Network Interface Buffer Elimination EECS 578 Final Report

نویسندگان

  • Qilu Guo
  • Jing Ji
  • Jiong Xue
چکیده

Aggressive transistor scaling continues to increase integration capacity with each new technology node. In recent five years, however, the performance gap between Moore’s Law and the art-of-design technology is getting larger, and area and power concerns are drawing much more attention, especially for networkon-chip design. Much attention has been paid on routers, and few on network interface. However, the storage of the data flits in the network interface causes an area overhead and therefore consumes more power. In this project, we propose eliminating buffers in the network interface and preserving the data in the cache instead to reduce overall area. From the simulation result, the proposed design reduces the area of the network interface by 6 times without degrading the performance. Keywords—Network interface, network-on-chip, area saving, buffer elimination

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Robust Cache Coherence Protocol Verification with Inferno EECS 578 Project Report Team FLY - Bee

Fault tolerant architectures are emerging to guarantee the reliable functionality of vulnerable silicon devices. However, the growing complexity of the coherence protocol and Network-on-Chip (NoC) design come to be a big challenge to presilicon verification. In this work, we implemented a custom designed robust MESI and directory based cache coherence protocol in System Verilog and practiced wi...

متن کامل

Eecs (eecs)

EECS 132. Introduction to Programming in Java. 3 Units. Introduction to computer programming and problem solving with the Java language. Computers, operating systems, and Java applications; software development; conditional statements; loops; methods; arrays; classes and objects; object-oriented design; unit testing; strings and text I/O; inheritance and polymorphism; GUI components; applicatio...

متن کامل

A Shared-Memory Packet Buffer Management in a Network Interface Card

This paper proposes a dynamic shared-memory packet buffer management algorithm for a protocol processor in a network terminal. The protocol processor is located in a network interface card (NIC). In general, two types of packet buffer management algorithms, static and dynamic, can be used in a NIC; the dynamic buffer management algorithms work better than the static ones for reducing the packet...

متن کامل

Optimization Techniques in VLSI Physical CAD

This paper presents an overview of the techniques used for solving optimization problems, in the area of VLSI Physical Computer Aided Design. The material is presented in the form of a general introduction, without going into quantitative or theoretic details. The work describes fundamental aspects of topics such as simulated annealing, genetic algorithms, linear programming, integer programmin...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2015