HP-SEE Wiki

From HP-SEE Wiki

1 Virtual Research Communities
2 Fast Track access to HP-SEE infrastructure for new user communities
- 2.1 Fast Track Applications
3 User Guidelines
4 Developer Guidelines
- 4.1 HPC programming techniques
5 HPC Infrastructure Operations
6 Network Operations
- 6.1 NRENS
  - 6.1.1 NAS RA
  - 6.1.2 AzRENA
- 6.2 Monitoring and Management Tools
7 Software Stack and Technology Watch
8 Dissemination and Training
9 HPC Initiatives, Procurement and Organization Guidelines
10 Templates

Virtual Research Communities

Computational Chemistry VRC

Overview

Computational chemistry and material science is one of the highlighted research areas in computational science and a typical heavy user of HPC resources. The computational technologies are an indispensable tool for investigations in domains like quantum molecular dynamics, molecular modelling, nano-technology and design of new materials. Considering the size of the problems to be studied, the required calculations are extremely computationally intensive. Thus HPC would greatly facilitate the proposed work allowing the researchers to deal not only with “pilot” or model systems but to work on big and complicated real systems, which are physically and technologically more significant and challenging. These studies will extend understanding of some fundamental science issues and are of practical importance for pharmaceutical industry, nanotechnology, biomedicine, and many others.

Initially Computational Chemistry VRC supports 7 applications with main developers in 6 SEE countries, collaborating with scientists from more than 20 advanced research centers in Europe.

Computational Chemistry Applications

Acronym	Application Name	Main developer	Stage
CFDOF	CFD Analysis of Combustion	Faculty of Mech. Engineering, University of Banja Luka (UoBL), Bosnia - Herzegovina	testing 09.2011
CompChem	Quantum Mechanical, Molecular Mechanics, and Molecular Dynamics computation in chemistry	Univeristy of Belgrade, Faculty of Chemistry, Republic of Serbia	production 09.2011
FMD-PA	Design of fullerene and metal-diothiolene-based materials for photonic applications	Computational Chemistry Group of NHRF, Greece	production 08.2011
HC-MD-QM-CS	Hybrid Classical/Quantum Molecular Dynamics – Quantum Mechanical Computer Simulation of Condensed Phases	UKIM, Institute of Chemistry, Faculty of Natural Science and Mathematics, FYROM	production 08.2011
ISyMAB	Integrated System for Modeling and data Analysis of complex Biomolecules	IFIN-HH/DPETI, Romania	production 05.2012
MDCisplatin	Molecular Design of Platinum Group Metal Complexes as Potential Non-classical Cisplatin Analogues	Acad. Roumen Tsanev Institute of Molecular Biology-BAS, Bulgaria	production 07.2012
PCACIC	Principal component analysis of the conformational interconversions in large-ring cyclodextrins	IOCCP-BAS, Bulgaria	production 08.2011

Computational Physics VRC

Overview

Computational Physics is nowadays the main beneficiary of the scientific HPC, large-scale numerical computations being necessary whenever the complexity of the physical systems investigated does not allow the derivation of an analytical solution. The main objective of the Computational Physics VRC is to join together the various physics research teams from the SEE area and to provide them access to a powerful HPC infrastructure and tools which will make possible their participation in multidisciplinary and international collaborations. For this purpose, software developers from 7 countries (Albania, Bosnia and Herzegovina, Bulgaria, FYR of Macedonia, Republic of Moldova, Romania, Serbia) contributed with 12 applications in the fields of High Energy and Particle Physics, Plasma Physics, Physics of Condensed Matter, Atomic Physics, and Computational Fluid Dynamics. The application range extends from nanoelectronics, micro-devices optimization and the modeling of robotic devices for biomedicine, to improved means for feature detection in satellite images, which leads to better mapping, localization and search services.

Computational Physics Applications

Acronym	Application Name	Main developer	Stage
AMR_PAR	Parallel algorithm and program for the solving of continuum mechanics equations using Adaptive Mesh Refinement	Institute of Mathematics and Computer Science, Laboratory of Mathematical Modelling, Republic of Moldova	production 09.2011
EagleEye	Feature Extraction from Satellite Images Using a Hybrid Computing Architecture	University Politehnica of Bucharest / Computer Science and Engineering, Romania	production 08.2011
FAMAD	Fractal Algorithms for MAss Distribution	Institute for Space Sciences, Romania	testing 07.2011
FuzzyCmeans	Parallel Fuzzy C Means for classification/Feature detection category	West University of Timisoara/Computer Science Department, Romania	production 08.2011
GENETATOMICS	Genetic algorithms in atomic collisions	Institute of Physics, Faculty of Natural Science and Mathematics, UKIM, FYROM	production 09.2011
GIM	Geophysical Inversion Modeling	Polytechnic University of Tirana, Albania	testing 05.2011
HAG	High energy physics Algorithms on GPU	Institute for Space Sciences, Romania	testing 09.2011
HMLQCD	Hadron Masses from Lattice QCD	University of Tirana, Albania	production 07.2011
NUQG	Numerical study of ultra-cold quantum gases	Scientific Computing Laboratory, Institute of Physics Belgrade, Serbia	production 08.2011
SET	Simulation of electron transport	Institute of Information and Communication Technologies, Bulgarian Academy of Sciences, Bulgaria	production 08.2011
SFHG	Self Avoiding Hamiltonian Walk on Gaskets	Faculty of Mechanical Engineering, Dept. of Thermomechanics, University of Banja Luka, Bosnia and Herzegovina	2012
SIMPLE-TS_2D	Finite Volume Method for calculation of 2D gas-microflows using standard MPI	Kiril Stoyanov Shterev and Stefan Kanchev Stefanov, Institute of Mechanics – BAS / Department of “Complex and multiphase Flows”, Bulgaria	production 08.2011

