Parallel
Programming to Increase Performance of Binary Search Tree
Cipta
Ramadhani1 , Wahyu Ramadhan2
1Jurusan
Teknik Elektro, Fakultas Teknik Universitas Mataram, Indonesia
2STMIK
Syaikh Zainuddin NW Anjani, Lombok Indonesia
Binary
Search Tree is
one of the most useful fundamental data structure for dynamic datasets. It can
be used to sort numerical
data, character string and any other data in a sophisticated way. However, the performance in term
of processing time for Binary Search Tree increases when a data become
large, especially
for data mining and detecting pattern in DNA sequences. To increase speed performance of
Binary Search Tree, there are several strategies were applied such as using
super computing
or conducting parallel programming. In this research we applied parallel
programming by using Message Passing Interface (MPI) to sort numerical
data. This method is more efficient and less cost compare with using super
computing. The
basic idea of parallel programming is dividing root into sub-root and
executing sub-root simultaneously on multiple cores. MPI provides a set of
send and receive function that allow data to be processedon multiple cores. This method is very
useful to find and delete data faster than in a sequential way, but the consequence is the coding time takes
little bit longer than usual.
Key
word : Binary search tree, Message passing interface, Parallel Programming.