Name of the Researcher(s)

  • Principal Investigator- Dr. D. N. Ranasinghe
  • Research Intern- Duneesha Fernando

Description

Sri Lankan computer science faculties lack high end computational resources for any meaningful advanced research on high performance computing which also makes it difficult to attract full time researchers or to interact with regional research centers for collaboration. Although the concept of ‘computing cloud’ has emerged which enables access to public computing resources as the systems are under 3rd party control, the setting up of a known environment and parameter control essential for research is not possible there. In keeping this objective in mind, we are currently working on setting up of a high end computing cluster at UCSC towards serving related research needs.

 

With the emergence of large scale data (in terms of real time generated streaming data and/or manmade data such as bulk scientific data) and complex computational resources (signified by the emergence of GPU’s), the computer scientist is confronted by the problem of extracting maximum performance out of the resources while at the same time accommodating ever increasing data within limited resources. Towards this, we suggest that as a higher educational institute engaged in research, to have a significant piece of infrastructure that is on par with regional centers of similar status. We are working on setting up a specialized but flexible structure either as a ‘GPU based cluster’, which has potential to develop novel programming methodologies. By the end of phase 1, we wish to implement 3 programming paradigms, namely distributed memory, map reduce and streaming analytics. All programming approaches are being implemented along with incorporation of GPU power. As a case study, Automatic Fingerprint Identification process will be implemented in a parallelized manner.

Related Images

Contact details of the researcher(s)

By admin | January 9, 2018 | distributed . research |