| 研究生: |
周凱楓 Chou, Kai-Fong |
|---|---|
| 論文名稱: |
晶片網路多核心處理器之模擬器設計與實現 The Design and Implementation of a Network-on-chip Based Multi-core Processor Simulator |
| 指導教授: |
楊中平
Young, Chung-Ping |
| 學位類別: |
碩士 Master |
| 系所名稱: |
電機資訊學院 - 資訊工程學系 Department of Computer Science and Information Engineering |
| 論文出版年: | 2007 |
| 畢業學年度: | 95 |
| 語文別: | 中文 |
| 論文頁數: | 61 |
| 中文關鍵詞: | 多核心系統晶片 、晶片網路 、網路模擬器 、平台模擬器 、多核心 、分散式 、嵌入式 |
| 外文關鍵詞: | MPSoC, Network-On-Chip, Network Simulator, Platform Simulator, Multi-Core, Distribution, Embedded |
| 相關次數: | 點閱:91 下載:1 |
| 分享至: |
| 查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
在嵌入式系統之中,System-On-Chip(後簡稱SoC)是目前最流行的系統架構,目的是在單一晶片之中建構出復雜的迷你電腦系統。完整的SoC平台,必須包含硬體以及軟體部份;對於軟體部份,以手機系統為例,當使用者在聽音樂而又需要接聽新的來電的同時,作業系統就必須將目前的執行程序中斷,將處理器(Processor)的時脈轉為處理通話之用。然而,面對愈來愈多需要同時處理的作業,除了在作業系統上加強即時排程的效能之外,晶片內建多個運算核心架構也是隨之而來的必要過程。
目前的研究課題就是如何將在晶片中各個獨立的核心利用與封包交換網路通訊系統相連結,也就是Network-on-Chip(後簡稱NoC)晶片網路。然而,因為傳統的電路/匯流排交換架構比較適合一對一的資料傳輸,對於多核心系統的定址空間有許多限制,除此之外,還有許多限制相當不適用於較複雜的多核心SoC系統;因此,NoC是採用將處理器透過區域網路的路由節點結構來互連的方式來實作的一種單晶片內部封包交換網路通訊系統。在越來越復雜的SoC平台開發過程,軟體的開發常常需要為特別的硬體進行客製化,所以為了加速軟體的開發而將嵌入式開發平台虛擬化是目前最熱門的課題,虛擬化後的開發平台將能提供軟體開發者一個友善、快速而且方便的開發環境,讓軟體開發者可以在沒有真實硬體的情形之下開發軟體。總而言之,本篇論文的主要目標就是實作出適合支援多核心晶片網路架構的模擬平台,讓軟體開發者可以透過此模擬平台在系統開發前期快速的進行軟體開發工作。
Among the embedded systems, System-On-Chip (SoC) is the most popular system architecture at present. The purpose of SoC is to build a mini but complex computer system of a single chip. A complete SoC platform must include the hardware and software part. Take the software used in cell phone system for an example, when a user uses a cell phone to listen the music, an incoming calling suddenly needs to be answered. OS has to interrupt the current process and change the clock of CPU to deal with the calling issue. However, as the demand of processing a large amount of data grows, besides improving the performance of immediate schedule, increasing more kernels embedded into a chip is required.
The present issue is how to make every independent kernel of a chip connect to network communication system by exchanging the packets, which also is called Network-on-Chip. However, because the traditional electric circuit data bus is suitable for one-to-one data transmission, it has many limits for addressing space of multi-core system. Except addressing limit, it also has many other limits for multi-core system. Therefore, NoC is implemented by making every processor of a chip to connect to each other through the router nodes of the Local Area Network (LAN). It is a network communication system of a single chip by exchanging inside packets.
In the more and more complex period of developing SoC, the development of application often needs the special hardware to be customized. Thus, the virtualization of the embedded system becomes the most popular issue recently in order to accelerate the
development of applications. After virtualization, the development platform is able to provide developers a friendly, fast, and convenient environment to make developers can develop software applications without hardware. In conclusion, the object of this thesis is to implement a virtual platform to support the multi-core NoC and make developers can accelerate the development of a system at earlier stage.
[1] NS2 Official Website, http://www.isi.edu/nsnam/ns/
[2] Ralph Bulter, Zach Lowry, and Chrisila C. Pettey :”Virtual Clusters”, Proceedings of the 18th International Conference on System Engineering, IEEE, 2005
[3] Bellard, Fabrice, QEMU, http://fabrice.bellard.free.fr/qemu/
[4] F. Bellard. QEMU, a fast and portable dynamic translator. In Proceedings of the USENIX 2005 Annual Technical Conference, FREENIX Track, pages 41–46, June 2005.
[5] R. Davoli, VDE: Virtual Distributed Ethernet, http://vde.sourceforge.net/
[6] R. Figueiredo, P. A. Dinda, J. A. B. Fortes. "A Case for Grid Computing on Virtual Machines". Proc. International Conference on Distributed Computing Systems (ICDCS), May 2003.
[7] S. Santhanam, P. Elango, A. Arpaci-Dusseau, M. Livny. "Deploying Virtual Machines as Sandboxes for the Grid". Proceedings of USENIX Worlds, 2005.
[8] Intel Corporation. Intel virtualization technology. http://www.intel.com/technology/
computing/vptech/.
[9] S.Y. Wang, C.L. Chou, C.C. Lin, "The Design and Implementation of the NCTUns Network Simulation Engine" , Elsevier Simulation Modeling Practice and Theory, 15 (2007) 57 – 81.
[10] S.Y. Wang and K.C. Liao, "Innovative Network Emulations using the NCTUns Tool", as a book chapter of the "Computer Networking and Networks" book, (ISBN 1-59454-830-7, published by Nova Science Publishers)
[11] Earth-Simulator Introduction, http://en.wikipedia.org/wiki/Earth_Simulator
-60-
[12] Earth-Simulator on Top 500 Super Computers Site, http://www.top500.org/system/5628
[13] SystemC-QEmu Project Website, http://cephis.uab.es/proj/public/qemu/%20QEMU
[14] VDE, Virtual Distributed Ethernet Website, http://sourceforge.net/projects/vde/
[15] Introduction to Turing Machines, http://www.unidex.com/turing/tm_intro.htm
[16] 葛立偉,電子系統級(ESL)工具成為EDA業者關注的焦點,”電子工程專輯, Nov. p.200
[17] Moor’s Law From Wikipedia, http://en.wikipedia.org/wiki/Moore's_law
[18]Intel’s Page for Moor’s Law, http://www.intel.com/technology/mooreslaw/index.htm
[19] Scratchbox Official Website, http://www.scratchbox.org/
[20] OLPC Official Website, http://laptop.org/
[21] The Xen Virtual Machine Monitor, http://www.cl.cam.ac.uk/research/srg/netos/xen/
[22] Ian Foster, “What is the Grid? A Three Point Check List”, http://www.cl.cam.ac.uk/research/srg/netos/xen/
[23] J. Joven Murillo, D. Castells-Rufas, J. Carrabina. HW-SW framework for distributed parallel computing on programmable chips". XXI Conference on Design of Circuits and Integrated Systems. (DCIS). Barcelona 22-24 November 2006.
[24] H. Chang et al, "Surviving the SOC Revolution: A Guide to Platform Based Design", Kluwer Academic Publishers. 1999
[25] Dijkstra’s Algorithm Explain, http://en.wikipedia.org/wiki/Dijkstra%27s_algorithm