Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

Assignment – 4

1. Choose the correct options.


a. On-board and on-chip delays are typically smaller than off-board delays.
b. Partitioning of a PCB into chips is called chip-level partitioning.
c. Critical components/critical paths should ideally be assigned to the same partition
during partitioning to reduce the critical path delay.
d. The reliability of the system is directly proportional to the number of boards in the
system.
2. A digital block is designed using full custom design style. The block is of square shape, with
each side length 1 mm. The terminal pitch is 8 µm. The terminal count of the block is _____
pins.

Answer: Terminal count = Perimeter of block/terminal pitch

Terminal count = 4*1mm/8µm = 4000/8 = 500

3. Choose the incorrect options regarding partitioning algorithms.


a. Kernighan-Lin (KL) and Fiduccia-Mattheyses (FM) algorithms are examples of group
migration algorithms.
b. Kernighan-Lin (KL) and Fiduccia-Mattheyses (FM) algorithms are examples of
probabilistic algorithms.
c. Constructive algorithms provide initial partitions which are typically not optimal.
d. The simulated annealing algorithm is not a deterministic algorithm.

4. Consider the graph given below. The initial partition is done as shown in the figure.

How many iterations are needed to complete one pass of the KL algorithm for partitioning the
given graph? ___

Answer: We have 8 nodes in the given graph. In each iteration, we will fix two nodes. ∴ 8/2 = 4
iterations are needed to complete one pass of the KL algorithm for partitioning the given graph.

5. The values of cost function D(1), D(3), D(6), and D(8) defined in the KL algorithm in the graph
given in question 4 are _____, respectively.
a. 1, 0, 0, 1
b. 0, 1, 1, 0
c. 0, 1, 0, -2
d. 1, 0, 0, -2
e. 0, 0, 1, -1
Answer: The definition of the cost function used in the KL algorithm is D(a) = Ec(a) – Enc(a)

Where Ec(a) is the set of a’s incident edges that are cut by the cut line, and

Enc(a) is the set of a’s incident edges that are not cut by the cut line.

D(1) = 1 – 0 = 1

D(3) = 1 – 1 = 0

D(6) = 2 – 2 = 0

D(8) = 0 – 2 = -2

6. The sum of gains Δg(2,5) and Δg(3,7) defined in the KL algorithm in the first iteration of the
KL algorithm for the graph given in question 4 is _____.

Answer: The definition of the gain used in the KL algorithm is Δg(a,b) = D(a) + D(b) – 2c(a,b)

Where c(a,b) = 1, if there is an edge between a and b else c(a,b) = 0.

Δg(2,5) = D(2) + D(5) – 2c(2,5) = 0 + 1 – 2*1 = -1.

Δg(3,7) = D(3) + D(7) – 2c(3,7) = 0 – 3 – 2*0 = -3.

Sum = Δg(2,5) + Δg(3,7) = -1 + (-3) = -4.

7. Number of times Δg is calculated in the first pass of the KL algorithm for the graph given in
question 4? _______

Answer: In the first iteration (k=1), there are 4 nodes in each partition. So, you need to calculate
Δg between each node in partition A and each node in partition B. ∴ In the first iteration, Δg is
calculated 42 times. In the successive iterations, one node in each partition is fixed.

So, the number of times Δg is calculated in one pass = 42 + 32 + 22 + 12 = 16 + 9 + 4 + 1 = 30.

8. Consider the following graph with the initial partition as shown in the figure. This graph is
provided as input to the FM algorithm.

The ratio factor is 0.4 . Areas of each node are given as

Node Area
a 3
b 2
c 3
d 1
e 4
f 5
g 2
Perform the FM algorithm on the given graph. The value of gain, Δg(d), defined in the FM
algorithm, is ____.

Answer: Gain ∆g(c) for cell c, ∆g(c) = FS(c) – TE(c).

Where FS(c) is the number of nets connected to c but not connected to any other cells within c’s
partition, i.e., cut nets that connect only to c, and

TE(c) is the number of uncut nets connected to c.

So, Δg(d) = FS(d) – TE(d) = 1 – 1 = 0.

9. Which of the following is the correct balance criterion for the graph given in question 8?
a. 4 ≤ Area(A) ≤ 13
b. 3 ≤ Area(A) ≤ 12
c. 3 ≤ Area(A) ≤ 13
d. 4 ≤ Area(A) ≤ 12

Answer: The balance criterion is given by

[ r ∙ area(V) – areamax(V) ] ≤ area(A) ≤ [ r ∙ area(V) + areamax(V) ]

Where area(V) = sum of areas of all nodes = 20 and areamax(V) = area of largest
node = 5.

[0.4*20 – 5] ≤ area(A) ≤ [0.4*20 + 5]

3 ≤ area(A) ≤ 13.

10. Choose the base cell(s) for the first iteration of the FM algorithm for the graph given in
question 8.
a. a
b. b
c. c
d. d
e. e
f. f
g. g

Answer: A base cell is a cell c with the greatest cell gain ∆g(c), among all free cells whose move
does not violate the balance criterion.

Node FS TE ∆g
a 0 0 0
b 0 0 0
c 1 0 1
d 1 1 0
e 1 0 1
f 0 2 -2
g 0 1 -1

You might also like