Life Sciences VRC

Overview

Life Sciences depend heavily on the use of HPC for both data mining and data integration as well as for the simulation of biological systems. HPC technologies are essential for research areas such genome analysis, expression profiling, -omics analysis and biological simulations, whereby a vast amount of experimental data needs to be analyzed and synthesized into reasonable hypothesis. Thus HPC would greatly facilitate the various applications described in this project, enabling the respective research teams to study questions that have thus far been intractable due to their high computational complexity. The use of HPC in the Life Sciences applications with help further our understanding of basic problems in the fields of DNA sequence analysis, comparative genomics, and brain modeling among others and can be of great importance for the health sector.

The Life Sciences VRC supports 7 applications with main developers in 5 SEE countries (Greece, Hungary, Montenegro, Armenia, Georgia) working in the areas of computational biology, computational biophysics, DNA sequence analysis and computational genomics. The various projects involve collaborations with numerous scientists both in Europe and the U.S. and will foster the development of new collaborations among the participant SEE countries.

Life Sciences Applications

Acronym	Application Name	Main developer	Stage
CMSLTM	Computational Models of Short and Long Term Memory	George Kastellakis, IMBB-FORTH, Greece	production 09.2011
DeepAligner	Deep sequencing for short fragment alignment	Miklos Kozlovszky, Gergely Windisch, Judit Márton,Obuda University (OU), John von Neumann Faculty of Informatics, Biotech Group, Hungary	production 02.2012
DiseaseGene	In-silico Disease Gene Mapper	Miklos Kozlovszky, Gergely Windisch, Judit Márton, Obuda University (OU), John von Neumann Faculty of Informatics, Biotech Group, Hungary	production 02.2012
MC-WND	High resolution medical tissue image classifier	Miklos Kozlovszky, Gergely Windisch, Obuda University (OU), John von Neumann Faculty of Informatics, Biotech Group, Hungary	production 03.2013
DNAMA	DNA Multicore Analysis	School of Computer & Communication Sciences, Laboratory for Computational Biology and Bioinformatics, RAxML software, Montenegro	production 11.2011
MDSCS	Molecular Dynamics Study of Complex systems	Dr. Armen Poghosyan, Dr. Hrachya Astsatryan, National Academy of Sciences of the Republic of Armenia	production 07.2011
miRs	Searching for novel miRNA genes and their targets	Anastasis Oulas, IMBB/FORTH, Greece	production 05.2011
MSBP	Modeling of some biochemical processes with the purpose of realization of their thin and purposeful synthesis	Jumber Kereselidze, Tbilisi State University Department of Natural Science, Georgia	production 11.2011

Fast Track access to HP-SEE infrastructure for new user communities

Since 2012, a continuous call for proposals is opened for fast track access to HP-SEE resources, targeted at new user communities. Researchers from the South-East Europe region covered by the project can run their scientific applications on resources provided by HPC centers in Bulgaria, Serbia, and Romania, and benefit from the technical support of HP-SEE experts during a two- or three-months period. The resources offered are limited to 4 nodes (48-64 CPU cores) and 4x480 GPUs cores per run. Developers of Chemistry and Life Science applications are especially encouraged to apply.

Fast Track Applications

Acronym	Application Name	Domain	Main developer	Stage
LNAWENR	Asteroseismic pipeline and Web based portal based on LNAWENR software code to be used in the KASC/KEPLER Space Mission	Space Science	Marian Doru SURAN, Astronomical Institute of the Romanian Academy, Romania	ended Dec. 2012
SyMBA_GPU	Study of the initial formation stages of the solar system	Space science	Kleomenis TSIGANIS, Aristotle University of Thessaloniki, Greece	running
ASEP2D	Simulation of 2D Asymmetric Exclusion Processes	Computational Physics	Margarita IFTI, University of Tirana, Albania	ended June 2013
SLVP	Selective Load Value Prediction	Engineering	Adrian FLOREA, 'Lucian Blaga' University of Sibiu, Romania	ended Aug. 2013
NAMD	MelanoMD	Computational Chemistry	Adina MILAC, Institute of Biochemistry of the Romanian Academy, Romania	ended July 2013
SeqSize	Sequencing Size and Power	Life Sciences	Dean PALEJEV, Bulgarian Academy of Sciences, Institute of Mathematics and Informatics, Bulgaria	ended Aug. 2013
TMDC	Electronic structure, optical and transport properties of layered transition-metal dichalcogenides and their intercalated compounds	Computational Physics	Yurii CHOMAKOW, Institute of Applied Physics, Academy of Sciences of Moldova	ended July 2013
PSoBBB	Population Status of Brown Bears in Bulgaria	Life Sciences	Anna JORDANOVA, Institute of Molecular Biology, Bulgarian Academy of Sciences	running
MUSICO	MUscle SImulation COde	Bioengineering	Djordje NEDIC, Faculty of Science, University of Kragujevac, Serbia	running

