efficient computing.