Downloaded from

Size: px

Start display at page:

Download "Downloaded from"

Caroline Rogers
5 years ago
Views:

1 MCA (Revised) Term-End F,xarnination June, 2OOT MCSE-011 : PARALLEL COUfPUTING Time : 3 hours Maximum Morks : 700 Note 3 Question number 7 is three questions from compulsory. Attempt any the rest. l. (a) What is meant by 'Temporal Parallelism'? With the help of an example, explain how breaking a task into multiple tasks increases speed of execution. (c) (d) Write and explain the algorithm for solving the matrix multiplication problem using the parallel model. State and explain Amdahl's law for measuring speed up performance of parallel systems. Also, list the outcomes of analysis of the Amdahl's law. Define'hyper threading technology (HTT). Explain its salient features. Also, discuss the functionality of hyperthread processor. MCSE-01 P.T.O.

2 2. (a) Consider a list of following elements : l0 g, 6, g, 6, 7, 5, 11, Sort it using odd even transposition. Show intermediate steps. What is the concept of message passing libraries? List the salient features of MPI-I interface. l0 3. (a) Identify the types of the following vector processing instructions : (i) C(I) : A(I) AND B(I) s (ii) C(I) : MAX (A(I), B(I)) (iii) B(l) : A(t) / S (S is a scalar) (iv) B(l) : SIN (A(l)) (v) C(l) : SIN (A(t)) / CoS (A(r)) What are the problems faced by superscalar (c) architecture? How are these problems removed in VLIW architecture? Explain the shared memory parallel programming model. s 4- (a) Explain Bens network. Show the interconnection of Bens network for the following permutation : 70 r P: J MCSE-O1 1 2

3 Elaborate Handler's classification of parallel computers. (c) Define a Parallel Virtual Machine (PVM) and list its salient features. 5. (a) "To implement any algorithm, selection of a proper data structure is very important." In this context, explain any two data structures which are used in parallel algorithms, with an example for each. {b) What is meant by vector processing? How is it different from scalar processing? List and discuss various vector instructions along with their function mappings. 1.0 MCSE-OI 1 7,000

5 Teirn-End Examination December, 2OO7 MCSE-011 : PARALLEL COMPUT NG Time : 3 hours Maximum Marks : 0 Note : Question number 1 is compulsory. Attempt any three"questions f rom the rest. l l - (a) State and explain the law which uses the notion of constant execution time. Explain with the help of an example. Explain the concept of permutation network with an example. Discuss Perfect Shuffle permutation and Butterfly permutation. (c) (d) State and explain different fundamental parameters required for the analysis of parallel algorithm. What are the problems faced in super scalar architecture? How can these problerns be removed in VLIW architecture?. 'r- MCSE-01 1 P.T.O.

6 2. (a) Explain parallel virtual machine and list its salient f e a t u r e s.. l i l (c) Explain message passing with the issues decided by the system in the process of message passing. 5 Explain Bens Network as a non-blocking network. Show the interconnection of Bens Network for the following Permutation : 70 t-o P:l '.' '.'' j, -' 3. (a) Describe the property of the.merge Sort " circuit sequence and sort out the following list of values in ascending order using an odd-even merging circuit consisting of a set of cornparators. A : (4, 6, g,) B : (2, 7, g, 12\ Also show the intermediate steps Draw instruction exetcution steps Flynn's _t" Classification and discuss the Flynn's Classification based on instruction and data stream' 4. (a) What is the concept of message passing libraries? Explain two different types of message passing ' libraries with their merits and dernerits. Explain pipeline processing and describe the architecture of pipeline processing. M CS E

7 i i i I 5. (a) Discuss Block distribution and Cyclic distribution with their example. Explain difference between the following ; 6 6 (i) Tightly coupled system and Loosely coupled system (c) (ii) Vector processing and Scalar processing Explain the i.on.npt of Thread with basic methods in. concurrent programming languages for creating and terminating of threads. Also give the advantages the thread offers over other processes. I I I I : MCSE ,OOO I

9 MCSE-011o MCA (Revised) Term-End Examination June, 2OO7 MCSE-0l 1O : PARALLEL COMPUTING Time : 3 hours Maximum Marks : 700 Nofe r Qu estion number 1 is compulsory. Attempt any three quesfions from the rest. l. (a) Define Array processing. Why are array processors called as SIMD array computers? With the help of a block diagram, explain the architecture of an SIMD array processor. State and explain Gustafson's law for measuring speedup performance of parallel systems. Explain with the help of an example (c) (d) Elaborate the concept of permutation network. What is butterfly permutation? How is it implemented? Discuss. 70 Define cluster computing. Explain the memory organisation in a cluster computing. Give details of any of the important projects based on cluster computing. MCSE-O1 P.T.O.

