Report On Research Paper:

Improving of Quicksort Algorithm Performance by Sequential Thread or

Parallel Algorithms

Abdulrahman Hamed Almutairi Abdulrahman Helal Alruwaili

King Saud University

2012
 1

Contents

Figures 2

Tables 2

CERTIFICATE OF ORIGINALITY 3

Acknowledgement 4

Abstract 5

Introduction 5

What is Quicksort 6

Analysis of Parallel Quicksort 7

Comparision between Sequential and Parallel Quicksort 7

Speed Up Ratio 8

Conclusion 10
 2

Figures

1 Comparision of sequential and parallel quicksort, T=1000 . . . . . . . . . . 9

2 Graph showing speed up ratio by using parallel quicksort . . . . . . . . . . . 10

Tables

1 Average run times for different threshold . . . . . . . . . . . . . . . . . . . . 9

 3

CERTIFICATE OF ORIGINALITY

This is to certify, that the research paper submitted by me is an outcome of my

independent and original work. I have duly acknowledged all the sources from which the

ideas and extracts have been taken. The project is free from any plagiarism and has not

been submitted elsewhere for publication.

Name of Author (s):Vivek Kumar

Affiliated Institution: Indian Institue of Information Technology,Vadodara.

Title of the paper: Improving of Quicksort Algorithm Performance by Sequential

Thread or Parallel Algorithms

E-mail: [email protected]
 4

Acknowledgement

The success and final outcome of this project required a lot of guidance and assistance

from many people and I am extremely privileged to have got this all along the completion of

my project. All that I have done is only due to such supervision and assistance and I would

not forget to thank them.

I would like to express my deepest appreciation to all those who provided me the

possibility to complete this report. A special gratitude I give to our Course Instructor ,Mr

VIKAS JAIN, whose contribution in stimulating suggestions and encouragement, helped

me to coordinate my project especially in writing this report on topic ”Improving of Quicksort

Algorithm Performance by Sequential Thread or Parallel Algorithms”.

Lastly, I thank almighty, my parents, and friends for their constant encouragement

without which this assignment would not be possible.

 5

Abstract

Quicksort is a sorting algorithm and also known as a Partion-Exchange Sort in which

partition is carried out dynamically. Quicksort is a comparision sort and is popular because

it is not difficult to implement, works well for a variety of different kinds of input data, and

is substantially faster than any other sorting method in typical applications.The method

compares very favourably with other known methods in speed, in economy of storage, and

in ease of programming.

Introduction

One of the basic areas of the computer science is Data Structure. Sorting is an

important issue in Data Structure which creates the sequence of the list of items. Although

numbers of sorting algorithms are available, it is all the more necessary to select the best

sorting algorithm. Therefore, sorting problem has attracted a great deal of research as sorting

technique is very often used in a large variety of important applications so as to arrange the

data in ascending or descending order. This paper presents that sorting is an important area

of study in computer science. Like searching, the efficiency of a sorting algorithm is related

to the number of items being processed.

With the appearance of parallel computing, new possibilities have appeared to remove

this bottleneck and improve the performance of known sorting algorithms [1] by modifying

them for parallel execution. At first, this was achieved using distributed computing, however

with the hardware available today, it is possible to do this even on home computers thanks

to their multicore processors. In this report, we discuss a parallel version of the well know

quicksort algorithm and compare its performance to the performance of its simpler, sequential
 6

quicksort algorithm. Comparing their performance, we look for the threshold T, the size

of the array, at which the parallel algorithm becomes actually slower than the sequential

algorithm.

What is Quicksort

The quicksort algorithm was developed in 1960 by Sir Charles A. R. Hoare commonly

known asTony Hoare while in the Soviet Union.Quicksort is a divide and conquer algorithm.

Quicksort first divides a large array into two smaller sub-arrays: the low elements and the

high elements. Quicksort can then recursively sort the sub-arrays. The steps are:

1. Pick an element, called a pivot, from the array.

2. Partitioning: reorder the array so that all elements with values less than the pivot

come before the pivot, while all elements with values greater than the pivot come after it

(equal values can go either way). After this partitioning, the pivot is in its final position.

This is called the partition operation.

3. Recursively apply the above steps to the sub-array of elements with smaller values

and separately to the sub-array of elements with greater values.

The base case of the recursion is arrays of size zero or one, which are in order by

definition, so they never need to be sorted.

