Author(s): Harshita G.Patil | Manish Narnaware
Journal: International Journal of Computer Technology and Applications
ISSN 2229-6093
Volume: 03;
Issue: 01;
Start page: 125;
Date: 2012;
VIEW PDF
DOWNLOAD PDF 
Original page
Keywords: Sequences alignment | dynamic programming algorithms | FASTA
ABSTRACT
This Project shows the issues involved in implementing adynamic programming algorithm for biological sequencecomparison on a general purpose parallel computingplatform based on fine –grain event –driven multithreadedprogram execution model. Fine –grain multithreadingpermits efficient parallelism in this application both bytaking advantage of asynchronous point-to–pointsynchronizations and communication with low overheadsand by effectively tolerating latency through theoverlapping of computation and communication. For thistiling technique can be implemented. Tiling is an importanttechnique for extraction of parallelism. Informally, tilingconsists of partitioning the iteration space into severalchunks of computation called tiles (blocks) such thatsequential traversal of the tiles covers the entire iterationspace. The idea behind tiling is to increase the granularityof computation and decrease the amount of communicationincurred between processors. This makes tiling moresuitable for distributed memory architectures wherecommunication startup costs are very high and hencefrequent communication is undesirable. Our work todevelop sequence-comparison mechanism and softwaresupports the identification of sequences of DNA
Journal: International Journal of Computer Technology and Applications
ISSN 2229-6093
Volume: 03;
Issue: 01;
Start page: 125;
Date: 2012;
VIEW PDF
DOWNLOAD PDF Original pageKeywords: Sequences alignment | dynamic programming algorithms | FASTA
ABSTRACT
This Project shows the issues involved in implementing adynamic programming algorithm for biological sequencecomparison on a general purpose parallel computingplatform based on fine –grain event –driven multithreadedprogram execution model. Fine –grain multithreadingpermits efficient parallelism in this application both bytaking advantage of asynchronous point-to–pointsynchronizations and communication with low overheadsand by effectively tolerating latency through theoverlapping of computation and communication. For thistiling technique can be implemented. Tiling is an importanttechnique for extraction of parallelism. Informally, tilingconsists of partitioning the iteration space into severalchunks of computation called tiles (blocks) such thatsequential traversal of the tiles covers the entire iterationspace. The idea behind tiling is to increase the granularityof computation and decrease the amount of communicationincurred between processors. This makes tiling moresuitable for distributed memory architectures wherecommunication startup costs are very high and hencefrequent communication is undesirable. Our work todevelop sequence-comparison mechanism and softwaresupports the identification of sequences of DNA