10 2. (a) With the help of suitable example, explain control dependence. (c) List any three scientific applications and engineering applications of parallel computing, With the help of an example for each, explain the following parallel programming models : (i) Message passing (ii) Data parallel programming 3. (a) What are Bernstein conditions? Show the operation TIT;Til: ::l'wngc'de: 12:y:(b+c) *d.i 13: z: *2 + (a* e) "Flynn's classification discusses the behavioural concept for classifying the parallel cornputers and doesn't take into consideration the computer's classification depending on their structure. 4- (a) What is Synchronization Latency problem in ' multithreaded processors? How can it be handled? 5 Define following asymptotic notations used for analysing functions : 6 (i) (ii) Theta notation Big-O notation (iii) O notation MCSE-01 2

11 (c) Describe the property of the bitonic sequence and sort out the following list of values in ascending order using a combinational circuit consisting of a set of comparators. 3, 5, 19, 29,, 2, L4,0, 2L, 4,9 Ahso show the intermediate steps.. I 5. (a) Write short notes on the following : (i) (ii) Parallel virtual machine Data parallel programming Draw the structure and explai.n the following interconnection networks : (i) Fat tree (ii) Systolic array MCSE-01 2,000

13 McsE-ol1o MCA (Revtsed) Term-End Examination June, 2OO8 MCSE-0l1 ) : PARALLEL COMPUTING Time : 3 hours Maximum Morks : 700 Note : Quesfion number 7 is compulsory. Attempt any thtee ques ons lrom the rest. 1. (a) Explain two applications of parallelism. Deline the following terms : (i) (iil Running Time Speed up (c) F:<plain Amdahl's Law. (d) List the various search based tools used in performance analysis. (e) List some data parallelism feature of a modern comduter architech.re. 5x8=40 2. lal Define the terms Temporal parallelism and Data parallelism in detail. Also provide examples for both. 1O Explain the life cycle o{ a process in detail. What are the four actions for process creations? Explain each. I0 P.T. O.

14 3. (a) What do you understand by Bernstein conditions? Find out Bernstein mndition in the following example. A = B x C C _ D + E C=A+B E=F-D H : l + J What are the differences between Control flow and Data flow architech-rre? 4. (a) How will you classify Parallel computers according io Handler's classif ication? Discuss the design issues of interconnection network. Explain Cross bar network. 70 5, (a) Draw a Clos network for [ l t l L s 9 4 1). Explain VLIW Architecture in detail. What is the condition for compaciing the instruction in a VLIW instruction word? McsE-ol 1@ 3,000

15 MCA (Revised) Term-End Examination June, 2OO8 MCSE-011 : PARALLEL COMPUTING Time : 3 hours Maximum Morks : 700 Note : Question number 7 is compulsory. Attempt ang thtee questions from the rest. f. (a) Discuss any two applications of parallel processing. (c) (d) (e) What are the primary attributes used to meazure the performance of a parallel computer system? Discuss the four levels of parallel processing. Define Flynn's classification of parallelism. Define the prop rties of interconnection networks. Ex8:40 2. (a) Define Benz Network and design it {or permutation f s P l I L l s J r c Give the classification of Vector instructions. Explain each. MCSE-OI 1 1 P.T.O.

16 (a) How are parallel algorithms analysed? Explain each What types of data structures are used for parallel algorithms? Discuss in detail D. Discuss the four parallel programming models in use' Discuss the lollowing : (a) Grid compuiing 20 4x5 (c) (d) Cluster computing Hyper threading Parallel virutal machine McsE-ol 1,000

17 MCSE-O11 MCA (Revised) Term-End Examination December, 2OO8 MCSE-011 : PARALLEL COMPUTING Time : 3 hours Maximum Marks: 700 Nofe : Question number 7 is compulsory. Attempt any three questions t'rom the rest. 1. (a) Define the following in brief : (i) Fine grain (ii) Coarse grain (iii) Grain packing (iv) Communication latency (v) Node duplication Write the salient features of the parallel computer series PARAM and MARK developed by India. 6 MCSE-o11 P.T.O.

