You are on page 1of 2

Code No.

: 362/N

FACUL_TICS

B.E. In/IV Year (I.T.) n Semester (Main) Examination, April/May 2009


(New)
ADVANCED COMPUTER ARCHITECTURE
(Elective :- I)
Time: 3 Hours]

[Max. Marks: 75
Answer all questions from Pari A.
Answer five questions from Pari B.
Part A

(Marks:

25)

1. Define the term "Degree of parallelism" ?


2. What do you mean by Networ~ communication
Network."

latency in a message passing


2

3. What causes a processor pipeline to be under pipe lined ?

4.

Explain the limitations of the conventional

UNIX kemal for parallel processing

5. A Binary tree with height 5 is used as interconnection Network in a


multiprocessor. What is the Diameter of this Network?

6. Write short notes on "Optimizing compilers for parallelism".

7.

Explain the temporal locality associated with program or data access in memory
hierarchy.
3

8.

Define processor design spaces related to modern processor technology.

9.

How can you maximize the pipeline utilization by using instruction


technique? Give an example.

10. Consider
1
81
82
83
.

the following pipeline


2

reservation

reordering
3

table.

4
X

tBili
X

X"

Let the pipeline

clock period be T= 20 hs. Determine

throughput

of this pipeline.

[P.T.O.

.~--~
""'" '...

, 1'/:

f/

:;

~\

/.. ~. p.

'Io

~\:.,,:;..,~'~'~ '

.~,~, ,
.;
~~.
~. . ..., ,r.
'.?

"\

/ . .oil.;:'f~.!\'," ~~
" '\.

~r.

.',

t~'

~..
~\ ~+.t:~/.
..~!!, A.~;?

.'

'II

JI...
,
..~- ~

'.

~-'-'.'" ?-;.
.

';""

362/N

:50)

.~

11. (a) Explain

the PRAM models and PRAM variants.

(b) Discuss multiplication


n3 flog n processors.
12. (a) Distinguish

of two n x n matrices

in O(log n) time on a PRAM with


5

between

CISC scalar processors

and RISC scalar processors.

(b) Differentiate between private virtual memory spaces and globally shared
virtual memory spaces in different processors.

'

13. (a) Explain the routing


(b) List out various
mechanisms.

mechanism

in Omega network.

Cache Coherence

problems

and give synchronization


5

14. (a) Explain various


examples.

vectorization

and parallelization

methods

with code
5

'

(b) Explain
15. (a) Explain

software

pipelining.

the various

parallel

programming

(b) Explain the UNPACK matrix factorization


16. (a) Explain

the different

dynamic

(b) Explain

the multiple

context

17. Explain

connection
processors.

models.
using domain
networks.

decomposition.

5
5
5

the following terms:

(a) Grain packing

and scheduling

(b) Systolic program


(c) Vectorization

graphs

compiler

or vectorizer.

You might also like