MESSAGE

ABOUT THE TALK
2019.10

Join us at d.lab!

Tadahiro Kuroda

MORE CLOSE

2019.10.05

Join us at d.lab!

The knowledge-driven society is here. How is the manufacturing industry going to adapt? Finding the answer is our mission at d.lab.

To realize Society 5.0 - a human-centric, ultra-smart society, it is necessary to bring about a paradigm shift from a capital-intensive to a knowledge-intensive society. While value-added in a capital-intensive society is primarily found in products, in a knowledge-intensive society it is delivered by services.

The key to creating services with high value-added is the innovative use of data that seamlessly connects between the physical world and cyberspace (virtual space). In other words, what is desirable is a system that utilizes IoT devices to collect information from the physical world and convert it into digital data, then performs sophisticated AI analysis on the data, and finally delivers valuable services through the internet based on the analysis results.

Such data-driven systems will require specialized semiconductor devices, both on the edge and in the cloud. There are tasks which are better suited for distributed processing at the edge near local sensors and memory, and tasks which are more efficiently performed through batch processing in the cloud. It is therefore necessary to utilize specialized chips that are optimized based on the required processing of the particular application.

As a result, the conventional capital-intensive chip development paradigm must transform into one that is knowledge-driven to deliver devices optimized for specific applications. This has led to the industry transitioning away from competing on transistor density and cost of general-purpose semiconductors to competing on value-added of specialized chips. Meanwhile, with the slowing of Moore's Law, it is becoming increasingly difficult to derive profit from high volume production of general-purpose products. Consequently, it is anticipated that, unlike the ASIC era, this new era of low-volume, high value-added specialized chips will continue for a long time to come.

Against the backdrop of this paradigm shift and from the perspective of a solution developer, we launch d.lab with the goal of creating a design methodology and constructing a manufacturing ecosystem to enable anyone with innovative ideas to easily develop specialized devices to realize their system. To that end, d.lab will endeavor to create a design platform to enable speedy development of data-driven systems while fostering the next generation of designers who will thrive in a data-driven society.

The “d” in d.lab embodies the key elements of our mission, which is to create a design platform that uses data as the starting point to drive the design of domain-specific systems that encompass everything between software and device, for the age of digital inclusion where every individual is empowered by digital technology. It also highlights our co-location with a university dormitory which provides an environment conducive to the exchange and collaborative creation of ideas in a diverse international community. Like garages giving birth to startups, we hope the dormitory environment will help bring about paradigm shifts.

To realize our target data-driven systems, we need to overcome three major challenges, one after another. The first challenge is to increase energy efficiency, which is why specialized devices are indispensable. Since specialized devices are required to provide only limited functionalities, the movement of data which contributes a large percentage of total system power consumption can be minimized. Unfortunately, specialized devices are expensive to develop. This brings us to the second challenge which is to improve design productivity. This in turn is achievable through design automation. The drawback of automation is suboptimal hardware performance. Hence the third challenge is to improve device performance, which can be realized using advanced manufacturing processes. Adoption of advanced processes also has the added benefit of improving energy efficiency.

The ecosystem we envision aims to overcome these three challenges to enable anyone to speedily develop prototypes using advanced foundry processes such as 7nm. We strive for 10x increase in energy efficiency by applying 3D integration and advanced device technology, as well as 10x improvement in design productivity by utilizing agile design methodology to enhance EDA tools and by adopting open architectures such as RISC-V.

At d.lab, it is also our goal to foster interaction and collaboration at the personal level between researchers from industry and university students, and promote collaborative innovation at the organization level between industry and the academia through various avenues from open innovation to large-scale joint projects. Furthermore, we endeavor to create the academic framework which seamlessly connects everything between systems and electronic devices, while advancing higher education in design automation.

We are building our lab on the Mejirodai International Village campus of the University of Tokyo which is a multi-function complex that also houses a dormitory for the university’s international community. As a result, d.lab will be interacting locally with a diverse population of about 1000 students and researchers on a daily basis. In addition, we will interact with universities and research centers nationally across Japan through SINET, a high speed digital network dedicated to research information exchange. Furthermore, we will extend our network beyond Japan through academic collaboration with the rest of the world.

We call on you to join us at d.lab in creating the digital solutions for Society 5.0.


Tadahiro Kuroda
Professor & Director of Systems Design Lab, University of Tokyo