CPU(Central Processing Unit) -- 中央處理器
GPU( Graphic Processing Unit ) -- 圖形處理器
VLSI(A circuit containing one hundred thousand to one million electronic units on a chip) -- 超大規(guī)模集成電路
Commodity cluster -- 常用集群? ASCI -- 先進計算計劃? network of workstations -- 工作站網(wǎng)絡(luò)
processor arrays -- 處理器陣列? multiprocessors -- 多處理器 multicomputers -- 多計算機
systolic arrays -- 脈動式陣列
Interconnection Networks -- 互連網(wǎng)絡(luò): shared memory or send message to each other
Shared versus Switched Media -- 共享介質(zhì)和開關(guān)介質(zhì)
A shared medium allows only one message at a time to be sent. Processors broadcast their message? over the medium. Each processor "listens" to every message and receives the one for which it is the destination.(共享介質(zhì)允許在某一時刻僅僅一個消息被發(fā)送馋辈，通過共享介質(zhì)停做，處理器得以廣播它們的消息檐涝，每個處理器監(jiān)聽每一條消息并接收發(fā)送至其上的消息)
Switched interconnection media support point-to-point message among pairs of processors. Each processor has its own communication path to the switch. Switches have two important advantages over shared media. They support the concurrent transmission of multiple message among different pairs of processors, and they support the scaling of the interconnection network to accommodate greater numbers of procrssors.(開關(guān)互連介質(zhì)支持處理器對之間點對點的消息互傳，每個處理器都有屬于它自己的通向開關(guān)通信的一條路徑扛施。與共享介質(zhì)相比，開關(guān)介質(zhì)有兩個重要的優(yōu)勢搓劫，它們支持在不同處理器對之間的多個消息并發(fā)傳輸畅买，而且支持互連網(wǎng)絡(luò)的擴展，從而能夠容納更多數(shù)目的處理器)
Switch Network Topologies 開關(guān)網(wǎng)絡(luò)拓撲結(jié)構(gòu)
vertices = processors or switches
edges = communication paths
type ：
Direct
? ? Ratio of swtich nodes to processor node is 1：1
? ? Every switch node is connected to
? ? ? ? ? ? * 1 processor node
? ? ? ? ? ? * at least 1 other switch node
Indirect Topology
? Ratio of switch to processor nodes is greater than 1：1
? Some switches simply comnect other switches.
Evaluating Switch Topology -- 評價開關(guān)網(wǎng)絡(luò)拓撲
* Diameter -- the maximum distance
* Bisection Width -- 對分帶寬
* Number of edge node -- 每個開關(guān)節(jié)點的變數(shù)
* Constant edge length -- 固定邊長