18 (c) (d) (e) A workstation uses a 15 MHz processor with a claimed MIPS rating to execute a given program mix. Assume a 7 cycle delay for each memory access. Find the effective CPI of this computer. Discuss the parameters used to analyse a generic algorithm. Write a parallel algorithm to rank the elements of a linearly linked list in terms of distance from each node to the last element of the list. Compare and contrast the following : (i) Vector processing and Array processing (ii) Instruction pipelines and Arithmetic pipelines (iii) Superscalar processor and Multithreaded processor (a) Elaborate the following in context of parallel algorithms : (i) Merge Sort circuit (ii) CRCW and CREW Give the output of following instruction in HPF : (i)! HPF $ pnocrssors Q(s, r) (ii)! HPF $ PROCESSORS P1(5)! HPF $ reuplrre Tr (22)! HPF $ DISTRIBUTE T1 (CYCLIC) ONTOPl MCSE-O1 1 2

19 (c) What is the function of Open MP directive? Write its syntax and meaning of the following directives : (i) First private clause (ii) Private clause (iii) Work sharing construct (iv) COPYIN clause 3. (a) Differentiate between control flow computing concept and data flow computing concept. Give example ol each. What is Flynn's classification of parallel computer systems? List salientfeatures of all categories of parallel systems. 4. (a) List various visualisation tools performance analysis. employed in Briefly discuss the following laws to n-reasure the speed-up performance : (i) Amdahl's Law (ii) Gustafson's Law (c) (i) Write a shared memory program for - parallel systems, to add elements of an array using two processors. (ii) Write a program for p'um (parallel virtual ' machine), to give a listing of the "slave" or spawned program, MCSE-O11 P.T.O,

20 Elaborate any four of the following : (i) Grid Computing 4x5:20 (ii) Intel Architecture IA - 64 (iii) Gantt Chart (iv) Parallel Virtual Machine (v) Multithreading MCSE-O1

21 r-c) v tc) O O MCA (Revised) Term-End Examination june, 2009 MCSE-011 : PARAttEt COMPUTING Time : 3 hours Maximum Marks : 1-00 Note : Question numbier 7 is compulsory. Attempt any three questions from the rest. 1. (u) Explain Bernstein conditions for detection of parallelism. (c) (d) (e) What are the Asymptotic Notations. Explain with the example, Write the Handler's classification of parallel computer. Explain permutation network with properties Explain the following interconnection networks. (t) (ii) Systolic Array Hyper cube MCSE-011 P.T.O.

22 2. (u) Explain basic concepts of dataflow computing and describe various applications of parallel computing. Describe the metrics involved for analysing the performance of parallel algorithm for parallel computers, also explain the various factor causing parallel system overhead. 3. (u) With the help of a diagram illustrate the concept of sorting using comprators for the L0 unsorted list having the elements values as [3, 5, g, g,, 12, 1.4, 20, 95, 90, 60, 35, 23, 18,01 Explain the pipeline processing and describe pipeline processing architecture. 4. (u) A three stage network is set so that ? ) P(s1)=[; s ) s 6 7 ) P(s2)=[l ) (0 1, ) P(s3)=[.z s s 6 4 z o t) What permutation realised by network. MCSE-011

23 Write short note on the following : (0 Spin Lock Mechanism f or synchoronisation. (ii) Synchronous and Asynchronous message passing J. (u) Discuss Handler's classification based on three distinct levels of computer. L0 Explain three work sharing constructs of open Mp. L0 -ooo- MCSE-011

25 I MCSE-0111 MCA (Revised) -14 Term-End Examination -14 December, " C MCSE-011 : PARALLEL COMPUTING Time : 3 hours Maximum Marks : 0 Note : Question number 1 is compulsory. Attempt any three questions from the rest. 1. (a) Explain and write the Algorithm for odd- 8 even transposition with an example. Explain various properties associated with 8 Inter Connection Networks. Explain and discuss all the performance 8 measurement tools and list the various search based tools and visualisation tools used in performance analysis. (d) Explain the following : 8 Vector processing Array processing (e) Explain two types of combinational 8 comparators. MCSE P.T.O.