User Guidelines

Developer Guidelines

HPC programming techniques

Code programming and optimization

Architecture-specific optimization

Setting up the software environment

Choice and usage of libraries and application tools

Compiler-level optimization

Compiler-level optimization

Using development tools

Input/Output methodologies

Usage of job submission systems

Scalability testing and optimization

Scalability testing and optimization

Benchmarking

HPC Infrastructure Operations

HPC Centers

Bulgaria

Hungary

Serbia

Resource centre PARADOX

Romania

MKD

Resource centre HPGCC

Core Services

Monitoring Tools

Pilot Call Resources

Network Operations

HP-SEE project connects South Caucasus National Research and Education Networks (NRENs), Azerbaijan Research & Educational Networking Association (AzRENA) and National Academy of Sciences of Armenia (NAS RA) to GEANT with 45 Mbps connections. These two NRENs were connected to GEANT at Q4 of 2011 and these connections will be valid during the project life time.

HP-SEE project built a virtual NOC composed of the NOC personnel from AzRENA, NAS RA and ULAKBIM and this virtual team operates the network. Network operations includes monitoring and management of Sofia-Baku and Sofia-Yerevan links.

NRENS

NAS RA

The backbone map of NAS RA is given in the following figure. This map basically shows the users of the GEANT link provided by HP-SEE project.

AzRENA

The backbone map of AzRENA is given in the following figure. This map basically shows the users of the GEANT link provided by HP-SEE project.

Monitoring and Management Tools

HP-SEE NOC uses Nagios, MRTG & RRDTool, CACTI and NfSEN for management and monitoring purposes and all of these tools are open-source. The findings of these tools and the usage graphics of HP-SEE link are available at: http://network.hp-see.eu

Nagios

Nagios is an open source network monitoring application which watches the hosts and services specified. Simple alerting mechanisms such as e-mail and sms are the equipments of Nagios to inform the NOC personnel in case of possible problems on the services or hosts being watched. In this deployment a PING service is associated with these defined hosts so that Nagios can continuously pole the link between ULAKBIM and the two NRENs.

MRTG and RRDTool

The Multi Router Traffic Grapher - MRTG is one of the most widely used to generate graphics of link usages. MRTG uses Simple Network Management Protocol (SNMP) to monitor network devices and draws pictures showing how much traffic has passed through each interface. Although MRTG generally uses SNMP as an input, it can also use other input methods such as output of files, plugins or custom scripts. HP-SEE monitor is equipped with MRTG and with the related configurations, it generates the graphics of Sofia-Baku and Sofia-Yerevan links in daily, weekly, monthly and yearly manner. In each 5 minutes, the usage statistics are gathered from AzRENA and NAS RA routers and the related graphics are updated.

WeatherMap

WeatherMap is a network visualization tool, uses data already collected from network devices and creates your network map. HP-SEE NOC uses WeatherMap to read the link usage data MRTG and generate HP-SEE links maps with link utilization ratios.

CACTI

CACTI is PHP based frontend by combining all the useful functionalities of RRDTool, SNMP and MRTG. HP-SEE NOC uses CACTI to aggregated data from Yerevan and Baku routers and to generate on demand graphs. HP-SEE NOC currently generates CPU usage, ping latency and traffic usage graphics via CACTI. All of these graphics are also generated in hourly, daily, weekly and monthly manner.

NfSen

NetFlow is a network protocol developed by Cisco Systems for collecting IP traffic information. NetFlow has become an industry standard for traffic monitoring. Today it is supported by almost all router manufactures providing similar functionality with a differed label and standards like sflow, ipfix. Open source flow generator software implementations such as softflowd, frobe can be used to generate a flow on Linux, FreeBSD, NetBSD and OpenBSD operating systems. In order to capture and store the generated flow in a analyzable format many flow capture and analyzer software were developed by open source community. NfSen, which was developed by GEANT2 project, is a graphical web based front end of nfdump and it is used to store and analyze network traffic flow traces in combination with NfDump. A flow trace is accounting information aggregated from the packets travelling over a router. Accounting information for all packets with the same 5-tuples (source IP address, source port number, destination IP address, destination port number and protocol) is aggregated under the same flow trace.

The HP-SEE Advanced Traffic Management System is installed by using two NfSen instances. One is used to capture the flows exported by Yerevan router and one is used to capture flow by AzRENA router. This system is a living one, which will be enhanced by addition of currently available NfSen plug-ins such as security related once in parallel to the growing needs of HP-SEE NOC.