Scribd Logo
Upload

Welcome to Scribd!

  • 6 views

    Mid 2 Solution

    Uploaded by

    Moazzam Sheikh

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Mid 2 Solution

    Uploaded by

    Moazzam Sheikh
    6 views5 pages

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    6 views5 pages

    Mid 2 Solution

    Uploaded by

    Moazzam Sheikh

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 5

    National University of Computer and Emerging Sciences, Lahore Campus

    Course Name: Parallel and Distributing Computing Course Code: CS3006


    Degree Program: BS (CS) Semester: Spring 2023
    Exam Duration: 60 Minutes Total Marks: 35
    Paper Date: 08/04/23 Weight 12.5
    Exam Type: Mid II Page(s): 5

    Student : Name:___ SOLUTION ___ Roll No.________________ Section:_______

    Question # 1 a: [6 marks, CLO # 1]

    From the given options, select the best answer.


    i. With distributed systems, we will have:
    a. Processors with a shared logical address space
    b. Processors with a disjoint logical address space
    c. Processors that interact with each other by message passing
    d. Both (b) and (c)

    ii. Failover clusters are also known as:


    a. High availability clusters
    b. Load balancing clusters
    c. High performance clusters
    d. None of the given option

    iii. For a 2*8 mesh, using a naïve solution, how many messages would you expect the sending process to send,
    for a 1-to-all broadcast?
    a. Less than 8
    b. More than 12
    c. Less than 4
    d. Any one of the given option

    iv. Google App Engine is an example of


    a. Infrastructure as a service
    b. Platform as a service
    c. Software as a service
    d. None of the above

    v. The transparency that enables access to local and remote resources using identical operations is
    a. Access transparency
    b. Location transparency
    c. Computation transparency
    d. Scaling transparency

    vi. With the naïve solution for one-to-all broadcast on a ring, we would expect to send ____ messages
    to the other _____ processes, and this may lead to an __________ of the communication network
    a. p; p-1; overutilization
    b. p-1; p-1; underutilization
    c. p-1; p-1; overutilization
    d. (p-1)2; p-1; underutilization

    Department of Computer Science Page1 of 5


    Question # 1 b: [3 marks, CLO # 1]

    True/False

    I. The purpose of community cloud is to combine different clouds, e.g. private and public cloud.
    a. True
    b. False

    II. The total exchange in a linear ring with p nodes would take p steps.
    a. True
    b. False

    III. Grid Computing is a form of Utility Computing.


    a. True
    b. False

    Question # 1 c: [6 marks, CLO # 1]

    Provide short answers for the given questions.

    I. What should be the total time for sending a message of size m across n hops?

    t = ts+ (twm + th+ tr) n


    II. Which form of Operating System completely hides the distribution and presents a single system
    image?

    Middleware

    III. Which form of Operating System is better suited for tightly coupled OS for multiprocessors and
    homogenous multi-computers?

    Distributed OS

    IV. Which form of Operating System is better suited for loosely coupled OS for heterogeneous multi-
    computers?

    Network OS

    V. eScience applications are better suited to run on which type of computing?

    Grid Computing

    VI. How is the gather operation different from all-to-one reduction?

    The Size of message doubles at each step in gather operation whereas in reduction operation the
    size of message remains the same.

    Department of Computer Science Page2 of 5


    Question # 2: [3+3+4 marks, CLO # 3]

    a) Explain the scatter communication operation on a hypercube with 8 nodes.

    b) Provide total cost estimation for this operation. You have to consider the size of each message!

    𝑻 = (𝒕𝒔 𝐥𝐨𝐠 𝒑 + 𝒎𝒕𝒘 (𝒑 − 𝟏))

    At first step, the size of the message is 4 then 2 and finally 1 thus amounting to 4+2+1 = 7 which is
    equivalent to p – 1.

    Department of Computer Science Page3 of 5


    c) What modification would be required in the following code to perform scatter operation instead of one to
    all broadcast?

    The only difference is the message to be sent and received so in this case X is an array of P messages or 2 d
    messages.

    Now at sender end, we need to keep the first half of the array and send second half to the receiver.

    Int arraysize = 2d; //declare a new variable

    If (I am a participant)
    {
    arraysize = arraysize/2;

    If (I am the sender)
    Send &X[arraysize], araysize

    else
    Receive X , araysize
    }

    Department of Computer Science Page4 of 5


    Question # 3: [6+4 marks, CLO # 2]

    a) Write the output for the following piece of code assuming that there are 4 MPI processes. Assume
    there is no syntax error.
    #include <mpi.h>
    #include <stdio.h>
    int main (intargc, char** argv) {
    MPI_Init (NULL, NULL);
    MPI_Status status;

    int p, b, my_rank;
    MPI_Comm_size(MPI_COMM_WORLD, &p);
    MPI_Comm_rank(MPI_COMM_WORLD, &my_rank);
    int a = my_rank;
    int sTag = a;
    int rTag = (a-1+p) % p;
    int next = (my_rank + 1) % p;
    int prev = ((my_rank - 1 + p) % p);
    MPI_Sendrecv(&a,1,MPI_INT,next,sTag, &b,1,MPI_INT,prev,rTag, MPI_COMM_WORLD, &status);
    printf("I am %d: Got:%d from %d and Sent:%d to %d\n ", my_rank, b, prev, a, next);
    MPI_Finalize();
    }

    I am 0: Got: 3 from 3 and Sent: 0 to 1

    I am 3: Got: 2 from 2 and Sent: 3 to 0

    I am 2: Got: 1 from 1 and Sent: 2 to 3

    I am 1: Got: 0 from 0 and Sent: 1 to 2

    The order may differ (vertically)

    b) What is the one extra parameter used in MPI_Recv() that is not used in MPI_Send(), and in what
    circumstances would it be of greater importance?

    The status parameter (of MPI_Status type) is the extra parameter. It may be of
    greater importance when we call the MPI_Recv() using wildcards for either source
    or tag, or possibly both. This parameter provides us with information on the (i)
    source, (ii) tag, and (iii) error from the incoming message from another process.

    Department of Computer Science Page5 of 5

    You might also like