Heterogeneous 3D-NoCs

Logo

Networks-on-Chips for heterogeneous 3D System-on-Chips

View the Project on GitHub jmjos/heterogeneous-3D-NoC

Introduction

This is the project repository containing a extensive suite of tools for the design of NoCs for heterogeneous 3D SoCs. We provide:

You can download the resources via this repo using a recursive git clone git clone --recursive https://github.com/jmjos/heterogeneous-3D-NoC or directly on the individual repositories.

NoCs for heterogeneous 3D SoCs

New manufacturing methods enable the production of heterogeneous 3D System-on-Chips (3D SoCs), in which dies, manufactured in disparate technology nodes, are stacked. Heterogeneity is one of the most promising paradigms to combine sensing and processing in a single 3D chip. For instance, it is possible to integrate sensors, analog-digital conversion and digital processing, which have different and even contradicting requirements to technology. Communication architectures, which use the advantages of heterogeneity, have not been considered so far. We propose dedicated 3D Networks-on-chips that target heterogeneous 3D SoCs and further exploit the specific properties of silicon dies in disparate technologies.

Tools, designs, models, etc.

This site collects tools, models, hardware designs for NoCs for heterogeneous 3D SoCs.

Getting started

Clone all project-related repositories with git clone --recursive https://github.com/jmjos/heterogeneous-3D-NoC

Simulator

The ratatoskr simulator can be used to simulate NoCs for heterogeneous 3D SoCs, because it accounts for technology features.

Router

A router model will follow soon!

Power models

We use the power models proposed in Bamberg, Lennart, and Alberto Garcia-Ortiz. “High-level energy estimation for submicrometric TSV arrays.” IEEE Transactions on Very Large Scale Integration (VLSI) Systems25.10 (2017): 2856-2866. and Bamberg, Lennart, Amir Najafi, and Alberto García-Ortiz. “Edge effects on the TSV array capacitances and their performance influence.” Integration 61 (2018): 1-10.

Implementations of the models will follow soon.

Design algorithms

The models can be used for exact design optimization. Source code to the heuristics will be made avaiable after publication.

Publications

We also provide preprints of the publicaitons as long as the publishers permit.

2018

2017

2016

2015

2014

People

The following people participated:

Acknolegdements

This work is funded by the German Research Foundation (DFG) project GA 763/7-1 and PI 477/8-1.

We are very grateful for the help of Prof. Dr. Volker Kaibel.