Evren Yurtesen - Code Optimization of the Biosequence Analysis Program HMMER

Code Optimization of the Biosequence Analysis Program HMMER Download

Thumbnails:

Description:

A computer system is a complex piece of hardware. Inefficient usage of the computer hardware by programs can cause excessively high memory usage patterns and slow processing speeds. Code optimization is the process in which the program code can be improved to use less memory space, less CPU cyles (faster run times), storage space or power. Optimization of program code is often a process with trade offs and usually it is not possible to improve the program code on all aspects of performance. The performance of the program code can be improved by removing redundant code, replacing inefficient algorithms with more efficient ones, using the advanced functions provided by recent CPUs or using optimizing compilers. In this thesis, the main optimization aspects which are presented will be optimized usage of CPU cycles and creating program code which can be optimized efficiently by the program compiler. Optimized usage of some advanced CPU functions like SSE operations will be presented as a more efficient way to use CPU cycles, as well as Altivec counterparts of these functions will be mentioned. The example program code used for practical examples is from HMMER, a program which does biosequence analysis using profile hidden Markov models.

Keywords: code optimization, SSE, Altivec, HMMER, biosequence analysis, Hidden
Markov Model

Note: Includes SSE2 source code for HMMER

Submitted By:

Evren Yurtesen (admin)

Submitted On:

04 Oct 2008

File Size:

2,847.22 Kb

Downloads:

File Version:

v4.0.1

File Author: