Scribd
Upload
23 views6 pages

Fiduccia Mattheysis Algorithm

The ultimate ma_ka_bhosda

Uploaded by

vermamaggi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
23 views6 pages

Fiduccia Mattheysis Algorithm

The ultimate ma_ka_bhosda

Uploaded by

vermamaggi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 6

FIDUCCIA MATTHEYSIS ALGORITHM

SUBMITTED BY
VINEET VERMA 2023H1230195H
INTRODUCTION

 FM algorithm is an extension of the KL algorithm. These


algorithms are used for partitioning a circuit which is an integral
process of Physical Design in the VLSI industry.
 The aim of the algorithm is to minimize the net cost of all cut-
nets while satisfying the area constraint of each partition by
assigning all nodes to disjoint partitions.

Differences between KL and FM algorithm:


 The KL algorithm didn’t have the area constraint as is the case
in FM algorithm.
 Also, in FM algorithm we can move single cells independently
instead of swapping pair of nodes. Thus, in FM algorithm area
of individual cell need to be given to the algorithm unlike KL
algorithm.
 In FM algorithm we can partition unequal size partitions unlike
KL algorithm
 While the KL algorithm was based on partitioning the graph
based on edges the Fm algorithm does that based on nets
 FM algorithm supports hyper edges which is basically an edge
of a hyper graph. A hyper graph is a net with more than two
pins.
Terminology in FM Algorithm

Gain

The gain d(g(c)) for cell c is defined as: d ¿ where the FS(c) is defined
as number of nets connected to c but not connected to any other cell
in c’s partition and TS(c) is defined as the retention force which
represent the number of uncut nets connected to c.

Higher the gain, the higher is the priority to move the cell c to other
partition.

Ratio Factor

The ratio factor is defined as relative balance between the two


partitions with respect to cell area. It prevents all cells from
clustering into one partition. The factor r is defined as
area ( A )
r=
areaA+areaB

Where area and area B are respective areas of partitions.

Balanced partitioning
The partitioning of graph V into two partitions A and B is said to be
balanced if:
Base Cell
A base cell is one that has the highest cell gain among all free cells
and whose move doesn’t violate the balance criterion.

Core mechanism of the algorithm

 Gain Calculation: For each vertex, we calculate the "gain,"


which indicates how much the cut size would decrease if that
vertex were moved to the opposite partition. This step
identifies the potential benefit of moving each vertex between
partitions.
 Locking and Moving: Then we select the vertex with the
highest gain and move it to the opposite partition. Once moved,
a vertex is "locked" and cannot be moved again during this
pass. The gain values of the affected vertices are updated.
 Iteration: Finally we repeat the locking and moving process
until all vertices have been locked. This constitutes one pass of
the algorithm

Optimization:

 After a complete pass, the algorithm examines all the


intermediate states (after each move) and then selects the one
with the smallest cut size. This state becomes the absolute
starting point for the next pass.
 The algorithm can then perform multiple passes, unlocking all
vertices at the start of each new pass, until no further
improvements can be made.

Efficiency and Effectiveness:

 The FM algorithm is efficient because it focuses on local moves


with the highest potential for reducing the cut size, rather than
considering all possible moves which is the case with KL
algorithm
 It is effective for large-scale problems because it quickly
converges to a solution that, while not guaranteed to be the
most optimal, is often close to the best possible solution.

Practical Usage and conclusion:


 In practice, commercial and open-source EDA tools implement
these algorithms as part of their suite of optimization
techniques. Engineers and designers rely on these tools for
automating and optimizing the design of complex circuits.
 The underlying principles of the FM algorithm, especially the
focus on minimizing interconnections (cut size), remain central
to these tools.
 In summary, the Fiduccia-Mattheyses algorithm plays a crucial
role in the partitioning tasks in the EDA industry, and its
principles are integral to the design and optimization of
electronic circuits and systems.
 Its ongoing relevance is maintained through continuous
adaptations and improvements that cater to the evolving
challenges of circuit design and system optimization.

You might also like