About HKU Grid Point System
The largest processing power Grid computing resource in Hong Kong region
It is the largest processing power Grid Computing resource in Hong Kong region powered by Dell PowerEdge servers with latest Intel Nehalem/Westmere CPU technology and NVIDIA Fermi GPU technology for high performance academic research.
The first phase of the HKU Grid Point project is implemented at early 2010. A new 1920 cores of 64-bit Linux Grid point system is installed to support System Research Group (SRG) of Department of Computer Science and Multiple Disciplines Research Project (MDRP) for HKU researchers on Grid computing, with the total processing power, in term of theoretical peak performance value of 19.43 Tera-FLOPS (and measured Rmax of 14.218 Tera-FLOPS). This system consists of latest hardware such as Dell PowerEdge Servers, Intel Nehalem Quadcore CPU, 10Gb Ethernet (10 Gb/s) switch and 4X DDR Infiniband (20 Gb/s) communication network.
The second phase of Grid Point project is expected to be completed by early-2011. It is planned that more computational power based on latest Intel Westmere 6-core Xeon CPUs, latest NVIDIA Fermi Graphics processing unit (GPU) and fastest communication based on 4X QDR Infiniband (40 Gb/s) communication network will be implemented. It is expected that the theoretical peak performance value of the whole system will be augmented by at least 7.62 Tera-FLOPS after the second phase of implementation.
Total processing power of HKU Grid Point system, by combining 2 implementation phase, will be at least 27.05 Tera-FLOPS in term of theoretical peak performance value.
About HKU Grid Point Project
HKU Grid Point for Systems Research and Applications in Multiple Disciplines is a UGC Special Equipment Grant project and is an extension of China National Grid (CNGrid) in Hong Kong. The project is initiated from the Computer Science Department, teaming up with Information Technology Services, together with the support from a number of researchers from Department of Chemistry, Department of Statistics and Actuarial Science, etc.
This project has a two-phase implementation for the equipment acquisition. The first phase (Sept 2008 – Feb 2010) aims to meet the requirements of Phase II of the China 863 CNGrid project and providing adequate processing power for supporting the compute-intensive researches in multiple disciplines at HKU. In this phase, equipment providing 19.43 Tera-FLOPS of theoretical CPU power was implemented. In the second phase (March 2010 – Aug 2011), the system would be expanded by 7.62 Tera-FLOPS. The system uses 64-bit processors for higher precision arithmetic, multi-cores (4-core or 6-core) CPUs for rapid parallel computation and energy saving CPUs with low heat dissipation generation. The system uses high-density blade servers for space saving, with extreme high-speed interconnect network and dedicated file servers with large disk capacity.
It has set up a state-of-the-art Grid Computing facility to support research in multiple disciplines. The system is connected to other computing resources locally and in Mainland China as a “grid point” of a nation-wide Grid Computing Infrastructure. The equipment is now serving as the main research platform for “systems research” in the Computer Science Department (HKUCS), and be used to support cutting-edge compute-intensive research projects in multiple disciplines that the HKU Information Technology Services (HKUCC) is currently supporting, including Quantum Chemistry, Computational Physics, Computational Fluid Dynamics, Financial Engineering, Environmental Studies, Global Hydrology and Nanotechnology studies. The collaboration between HKUCS and HKUCC will provide enhanced support to these latter projects such as performance optimization of application code running on the new platform and with extension to exploit resources available in the grid infrastructure.
Grid computing becomes the blueprint for the new research computing infrastructures. In Grid computing, multiple independent computing clusters will be joined to form a grid. The Grid approach enables resource virtualization, on-demand provisioning, and service (resource) sharing between organizations. It defines a set of open standards and protocols, to gain access to applications and data, processing power, storage capacity and a vast array of other computing resources over the Internet.
HKU constructed the initial version of the HKU Grid Point with the Gideon cluster installed in 2002. Because of HK’s strategic geographic location, the grid point has been used extensively during the development and testing for the Grid middleware of CNGrid versions 1 and 2. HKU spearheaded the formation of the Hong Kong Grid in 2003. The project aimed at constructing a grid testbed across the six departments from five local institutions (HKU, HKBU, HKUST, PolyU, and CityU) in Hong Kong. By 2010, with the availability of the new HKU Grid Point equipment based on latest computing technology, Hong Kong’s role as one of the key centers in research in High Performance Computing and Grid Computing can be maintained.
This new HKU Grid Point system will foster collaboration among tertiary institutes in grid computing development and compute-intensive research projects. It provides a solid and flexible virtual computing platform for supporting collaboration among universities in Hong Kong.
This new HKU Grid Point system is supported by a high speed direct network route between HKU network and China Science and Technology Network (CSTnet) established at late 2009 to provide a dedicated and high-speed network connection to China National Grid. In addition, HKU also has special research network set-up connecting through Internet2, Trans-Eurasia Information Network (TEIN2), Korea Research Environment Open Network (KREONET) and Taiwan Academia Sinica Grid Computing Network (ASGCNET) directly to the research networks of US, Japan, Singapore, Korea and Taiwan for the purpose of fast data transfer between HKU network and other international grid networks.
 |
 |
 |
 |
||
Grid Computing Services
It provide a collaborative platform with other China National Grid (CNGrid) research institutes for Grid computing research projects as national Grid point based on Grid Operating System (GOS) Grid middleware developed by Computer Network Information Center (CNIC) Institute of Computing Technology (ICT), Chinese Academy of Sciences.
