Fabien Clermidy

Position statement:

Computing goes 3D

As new applications like IoT or dedicated home servers require more highly dedicated, performing and power efficient hardware components, 
high variability of emerging or advanced technologies lead to high volume generic, programmable or reconfigurable components with, as a result, 
limited power efficiency.
Heterogeneous computing is needed for addressing application demands: generic processors, dedicated computing units, I/Os, high density and non-volatile memories… 
However, how to build such dedicated systems at acceptable cost and relevant features when considering advanced technologies required for 
reaching the performance targets?

We think 3D can be a solution for solving this issue: it is pervasive in technology developments and can help associating heterogeneous components 
with high shared bandwidth, adding non-volatility or high-density memories or reducing routing congestion.

In this panel, I will discuss on different 3D technologies and their potential applications from servers architectures to IoT.
 

About the panel member:

Fabien Clermidy obtained his master degree in 1994, his Ph.D in Engineering Science from INPG, Grenoble in 1999 and his supervisor degree from INPG in 2011.

Fabien is a pioneer in designing Network-on-Chip based multi-core and was the integrator of the first asynchronous-based Network-on-Chip published at ISSCC in 2007. He then took the lead of the second generation of Network-on-Chip based multicore dedicated to 3GPP-LTE applications published at ISSCC in 2010 and was the leader of the CEA team which has been working on the P2012 multi-core architecture in collaboration with STMicroelectronics and ST-Ericsson. At this period, his team elaborated one of the first 3D multi-core prototypes embedding a WIDE-IO DRAM memory called WIOMING.

Fabien is currently head of the digital architecture and design laboratory at CEA-LETI working on multi-core architectures and design with a focus on emerging technologies and new architectures using emerging technologies such as 3D TSV, 3D monolithic integration and emerging memories such as RRAM.