Comparing parallelisation strategies for a bioelectrical brain imaging algorithm
Juhász Zoltán Dr. <juhasz@virt.uni-pannon.hu>
Pannon Egyetem
Brain EEG (electroencephalography) is a routine clinical method to record the electrical activity of the brain. During measurement, typically 25 electrodes (channels) are placed on the scalp, and the electrical signals detected on the scalp are measured and displayed in wave form. The main advantage of the EEG in comparison to other methods (CT, MR) is its high temporal resolution (msec range), the disadvantage is the low spatial resolution and the fact that it measures activity on the scalp not within the brain. In our research programme running at the University of Pannónia, we are developing new methods and algorithms that make it possible to reconstruct the original activation sources of the brain for a given signal pattern and localise it with high spatial accuracy. Such a new imaging method could open up new opportunities in diagnostics.
One of the biggest obstacles in solving this problem is the very high computational intensity of the measurement evaluation algorithms. We are using 64 channels - soon 128. We need to solve the so-called inverse problem; we need to find an electrical source from the detected signal pattern. In this paper, we show one potential algorithm for solving this problem that detects one source only. The number of channels, the large number of measurements and the many potential locations within the brain result in 30-60 second running times for a single measurement point in time. For a typical measurement (several seconds), the evaluation time is measured in hours which is clearly unacceptable in clinical use. The only solution is to use parallel computing.
To find the best implementation, we have implemented the algorithm in several languages an in various parallel programming dialects (Java, C/MPI, C/OpenMP), then executed these on different parallel architectures (PC clusters, multi-core processors, NIIF supercomputer) to understand their behaviour, analyse running time, speedup and scalability.
The paper presents our results and conclusions drawn from these experiments, and in the end show a new possibility, the use of GPUs for speeding up the calculations. We hope to be able to show results obtained on our 4 teraflops NVIDIA Tesla S1070 device.
Scientific applications of the SEE-GRID-SCI project
Kozlovszky Miklós <nt.kozlovszky@sztaki.hu>
MTA SZTAK1
The SEE-GRID-SCI (SEE-GRID elnfrastructure for regional eScience) project is focusing on three strategic important research fields (meteorology, seismology and environmental protection) instead of on the infrastructure as its predecessor south-east European grid project did. These research fields are hosted as Virtual Organizations by the project. The presentation gives an overview about the Virtual Organizations (seismology, meteorology, environmental protection), and the available research applications and services developed during the last two years for these communities. It details how Hungarian researchers can benefit from these applications and can access their services.
Vulnerability assessment for grid systems
Acs Sándor <acs@sztaki.hu>
MTA SZTAKI
There are hundred of computers in a grid infrastructure and the proper operation of such large scale system depends on the security status of these computers. In order to increase the security level of the grid infrastructure a security assessment tool: Grid Site Software Vulnerability Analyzer (GSSVA) has been designed and implemented by SZTAKI. This developed monitoring tool overcomes the issues with communication through firewalls and client side installation without root privileges. GSSVA is using basic grid services, and can automatically explore the installed Linux packages locating on the grid machines. Our approach is extended with a modified PAKITI status monitoring system, which can be re-used as a toolkit for further security parameter monitoring. Some new features were added to the original software also (multi level authentication method based on the user's certificate, history). The paper introduce the inner architecture of the developed GSSVA monitoring tool and assess its usability on several virtual organization's resources. As an outcome of the security/vulnerability assessment of the targeted grid infrastructure, the tool reported a large set of valid security problems with high impact on the infrastructure. We also show some statistic and draw the conclusions based on the observations
ClusterGrid+ "Reorganized Resources for a Better Science"
Márton Iván <niartoni@tiiif.hu>
NIIF Intézet
Aiding the Hungarian scientific research and development NIIF has been involved in offering distributed computing facilities, supercomputers and a country-wide computational grid system, to its member institutions since 2001. One of our principal objectives was to make our services to be the organic part of the cutting-edge European infrastructure via continuous maintenance and development.
An important result of our past years developments has been the establishment of an integrated infrastructure that encompasses the former ClusterGrid as well as additional supercomputing devices. The new ClusterGrid+ thus allows grid users to access modem and dynamic computational services.
In the presentation the already-existing part of the infrastructure and its future development plans will be shown. Besides the refurbished existing sites the new system will provide a magnitude of larger computing capacity than the previous versions has provided, thus further improving the efficiency of the scientific applications.
In the technical developments we put special emphasis both on the ergonomic and broad-scale grid usability as well as on making the new grid infrastructure be attractive to the larger set of users.
(The improvements are developed under the TIOP 1.3.2 project of The New Hungary Development Plan.)