The pivot selection and partitioning steps can be done in several different ways; the

choice of specific implementation schemes greatly affects the algorithm’s performance.

Mathematical analysis of quicksort shows that, on average, the algorithm takes O(n

log n) comparisons to sort n items. In the worst case, it makes O(n2) comparisons, though

this behavior is rare.

 7

Analysis of Parallel Quicksort

Quicksort’s divide-and-conquer formulation makes it amenable to parallelization using

task parallelism[2]. The partitioning step is accomplished through the use of a parallel prefix

sum algorithm to compute an index for each array element in its section of the partitioned

array. Given an array of size n, the partitioning step performs O(n) work in O(log n) time

and requires O(n) additional scratch space. After the array has been partitioned, the two

partitions can be sorted recursively in parallel. Assuming an ideal choice of pivots, parallel

quicksort sorts an array of size n in O(n log n) work in O(log 2 n) time using O(n) additional

space.

Quicksort has some disadvantages when compared to alternative sorting algorithms,

like merge sort, which complicate its efficient parallelization[3]. The depth of quicksort’s

divide-and-conquer tree directly impacts the algorithm’s scalability, and this depth is highly

dependent on the algorithm’s choice of pivot. Additionally, it is difficult to parallelize the

partitioning step efficiently in-place.The limitation to speed increase of a parallel algorithm

as compared to a sequential algorithm are the overhead caused by the need to create new,

parallel processes and their management.

Comparision between Sequential and Parallel Quicksort

The parallel implementation is only slightly different. First, the size of the array to

be sorted is checked against a threshold. If the size is smaller than the threshold, sequential

quicksort is used as the overhead of creating new tasks would slow the sorting process too

much. If the size of the array is bigger than the threshold, instead of calling the quicksort for
 8

each part directly, new Task is created for each of the call and let’s them handle the sorting

of each part of the array.

Pseudocode for sequential quicksort :

With the threshold set at 50000, the parallel algorithm is actually slower, as the

computational cost of creating new tasks increases the total run time, but the parallelism is

not utilized enough to offset this. The use of scratch space simplifies the partitioning step,

but increases the algorithm’s memory footprint and constant overheads.

Speed Up Ratio

The performance of a parallel algorithm is determined by calculating its speedup.

Speedup is defined as the ratio of the worst-case execution time of the fastest known se-
 9

Figure 1. Comparision of sequential and parallel quicksort, T=1000

T=1000 T=5000 T=50000

10 0.01 0 0.010001
100 0.001 0.020001 0.050004
1000 0.250016 0.270011 0.260018
10000 2.010118 2.880166 3.060169
25000 5.380318 6.120344 9.15052
50000 11.36065 11.320644 19.61112
75000 18.14103 18.251045 28.60164
100000 24.91142 22.591294 34.19196
Table 1. Average run times for different threshold

quential algorithm for a particular problem to the worst-case execution time of the parallel

algorithm.

Speed up formula:

As can be see here, at low number of elements no speed up is achieved. As the number

of elements increases, the speed of sorting actually decreases. This is caused as stated before

by the overhead needed for creating the parallel tasks and as there is not enough elements

for the parallelism being able to compensate for this. After the number of elements increases

enough, the overall speed and speed gain increases as well by about 20%.
 10

Figure 2. Graph showing speed up ratio by using parallel quicksort

Conclusion

We successfully implemented a parallel version of quicksort algorithm. After choosing

a appropriate threshold value to switch from parallel to sequential sorting, we observed the

performance of the algorithm. The results are obviously in favor of the parallel quicksort

algorithm. Using a reasonable threshold to return to sequential quicksort, we are able to

circumvent the increased computational cost of creating new tasks for small datasets, while

with the bigger datasets we take advantage of the parallelism possible by today’s hardware.

And thanks to simplicity of the parallel implementation of quicksort algorithm it is easy to

achieve major, 20% increase of performance when sorting a larger dataset.

References

[1] Steven S Skiena. Sorting and searching. In The Algorithm Design Manual, pages 103–144.
Springer, 2012.

[2] Philippas Tsigas and Yi Zhang. A simple, fast parallel implementation of quicksort and
its performance evaluation on sun enterprise 10000. In Eleventh Euromicro Con-
ference on Parallel, Distributed and Network-Based Processing, 2003. Proceedings.,
pages 372–381. IEEE, 2003.

[3] ERNEST SCHEIBER. A parallelization scheme of some algorithms.