[Members-Internal]]

First, My name is Yuki Tanaka. Not Hiroki.

Physical Design of 3D OASIS Router
pipeline.jpg
3D-OASIS-NoC system architecture

Background

During this last decade, 3D- Network-on-Chips (3D-NoCs) have been proposed as a promising solution for future systems on chip design. It offers more scalability, higher bandwidth, and less interconnect-power than the 2D-NoC architectures.

One of the major issues in 3D-NoC systems is how to verify the system correctness and integrity. 3D-NoC architecture research include a lot of trade-offs between topology, routing, flow control, buffer size, packet size, and other optimization techniques. It is difficult to analyze these trade-offs using only high-level simulations. Therefore, prototyping is an essential design phase for evaluating the performance of NoC architecture under real applications.

Previously, 3D-OASIS- NoC (3D-ONoC) has been designed. 3D-FTO showed great performance under real benchmarks and synthetic traffic patterns [1]. However, logic modules were connected to routers instead of real cores. This make the evaluation less accurate.

Research goal

The main goals of this research are:

Research plan

Prerequisite

References

Benchmark suite

My memo


トップ   新規 一覧 検索 最終更新   ヘルプ   最終更新のRSS