It is the main research platform for SRG and used to support cutting-edge compute-intensive research projects in MDRP, including Quantum Chemistry, Computational Physics, Computational Fluid Dynamics, Financial Engineering, Environmental Studies, Global Hydrology and Nanometer studies.
HKU is already preparing Grid collaboration with Institute of High Energy Physics (China) for simulation on Beijing Spectrometer (BESIII) experiment. The BESIII Monte Carlo simulation generates events at the interaction point, propagates the resulting particles through the detector and records response of detector to these particles. The simulation software inherits about 30 event generators from BES/BESII experiments.
HKU also serves as collaborative research platform for other International Grid organizations that HKU has joined, such as Asia Pacific Grid Policy Management Authority (APGrid PMA), Pacific Rim Applications and Grid Middleware Assembly (PRAGMA) and Enabling Grids for E-sciencE (EGEE).
Computing System Overview
In the first phase implementation, the system is consisted of master/gatekeepers nodes (reserved for login, program development and job queue submission/manipulation), 160 x Dell M610/R610 compute nodes interconnected with 10GbE Ethernet network (of 12.96 Tera-FLOPS processing power) and 80 x Dell M610/R610 compute nodes interconnected with 4X-DDR Infiniband network (of 6.47 Tera-FLOPS processing power), together with 3 dedicated NFS servers & 2 backup servers providing total of 21TB of active and backup data storage. Each of the Dell M610/R610 compute node contains 2 x Intel Nehalem Quad-Core Xeon E5540 processors, 32GB memory and 2 x 250GB local disk storage.
In the second phase implementation, it is expected there will be at least 3.5 Tera-FLOPS additional processing power contributed by compute nodes based on latest Intel Westmere 6-Core processors interconnected with 4X-QDR infiniband (40 Gb/s) network, together with another 4.12 Tera-FLOPS processing power based on latest NVIDIA Fermi Graphical Processing Unit (GPU). Therefore by early 2011, the total processing power of HKU Grid Point system will be at least 27.05 Tera-FLOPS in term of theoretical peak performance value by combining 2 implementation phases.
The 64-bit Linux clusters are installed with several Grid middleware for national Grid and international Grid collaboration, such as Globus, Grid Operating System (GOS) for CNGrid and gLite for EGEE Grid. Some High-performance computing (HPC) and Message Passing Interface (MPI) software packages and libraries are installed such as MPICH, MVAPICH, PGI compiler suites for C, C++, Fortran 77, Fortran 90 and High Performance Fortran, Compute Unified Device Architecture (CUDA) compiler option for General Purpose GPU computing (in second phase), Gaussian 03/09 with TCP LINDA, Torque Resource Manager and Maui job scheduler.
Pioneer in HPC & Grid Computing Development
During the last more than a decade, HKU has been a pioneer in High Performance Computing (HPC) and Grid computing development support advanced research and innovations in Hong Kong. HKU Information Technology Services installed its first large-scale parallel computer system in 1993 (HKUSP2) to support research projects that required powerful computation tools at the University. To meet the continuously increasing demand for more computational power, we keep on advancing the University’s HPC facilities and provide technical assistance to researchers in developing HPC parallel programs and expand the Grid computing development to join various virtual organizations, thus establishing collaboration with other research institutes.
At 2002, HKU Gideon 300 Cluster was installed with processing power of 0.356 Tera-FLOPS and ranked No. 175 on the TOP500 for joining as one of the CNGrid point. At 2004, HKU HPCPOWER cluster was set up with processing power of 1.434 Tera-FLOPS and ranked number 240 on the TOP500. At 2008, HKU has installed the grid nodes for connecting international Enabling Grids for E-sciencE(EGEE) Grid resources and thus be connected to the Taiwan Grid. At 2009, HKU has formally joined as an institutional member of Pacific Rim Application and Grid Middleware Assembly (PRAGMA) Grid. The setup of HKU Grid Certificate Authority was accredited as the first International Grid Trust Federation (IGTF) Accredited Certificate Authority for Grid Computing in Hong Kong in 2009. At this year 2010, we completed the first phase implementation of HKU Grid Point system with processing power of 19.43 Tera-FLOPS. At next year 2011, we will host the prestige international research computing conference, the PRAGMA 20th Workshop cum HKU Centennial IT Conference on Grid Applications and Research Collaboration.
Many innovative research projects requiring powerful computing tools have been facilitated by HKU Information Technology Services. These projects cover a wide range of disciplines, including large-scale statistical simulations, nanotechnology, studies of turbulent flow, bioinformatics, etc. With the help of the new Grid computing services, research collaborations among different institutes from China and Hong Kong region are made feasible.
Research Community
With more than 20 years of experience in supporting intensive computation for cutting-edge researches in the university, we already has formed a large group of research community inside HKU. We welcome any eResearch collaboration on Grid Computing platform with our IT and academic research peers from Hong Kong, China and oversea.