ภาษาไทย | English

หมวดหนังสือ -> เอกสารสถาบันเทคโนโลยีพระจอมเกล้าเจ้าคุณทหารลาดกระบัง -> Development of parallel algorithms based on openMP : SNPHAP and PPMQSORT case studies

คะแนนหนังสือ :
ให้คะแนน  

อ่านหนังสือ  ยืมหนังสือ
 

Title
Development of parallel algorithms based on openMP : SNPHAP and PPMQSORT case studies
Title.Alternative
การพัฒนาอัลกอริทึมแบบขนานโดยใช้ OpenMP กรณีศึกษา SNPHAP และ PPMQSORT
Creator
Ratthaslip Ranokphanuwat
Creator.Orgname
King Mongkut's Institute of Technology Ladkrabang
Subject
Algorithms
Subject
Sorting Electronic computers
Subject
Data processing
Subject
Open source software
Subject
Electrical Engineering--Thesis (Degree of Doctor)
Subject
King Mongkut's Institute of Technology Ladkrabang. Electrical Engineering--Thesis
Description.Abstract
Abstract: This thesis has introduced the methodology for parallelizing sequential algorithms using OpenMP 3.0 library to run on any shared memory/multicore/multi-socket systems. It can achieve high Speedups especially on demanding applications such as bioinformatics, Big Data analytics and so on. As the first case study, acomputation-intensive problem called SNPHAP, an Expectation Maximization haplotype inference algorithm.The methodology starts with profiling SNPHAP to identify hotspots. Then, those hotspots are parallelized using OpenMP parallel For and Task constructs. The Speedups are 316%, 410%, and 488% on an 8-core Xeon E5405, an 8-coreHyperThread Xeon E5520 and a 32-core AMD Opteron 8356 Linux machines, respectively. The second case study is the QuickSort, a well-known sequential sorting algorithm in Computer Science. An explosive amount of data has tremendous impacts on run time complexity. This thesis proposes an efficient and scalable algorithm called Parallel Partition and Merge QuickSort (PPMQSort), the PPMQSort is developed to be compatible and benchmarked with the fastest C/C++ Stdlib qsort(). The PPMQSort recursively divides an unsorted input data array into partially sorted partitions using nested OpenMP Task parallelism. Finally, those independent partitions are sequentially qsort() (conquered). As such, it can be up to12.29 times faster on an 8-core HyperThread Xeon E5520 while sorting random 200M 32-bit integer data at 16 threads. With the same data type, a 4-core AMD A6-3600 CPU (non- HyperThread) can reach up to 4.67x yielding a superlinear Speedup. It has been proved that the proposed PPMQSort can exploit all available cache levels and HyperThread CPU cores well thus utilizing up to 83% and 96% of CPU on Xeon E5520 and AMD A6-3600, respectively
Publisher
สถาบันเทคโนโลยีพระจอมเกล้าเจ้าคุณทหารลาดกระบัง. สำนักหอสมุดกลาง
Publisher.Place
Bangkok
Publisher.Email
library@kmitl.ac.th
Contributor.Role
Surin Kittitornkun
Contributor.Email
surin.kh@kmitl.ac.th
Date.Created
2016
Date.Issued
2018-12-27
Date.Modified
2018-12-27
Type
วิทยานิพนธ์/KMITL E-Thesis
Format
application/pdf
Identifier.KMITL CODE
KMITL-2016-EN-D-018-027
Source.CallNumber
EThesis
Language
eng
Coverage.Spatial
King Mongkut's Institute of Technology Ladkrabang. Central Library
Rights
King Mongkut's Institute of Technology Ladkrabang
Thesis
Thesis (D. Eng. (Electrical Engineering)) -- King Mongkut's Institute of Technology Ladkrabang, 2016
Thesis.DegreeName
Doctor of Engineering (Electrical Engineering)
Thesis.Level
Doctor
Thesis.Faculty
Faculty of Engineering
Thesis.Descipline
Electronics Engineering
Thesis.Grantor
King Mongkut's Institute of Technology Ladkrabang


แสดงความเห็น
หัวข้อ
รายละเอียด
 
 
 
     แสดงความเห็น

107.23.37.199