26 (a) Flynn's classification is based on multiplicity of instruction streams and data streams observed by the CPU during program execution. Explain in detail. Define Hyper Threading Technology with it's features and functionality. (a) Explain the concept of message passing programming. Define structural classification based on different computer organisations. (a) Show the 8 x 8 Network created from the clos Network by setting n= m=n/ 2 and k = 2 and recursively decomposing any switches greater than 2 x 2 in size. What is the hardware complexity of this Network? Describe at least seven recent parallel programming models with example. (a) Explain the following basic concepts : Program Process Thread Concurrency (v) Granularity Explain the concepts of multithreading and its uses in parallel computer architecture. - o 0 o - MCSE-011 2

27 No. of Printed Pages : 3 MCSE-011 cv O MCA (Revised) Term-End Examination June, 20 MCSE-011 : PARALLEL COMPUTING Time : 3 hours Maximum Marks : 0 Note : Question number 1 is compulsory. Attempt any three questions from the rest. 1. (a) What do you understand by parallel 8 processing? Discuss the various levels of parallelism. Define the following asymptonic notations 8 used for analysing functions. Explain visualisation method for evaluating 8 the performance of parallel programs. (d) Explain the following terms : 8 Single instruction and single data stream (SISD) Single instruction and multiple data stream. (SIMD) (e) Which issues should be considered while designing an interconnection network? MCSE P.T.O.

28 (a) What are the various parallel programming models? Discuss each briefly. Define Bitonic sequence. Discuss a Bitonic sorting algorithm. Further using the algorithm, sort the following sequence. [15, 17, 19 20, 25, 27, 29, 34, 37, 18, 16, 13,, 8, 7, 6, 2] (a) What are the problems faced by super scalar architecture? How are these problems removed in VLIW architecture. Using Bernstein's conditions, detach maximum parallelism between the instruction of the following code. P1 :X = Y * Z P2 :P=Q+X P3 :R=T+X P4 :X=S+P P5 :V=Q+Z 4. (a) Explain different compiler directives in openmp in details. What do you mean by tightly coupled system? Give its characteristics. MCSE-011 2

29 5. (a) Explain the Gustafson's Law for measuring speed up performance with the help of an example. Explain the concept of Permutation Network with an example. Discuss perfect shuffle permutation and Butterfly permutation. MCSE-011 3

33 No. of Printed Pages : 3 MCSE-011 N N O MCA (Revised) Term-End Examination O June, 2011 MCSE-011 : PARALLEL COMPUTING Time : 3 hours Maximum Marks : 0 Note : Question number 1 is compulsory. Attempt any three questions from the rest. 1. (a) Explain the basic concepts of dataflow 8 computing and describe various applications of parallel computing. Explain PRAM Model with its 8 components. (c) Explain Hypercube Network with 8 properties. (d) Explain Bernstein conditions for detection 8 of parallelism. (e) Explain the Amdahl's law for measuring 8 speed up performance with the help of an example. MCSE P.T.O.

34 2. (a) Flynn's classification is based on multiplicity of instruction stream and data stream observed by CPU during program execution. Explain in detail. Discuss the following with respect to a parallel virtual machine. (i) Compiling and running of a PVM program. Creating and managing Dynamic process group. 3. (a) Explain the concept of multithreading and its use in parallel computer architecture. Give the classification of vector instruction. Explain each. 4. (a) Define array processing. Why are array processors called as SIMD Array computers? With the help of a Block diagram. Explain the architecture of an SIMD array processor. With the help of a diagram illustrate the concept of sorting using comparators for the unsorted list having the elements value as (3, 5, 8, 9,, 12, 14, 20, 95, 90, 60, 35, 23, 18, 0) MCSE-011 2

35 5. (a) A three stage Network is set so that. P(S1) = ( ) ( ) P(S2) = ( ) ( ) P(S3) = ( ) ( ) with permutation realised by Network. Define Cluster computing. Explain the memory organisation in a cluster computing. Give details of any of the important project based on cluster computing. MCSE-011 3

Fundamentals of. Parallel Computing. Sanjay Razdan. Alpha Science International Ltd. Oxford, U.K.

Fundamentals of. Parallel Computing. Sanjay Razdan. Alpha Science International Ltd. Oxford, U.K. Fundamentals of Parallel Computing Sanjay Razdan Alpha Science International Ltd. Oxford, U.K. CONTENTS Preface Acknowledgements vii ix 1. Introduction to Parallel Computing 1.1-1.37 1.1 Parallel Computing