Professional Documents
Culture Documents
PC (3rd) Dec2016
PC (3rd) Dec2016
INSTRUCTION TO CANDIDATES :
17912 1.
17912 Attempt17912
any FIVE questions
17912 out of EIGHT questions.
17912 17912 17912 17912 17912
2. Each question carries T WENT Y marks.
17912 1.
17912 a) Describe
17912 Flynn’s 17912
classification 17912
for parallel computers
17912 by giving
17912 examples of each
17912 17912
17912 17912
b) What do you understand
17912 17912
by Systolic
17912
Array and
different than SIMD and pipelined processors?
o m
17912
vector Processors?
17912
How 17912
these are 17912
17912
2.
17912 computing
17912 with suitable
17912 examples.
17912
.r c
a) Describe the issues involved in achieving data and control parallelism in parallel
17912 17912
m
17912 17912
p e o
b) What do you understand by performance metrices? Explain various performance
.r c
matrices used for17912
the communication overheads in parallel 17912
computers with suitable
17912 17912 17912
examples.
17912
3.
17912 Also
17912
b r
state the limitations
17912 of each.17912 17912
p e
a) Describe various types of distributed memory networks used in parallel processors.
17912 17912 17912
a
b) Differentiate between static and dynamic interconnections of parallel computing.
rp
17912 17912 17912 17912 17912 17912 17912 17912 17912
4. What do you understand by dataflow programming? Write a version of the C-Linda
17912 5.
17912 a) Describe
17912
example.
a debugging
17912 technique b
program to compute the value of pi () that uses a fixed number of processes.
17912for distributed
17912 memory 17912
parallel program
17912 with 17912
17912 17912
b) Differentiate
17912
between
17912
the following
17912
: 17912 17912 17912 17912
1 | M- 35417 (S9)-1270
b) Write the PRAM algorithm to compute depth of each node in a given binary tree;
17912 7.
17912 For the17912
following computation
17912 : 17912 17912 17912 17912 17912
A×B+C
17912 17912 17912 17912 17912 17912 17912 17912 17912
compute the speedup in execution with chaining and without chaining, on a vector
processor with two pipeline functional units, one for floating point multiply and the other
for add, which are seven stages and six stages long respectively. Assume that, A, B and C
17912 17912 are three vectors each17912
17912 of 64 elements.
17912 17912 17912 17912 17912
17912 17912
(b) Shared memory networks
17912 17912
(c) Problems in Loop Scheduling
17912
o m
17912 17912 17912 17912
17912 17912
(d) Array processor
17912 17912 17912
.r c17912 17912
m
17912 17912
p e o
a .r c
17912 17912 17912 17912 17912 17912 17912 17912 17912
r p e
17912 17912 17912
b 17912 17912
a
17912
p
17912 17912 17912
rp
17912 17912 17912 17912 17912 17912 17912 17912 17912
2 | M- 35417 (S9)-1270