Some examples of our HPC research accomplishments.
- Thanavanich, T., Uthayopas, P. Efficient energy aware task scheduling for parallel workflow tasks on hybrids cloud environment (2013), 2013 International Computer Science and Engineering Conference, ICSEC 2013, art. no. 6694749, pp. 37-42. http://www.scopus.com/inward/record.url?eid=2-s2.0-84893589944&partnerID=40&md5=cbc327ad115f242533ae2f290b41e089 DOI: 10.1109/ICSEC.2013.6694749 DOCUMENT TYPE: Conference Paper SOURCE: Scopus
In our research, the challenge of scheduling a parallel application on a cloud environment to achieve both time and energy efficiency is addressed. Two energy-aware task scheduling algorithms called the EHEFT and the ECPOP are proposed to address the challenge. These algorithms have the objective of trying to sustain the makespan and energy consumption at the same time. The concept is to use a metric that identify the inefficient processors and shut them down to reduce energy consumption. Then, the task is rescheduled to use fewer processors to obtain more energy efficiency. The experimental results from the simulation show that our enhanced algorithms not only reduce the energy consumption, but also maintain a good quality of the scheduling. This will enable the efficient use of the cloud system as a large scalable computing platform.
- Chidchanok Choksuchat and Chantana Chantrapornchai. 2013. On the HDT with the tree representation for large RDFs on GPU. Proceedings of the 19th IEEE international conference on Parallel and Distributed Systems (ICPADS), Dec 15-18, 2013, Soul, Korea, 651 – 656.
In this research project, we aim to construct efficient data representations and algorithms for searching large volumes of Resource Description Framework (RDF) data that are widely used for representing semantic data on the Internet. RDF data are organized in triples consisting of a subject, a predicate and an object, which together express the semantics of the considered domain. This research is funded by Thailand Research Funds (TRF) and TRF-DAAD under the co-supervisor by Prof. Sergei Gorlatch, TRF-Newton Fund under the co-supervisor by Prof. Jeff Z. Pan, as well as the NVIDIA hardware grants.
Related publications on this research are as:
– Chidchanok Choksuchat and Chantana Chantrapornchai. 2013. Large RDF representation framework for GPUs case study key-value storage and binary triple pattern. In Proceedings of International Computer Science and Engineering Conference (ICSEC), September 4-6,2013, Bangkok, Thailand, 13-18.
– Chidchanok Choksuchat and Chantana Chantrapornchai. 2013. Experimental Framework for Searching Large RDF on GPUs based on Key-Value Storage. In Proceedings of the10th International Joint Conference Computer Science and Software Engineering (JCSSE), May 30-31, Khon Kaen, Thailand, 171 – 176. DOI:10.1109/JCSSE.2013.6567340.
- Banpot Dolwithayakul, Chantana Chantrapornchai and Noppadol Chumchob. 2015. Utilizing the pipeline framework and state-based nonlinear Gauss-Seidel for large satellite image denoising based on CPU-GPU cores, Int. J. Computer Applications in Technology, Inderscience. 52(2). 2015.
Our research group also works on many-core computing algorithm and applications, particularly GPUs. We are also running another research for parallel image and video denioising. The algorithm is based on parallel Gauss-Siedel and sliding windows. It runs on many-core architecture and multi-core architecture as well. Several papers are published earlier about the methods and applications. The algorithm is planned to extend to support multi-GPUs considering streaming and asynchronous memory transfer. Also, the peer-to-peer memory will be considered. The research is partly funded by Thailand Research Funding and the NVIDIA hardware grant.
- Kanok Huankumnerd, Banpot Dolwithayakul and Chantana Chantrapornchai. 2014. Parallel simulation of HGMS of weakly magnetic nanopartilces in irrotational flow of inviscid fluid, Scientific World Journal, Vol. 2014, Article ID 519654, 12 pages. http://dx.doi.org/10.1155/2014/519654.
Furthermore, one of the master student is working on parallel simulation on nanoparticle capturing using HGMS method which is a useful method for applications such medical area, eg. drug targeting for tumor treatment, or in a factory, eg. for water treatment. The work runs using OpenMP support on our Xeon Phi. (e.g. https://www.youtube.com/watch?v=F0jQNMshJu4) as in Figure 1.7 and this work is being extended to test against multi-GPUs with streaming features for scalable problem size while the parallel visualization of the simulation is being constructed. The research is partly funded Silpakorn University Research and Development Institute Funding, Thailand, Thailand Research Funding as well as the NVIDIA hardware grant.