My Experience of Attending the VLSID24 conference

Introduction

The VLSID24 conference was held in Kolkata from Jan 6th to Jan 10th 2024. The conference was a premier platform for VLSI knowledge transfer in India. It consisted of tutorials, paper presentations and talks in the areas of Analog chip design, AI infrastructure, Neuromorphic computing and more.

VLSI Design Conference was initiated as a simple idea in 1985 to recognize the level of VLSI activities in India, focusing on engineering education & research.

I attended the VLSID24 conference as my first conference. Before this, I had just heard that a conference is a place where people present their research papers. But in the next 5 days, I came to know that there is a lot more to that.

Tutorials (Jan 6th – Jan 7th)

Going day by day, the first two days consisted of tutorials. These tutorials were designed so that attendees could get a brief idea of the papers being presented in the conference days ahead.

CDAC tutorial on Analog/Mixed signal design flow

There were many tutorials to choose from. On the first day, I attended the tutorial named “Analog IC Design” by C-DAC. In this tutorial, they explained the entire analog IC design flow using Cadence tools as well as described the theory of the circuits being implemented.

After this, I attended a session on neuromorphic computing by Arindam Basu, a professor at Hong Kong University. He described how we can move ahead of our conventional von Neumann architecture and integrate the memory and the compute into one unit. He discussed how neuromorphic architectures draw inspiration from the brain’s neural networks, where neurons serve as fundamental elements for data storage and communication. Additionally, he explored utilizing neuromorphic processors in edge AI devices through event-based sensor integration.

Prof. Priyadarshini Panda giving a talk on SNNs

On the next day of the tutorial, I attended another session on neuromorphic computing, this time by Professor Priyadarshini Panda of Yale University and her student Abhishek Moitra. They described how they developed a simulation platform for spiking neural networks as a low-power option to artificial neural networks.

In the second half of the day, there was a tutorial by C-DAC on the physical design flow.

Conference days (Jan 8th – Jan 10th)

Moving on to the day of the conference, it was held at the ITC Royal Bengal, Kolkata. The conference consisted of an inauguration ceremony which was held in the Plenary room of the hotel. The room was almost the size of two football fields.

The inauguration ceremony consisted of talks from eminent people in the VLSI industry as well as people from the government of Bengal. After the inauguration ceremony, the conference had a detailed schedule consisting of paper presentations, talks, panel discussions and networking breaks in between.

Opportunities and Insights

After the inauguration ceremony the Plenary room was divided into 8 ballrooms. Each of these ballrooms held a paper presentation session on a particular research area. One new thing I got to know about conferences is that they are opportunities where students compete with their research work with others and after all students of a particular domain have presented their papers, the best three get awards. This competition is domain specific, meaning top three for each domain.

Micron’s session on high bandwidth memory for AI workloads

The talks held during the conference were from eminent industry people. They were technical talks related to high bandwidth memory, AI chipsets, as well as panel discussions that I used to see on television.

One noticeable talk that I remember was from the director of the Indian Semiconductor Mission, Gupta, who gave a session on the government initiatives being taken for developing an electronics culture in India. He mentioned about the government’s MoU on building the Semiconductor supply chain the United States and Japan.

Synopsys’ flyer on their AI integrated EDA

The conference stage also consisted of exhibitions of different companies where they had stalls set up advertising the latest research and products that they offer. This was one of the most interesting parts of the conference where you get to know the vast number of companies working in the area.

Novel experiences

AMD’s booth at the conference (you can see that Korean guy in extreme left)

I met two students from South Korea who had built a startup named Hyper Accel. Hyper Accel was a company that made Latency Processing Units. Latency processing units (LPUs) are processors that are aimed to compete with GPUs for training generative AI models. They had set up their booth in collaboration with AMD. (They were using AMD’s cards in their LPUs).

Innatera’s talk on their Analog Spiking Processor

Apart from the technical talks, the conference was a great opportunity to network with people. I met some of the most promising guides in the conference and some friends too. One such meet was with the principal engineer of Innatera systems, Aditya Dalakoti. Innatera makes edge AI processors based on Spiking neural networks. Aditya advised me on the unnecessity of following the crowd and learning Verilog.

“People also do good with computer architecture. For example, I have been in the industry for 3 years and I don’t know Verilog, but I am still here giving a talk. There is one teacher Onur Mutlu. His lectures on computer architecture are great”
Aditya Dalakoti, Principal engineer, Innatera Systems

Conclusion

Conferences like these are a great opportunity to get to know people and build connections because industry people come to give talks and after talks, you can meet them and gain knowledge and showcase your skills. Some of these are CEOs of companies, others are domain experts, some are doing research in foreign universities, while others are MTech students. So, it’s a great opportunity to get to know about your domain. I would recommend every student, whether BTech or MTech, to attend such conferences. It’s not necessary to have a research paper to present. Networking is equally beneficial as personal research.