Project Proposed: This project, building a high-performance parallel computing instrument for numerical and network simulations, aims to support the following research projects: - Reducing simulation time for uniprocessor and multiprocessor systems; - Parallel sparse matrix solution using multi-core cluster with applications in analogue and digital VLSI systems; - Biochemical computations with parallel numerical simulations; - Future Internet switch architecture simulations and protocol validations; - Investigation of the mixing behavior of the nano-particles in a polymer-based binder system for powder injection molding applications; and - Quantum chemistry simulations on biomolecular active sites. The proposed cluster, called High Performance Pan American Cluster (HiPAC), will be used by multiple scientists at the University of Texas-Pan American (UTPA), a predominantly a Hispanic institution. Broader Impacts: The instrument enables research in a minority serving institution that contains an 88% Hispanic and a 63% female population. The project involves student training in multi-core cluster technologies, including software and hardware parallelization. Moreover, student training encourages undergraduates to join graduate studies.
The goal of this MRI project was the acquisition of a PC cluster to develop high-performance parallel-computing algorithms and to conduct large-scale parallel numerical and network simulations. This MRI Project, called High Performance Pan American Cluster (HiPAC), has supported ongoing multidisciplinary research and provided student training using multi-core cluster technologies at The University of Texas-Pan American (UTPA). UTPA is nationally ranked among Hispanic serving institutions: the 2nd for full-time Hispanic students enrolled and the 3rd for bachelor’s and master's degrees awarded to Hispanics. The HiPAC has provided student training, summer program, and workshop for UTPA and other institution, such as South Texas College, in the Rio Grande Valley Region. Fourteen faculty members from eight departments have used the HiPAC for their parallel-computing related research and teaching in three colleges, including electrical engineering, computer engineering, computer science, computer information systems, manufacturing engineering, mechanical engineering, mathematics, and physics and geology departments. Intellectual Merit The HiPAC has supported six research projects. The titles and goals for research projects were as follows: 1) Reducing simulation time for uniprocessor and multiprocessor design with SPEC Benchmark programs and NIC packet buffer management; the purpose of this project was to reduce long simulation time (several days to collect data for one out of 16 benchmark programs) for uniprocessor or multiprocessor design verification using HiPAC; 2) Parallel sparse matrix solution using multi-core cluster; the purpose of this project was to develop an efficient sparse matrix solution based on a weight-directed graph theory for modern high-speed analogue and digital VLSI systems; 3) Computational enzymology; the purpose of this project was to develop parallel numerical simulation to collect data regarding biochemical reactions catalyzed by enzymes; 4) Large scale network simulations for new protocols and high performance Internet switch architectures; the purpose of this project was to develop parallel computation algorithms to perform high-speed packet-level simulations to validate protocol designs for very large networks and the Internet; 5) Mixing of nanoparticles in continuous mixing domains by using particle tracking method; the purpose of this project was to investigate the mixing behavior of the nano-particles in a polymer-based binder system for powder injection molding applications; and 6) Quantum chemistry simulations on biomolecular active sites, and on adsorbate-surface interactions for direct methanol fuel cells anode catalysts (DMFC); the purpose of this project was to use quantum chemistry methods to a) enhance the ability of experimental structural probes of biomolecular active sites by computationally calculating appropriate properties, and b) provide a theoretical basis for designing high efficiency anode catalysts for DMFC by probing carbon monoxide adsorption on the catalyst surface. Key Outcomes The HiPAC has been a core bridge in a parallel computing area between research and training by providing high technology infrastructure for ongoing projects and by enhancing educational programs. For the projects, we developed parallel computing algorithms and simulation models using the HiPAC. Graduate students were hired to work on the projects and received training how to use cluster. The research activities through the HiPAC project also attracted undergraduates to STEM areas through well-prepared education programs. Our research findings from those projects have resulted in the following publications: 1) 20 journal papers have been published; and 2) 24 conference papers have been published or accepted. The education programs include several new courses, parallel computing labs, and a supercomputing workshop for the local community using the HiPAC. Broader Impacts UTPA, which is located in the Rio Grande Valley Region of South Texas, is one of the fastest growing universities in the UT system with an enrollment more than 20,000 (fall 2013). UTPA is nationally ranked among Hispanic serving institutions with the 3rd highest Hispanic students enrolled. The PIs hired graduate students and conducted high-performance computing research using the HiPAC. Supercomputing using the cluster has already been a major research area for an explosive amount of big data in the field of science and engineering. Academic enrichment for this project includes student training based on two essential aspects of cluster technologies: software and hardware parallelization. Software parallelization uses scientific computations of parallel array-dominated codes, such as quantum chemistry biomolecular simulation. Since the cluster project, HiPAC, consists of 860 computing cores, student training courses for undergraduates have provided cluster programming labs to experiment and carry out their research ideas related to parallel simulations, data networking, and matrix computations. One of the objectives for the student training was to encourage undergraduates to join graduate school. URL Project: www.utpa.edu/hipac
