AI Hardware Summit 2021 Preview

The AI Hardware Summit is coming up on September 13-16. It will take place in-person at the Computer History Museum. The website for the event says that most presentations will be given in person, but they will all also be streamed for virtual attendees.

When I rejoined Cadence in 2015, we were still in an AI winter. Artificial Intelligence had been heavily hyped in the 1980s and early 1990s, with several high-profile projects and companies. But then nothing really worked as hoped and work in the area largely stopped. You can read a more detailed history of what happened next in my post A History of Neural Networks. The short story is that ImageNet happened (read about that in my post ImageNet: The Benchmark that Changed Everything). What happened next was that Geoff Hinton, Yann LeCun, and Yoshua Bengio made back-propagation work. In 2012, a convolutional neural network won the annual ImageNet competition, halving the error rate of other approaches. Neural networks were a thing. Read more in my blog post Geoff Hinton, Yann LeCun, and Yoshua Bengio Win 2019 Turing Award.

Huge advances have been made in the last decade, both in algorithms, quantization, sparsity. And, in particular, in hardware. If you are interested in what is happening in AI hardware, then the only place to be in mid-September is at the CHM or in front of your computer.

Here's the conference's own summary of why you should attend:

AI hardware is evolving—and so are we! As machine learning models continue to grow in size and complexity, and more and more models enter production in enterprises worldwide, the way we approach accelerating these workloads is changing.

At the front end, data-centricity is taking precedence over model-centricity. At the back end, AI practitioners want systems that are performant and efficient, but also sustainable, explainable and accountable. From massive research models like GPT-3, to day-to-day models deployed by enterprises around the world, we are lifting the hood on how to make AI fast, efficient and affordable.

AI Hardware Summit is evolving, we continue our mission to help those who are accelerating AI workloads in the cloud and at the edge, and this year is all about systems-level AI acceleration.

Cadence

Cadence is involved in the event, with keynotes, presentations, and workshops.

Keynote

Lip-Bu Tan is giving the keynote on Wednesday, September 15 at 9:00am titled Fueling the Semiconductor Renaissance.

The electronics industry’s renaissance continues to drive change and growth in semiconductor design. With the increasing influence of hyperscalers comes a restructuring of the ecosystem as they drive priorities for process nodes, packaging, PCBs, and complete systems. The productivity of designing software for intelligent systems is an increasingly important bottleneck. Data analytics are a vanguard of this wave of change, and Cadence is a key enabler of the technologies fueling the innovation.

Technology Ecosystem

Following that, at 10:05am, Sanjive Agarwala, who heads up our IP business, will present Scalable On-Device to Edge AI for Pervasive Intelligence.

In this talk, we’ll present the on-device AI IP requirements for intelligent sensor, IoT audio/ vision, mobile, and automotive/ADAS markets. The presentation will cover the full range of Cadence’s Tensilica on-device AI solutions for low-cost voice-activated consumer devices to high-throughput autonomous vehicle perception. These IP are widely deployed in high-volume AI-enabled end products such as smart speakers, mobile phones, surveillance cameras, and automotive subsystems. We’ll show how the wide portfolio of Cadence’s low power programmable DSP and AI engines in on-device AI IP meets the need of each specific market. Finally, we’ll outline our on-device AI software tools and support for a wide range of software frameworks and a broad market.

Workshop

On Monday, September 13 at 9:00am, Cadence is presenting a virtual workshop titled Enabling Scalable On-Device and Edge AI Using Cadence Tensilica IP. The speakers will be Pulin Desai—Product Management Group Director, Adam Abed—Product Marketing Director, and Suhas Mitra—Product Marketing Director.

In this workshop, we will dive deeper into the Cadence Tensilica artificial intelligence (AI) product offering and specifically address how it solves key challenges in the on-device and edge AI space. The presentation will cover our comprehensive software offering that addresses a full range of devices, from low-cost voice-activated consumer devices to high-throughput autonomous vehicle perception. We will also cover key technical aspects and value points of our three hardware AI platforms that address the diverse needs of the market. We will show how the products within these platforms enable compelling use cases with industryleading performance and efficiency. Finally, we will summarize the key technical aspects that drive differentiation in our solutions and also outline how we plan to innovate and maintain differentiation moving forward.

Roundtable

On Thursday, September 16 at 9:00am there will be a virtual roundtable titled Scalable On-Device to Edge AI for Pervasive Intelligence. The participants will be Adam Abed—Product Marketing Director and Suhas Mitra—Product Marketing Director.

This roundtable will cover the full range of Cadence’s Tensilica on-device AI solutions for low-cost voice-activated consumer devices to high-throughput autonomous vehicle perception. These IP are widely deployed in high-volume AI-enabled end products such as smart speakers, mobile phones, surveillance cameras, and automotive subsystems. We’ll show how the wide portfolio of Cadence’s lowpower programmable DSP and AI engines in on-device AI IP meets the need of each specific market

Agenda

Here is an overview for all three days of the whole conference (not just the Cadence content):

Full details on everything are on the conference website.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.

.