UNIT -1 Flynn's and__Handler's__ Classification of _parallel_computing structures. Pipelined andVector Processors. Flynn's Classification of Computers The sequence of instructions re: an instruction stream. The operations performed ot a data stream. stream, single data stream (SISD) 2. Single instruction stream, multiple data stream (SIMD) 3. Multiple instruction stream, single data stream (MISD) 4. Multiple instruction stream, multiple data stream (MIMD)Flynn's Classification of Computers Data stream <—________88 Stream single Multiple Instruction stream SISD and fe Data Stream’. It single. computer containing a unit. SISD stands for ‘Single Parallel processing, in this case, may be achieved by means of multiple functional units or by pipeline processing.SISD: weans uoRINgSuL Where, CU = Control Unit, P| ing Element, M = Memor y Instructions are decod: trol Unit and then the Control Unit sends the instr ocessing units for execution. Data Stream flo ssors and memory bi- directionally. Examples: SIMD stands for ‘Single Instruction and Multiple Data Stream’. It represents an organization that includes many processing units under the supervision of a common control unit. All processors receive the same instruction from the control unit but operate on different items of data.The shared memory unit must contain multiple modules so that it can communicate with all the processors simultaneously. SIMD: {Memory | Data Bus f { Controt unit | oe} } a i wea) Processor units t "pats stream tT. Alignment network J { SIMD is mainly dedicated to array processing machines. However, vector processors can also be seen as a part of this group. MISDi MISD stands for ‘Multiple Instruction and Single Data stream’. MISD structure is only of theoretical interest since no practical system has been constructed using this organization. In MISD, multiple processing units operate on one single-data stream. Each processing unit operates on the data independently via separate instruction stream.teens eG Instruction Stream ~ Where, M = Memory Mod | Uni = Processor Units Example: The experimental computer (1971) MIMD stands fot Itiple Instruction and Multiple Data Stream’. In this organization, all processors in a parallel computer can execute different instructions and operate on various data at the same time. In MIMD, each processor has a separate program and an instruction stream is generated from each program.MIMD: Where, M = Memory Mod i ment, and CU = Control Unit Examples: Handler's Classifi Arghitecture In 1977, sented a computer architectural classific: ining the degree of parallelism je computer system hardware. Parallel systems are i to the program as compared to the single processol stem because parallel system architecture varies according the multiple CPUs and they must be coordinated or synchronized. In Handler's classification pipeline processing systems are divided into three subsystems:Processor Control Unit (PCU): Each PCU corresponds to one processor or one CPU. Arithmetic Logic Unit (ALU): ALU is equivalent to the processing element (PE). Bit Level Circuit (BLC): BLC corresponds to the combinational logic circuit required for 1-bit operations in ALU. ALU is a small unit that has fewer features tha and it works under the instructions of designed for performing arithmetic ani its name. A system comprises mul increase the performance of the s\ 7 vel circuit) is required to perform single or bit,opt Handler's classification use; independent entities that processor . ALU is Computer = Where CeSsol ‘CUs) within the computer hat can‘Be pipelined 's) under the control of PCU in be pipelined Example 1: Let us consider an example of Texas Instruments’ Advanced Scientific Computer (Tl ASC), which has one controller that controls 4 arithmetic pipelines, each having a 64- bit word length and 8 pipeline stages. From this data, we get K = 1, K’= 1, D=4, D’= 1, W = 64, W' = 8. So, we can represent Tl ASC according to Handler's classification as follows :TIASC = Let us now look at one more exercise on Handler's classification. Example 2: CDC 6600 has only a single CPU with an ALU, that has 10 specialized hardware functions each of 60-bit word length and up to 10 of these functions can be linked into a longer Here, we have two parts to consid and I/O Processor(IP). So, the repre protocol, whereas UDP is one of the core i. ernet protocol suite. UDP is used as itive communicative protocol. R in RUDP st a UDP-based data transfer but with hi It provides reliability by using the sliding windo is basically the solution to the UDP where data rellability along with confirmation is needed. As UDP provides “unreliable data transfer protocol which is unreliable services of things and the worst it sometimes goes missing without notice. Reliable UDP uses both positive and negative feedback to provide data reliability which provides reliable data transfer. In UDP sender's simply send a message without a piece of prior information about the receiver's availability which results in afaster rate but it can result in the loss of some packages, also the receiver can receive duplicate packets and UDP also does not provide information that the package has been received or not. RUDP used a sliding window protocol that delivers or transfers the datagram with reliability. RUDP Architecture The sending process and receiving pro both stand in the application layer. The si window size is maintained by both) the receiver, the window consists of some i tending to maximum avoiding the com icat! ing all the edge cases where the packets e drop| ‘Application — — layer Transport rerabie rot sone} [Bat] [oat] [aetiver eatay faye channel ce protocol (senaing sae) (recenng ice) ust_ sendy) rot_revo) unreliable channelImplementation of RUDP protocol: Use synchronized shared buffers using counting semaphores so that only one thread access the buffer at a time to prevent the deadlock situation from occurring. Let's keep two-variable called as the base an track of the window functioning. If the sender si the variable next is incremented to i le way of immediately after sending th, annel, simulate the packet loss rate and ré the protocol. packets are kept in a q igne e that values the system's current tim: work delay. When the current wi e assigned value the Classes in RUDP Protocol: Following are the classes that are used to implement the RUDP protocol: RUDP: This contains the send() and receives () functions called by the client and server.Buffer_RUDP: contains the implementation shared buffer using counting semaphores. Receiver_Thread: Implements a threaded architecture that simultaneously waits for the packets in the sockets while sending and receiving the data and waits to process the incoming data. Segment_RUDP: defines the structure of RUDI Timeout_Handler: to handle the timeout: Support_RUDP: provides function: e send \ Client: It sends the data. Server: It receives the dai J) knownas RUDP’s SWP on which RUDP is based, the reliability comes into play and it is ve i r security reasons where the insferred along with confirmation is ments which means it divides the provided data segments then packets or data packets are prepared and si over the network, resulting in when the receiver receives the package it checks the order of the packets and discards duplicates and after that, is sent it sends the acknowledgment to the sender for the correctly received packets. There are 3 Sliding window Protocols: 1. One Bit Sliding Window Protocol 2. Go back to N protocol3. Selective Repeat Protocol Piggy-backing Technique: Sliding window Protocol uses the piggy-backing technique, the technique is based on the acknowledgment that is received but, how is received? So, the answer is it provides the sliding window protocol to attach the acknowledgment in the next frame so that when a receiver receives the data it maintains numbers which is corresponding to the frames th be called acknowledgment frames withi same or different sizes. The mathematics of frame calcul: When a new packet is sel e receiver it is given the highest packet window's upper edge is incremented wil e acknowledged frames and vice versa for the mark of unackn, frames and wi i to zero that means all the d in data-intensive applications in a very less amount of time over high- Client Class sets the rate of loss and sends the data to the RUDP class, where window sizes of both the sender and receiver are set in the class. The RUDP class as per the working divides the data into segments of equal sizes and hence assigns the frames i.e. sequence number and checksum to each segment separately then put it in the sender’s window if it has any available slot.