CancerGrid - Application of grid computing to accelerate the selection of potential anticancer agents
Kovács József Dr. <smith@sztuki.hu>
MTA SZTAKI
Grid computing aims at helping the acceleration of computational intensive calculations for its users. In this presentation, a combined framework is introduced where a DesktopGrid system is utilised through a web-based portal called gUSE. The infrastructure can be customised by adding new algorithms, by attaching computing machines and by defining additional workflows describing flow of computations. Moreover, the user interface can also be customised for the users' needs. As an example, the infrastructure built in the CancerGrid EU FP7 project is detailed.
The role of Hungrid in shaping the Hungarian NG1
Hernáth Szabolcs <hernath(jamail.kfki.hu>
MTA KFKI RMKI
Grid resources and services provide an essential part of a successful research and education network. The common European e-science infrastructure reached a milestone with the paradigm shift that reforms the collaboration of the member countries' grid networks, giving a pivotal role to national grid initiatives or NGIs. Thus the establishment and effective operation of the Hungarian NG1 is instrumental in the success of the national e-science program. In view of this, we focus our attention on the tasks and goals that Hungrid, Hungary first and only full-featured gLite grid infrastructure can accomplish in order to help the NGI fulfill its mission.
Morphology classes of convex bodies based on static equilibria
Kápolnai Richard <kapolnai@iit.bme.hu>
BMEIIT
Domokos Gábor Prof. <donwkos@iit.bme.hu>
BME Szilárdságtani és Tartószerkezeti Tanszék
This work is built on a new approach of classifying convex 3D bodies (e.g. pebbles) based on the topology of their static equilibria. An equilibrium point is a surface point on which the body is able to rest on a horizontal plane. An equilibrium point is called stable if the body returns into its original position after some small perturbation, but called unstable if after some small perturbation it will roll away from this point..
Two bodies belong into the same class if they have the same combination of the numbers of stable and unstable equilibrium points. For example, the body christened Gömböc has exactly one stable and one unstable equilibrium thus belongs to the class (1,1). It was proven that for every possible combination of stable and unstable equilibria there exists such a body, however it is unlikely to find a pebble to each class in Nature. Despite their theoretical feasibility, some equilibria classes are so fragile that they can be probably found only for a short, transitional time before the pebble suffers a change due to some natural effect (e.g. abrasion). This is a consequence of the fact that the shape of a pebble could evolve by time and could change its classification as well. Developing a realistic abrasion model of shape transformation could enable us to reveal some of the historical and future evolution of the shape of pebbles.
The surface of the body determines connections between equilibrium points and we define subclasses in every class based on the topology of these connections. Our first goal is to design a combinatorial algorithm which is able to enumerate every possible subclass. Listing all these subclasses may deepen our understanding of the characteristics and time-evolution of natural shapes.
To the best of our knowledge, the complexity of enumerating these subclasses is unknown and our algorithm requires exponential computational resources in the number of equilibria.
We mention as an example that limiting the number of equilibria N<=18 would yield a computational task that needs approximately 3 months on a common personal computer.
Nevertheless to obtain practically useful results it seems this limit is sharp so needs to be reached. Our second goal is to develop a parallel grid application which is able to reach this limit in reasonable time. We present the structure and the operation of the grip application which we run in Hungrid.
Achievements of the EDGES project: infrastructure, applications, tools, and methods
Lovas Róbert Dr. <rlovas@sztaki.hu>
MTA SZTAKI
Kacsuk Péter Dr. <kacsuk@sztaki.hu>
MTA SZTAKI
Recently the EDGeS infrastructure that connects Desktop Grids to EGEE and other Service Grids, announced that the total number of computers connected to the Desktop Grids in EDGeS, has surpassed the 100.000. This milestone shows there are significant additional computational resources available to European scientists that are connected to EGEE via the Bridge technology of EDGeS. The combined size of the Grids interconnected in this way is now over 250.000 processors - one of the largest European computational infrastructures.
The Desktop Grids that collect volunteer resources to EDGeS are:
• SZTAKI Desktop Grid 77.000 computers
• IberCivis 25.000 computers
• Extremadura@home 1.000 computers
• AlmereGrid 3.000 computers
• University of Westminster 1.000 computers
• IN2P3, Paris 1.000 computers
The largest of these systems are based on BOINC technology, while others use XtremWebHEP-E technology developed by INRIA and IN2P3. The EDGeS developed 3G Bridge allows scientists to use more computational power for their applications and send their work to the best suited infrastructure. This way they can reduce the time that is needed to get answers to their scientific problems and also ensure the more efficient utilization of IT resources.
The EDGeS project already ported and installed over 20 applications to the combined infrastructure. This includes medical applications - Patient Readmission; Protein Simulation;
Blender Video rendering; and Fusion research.
"Reaching a milestone of 100.000 computers in Desktop Grids to EGEE through the EDGeS bridge, shows there is a large computing capacity that can be made available to European researchers", said Peter Kacsuk, EDGeS project coordinator." And this is just the beginning:
there are millions of computers of citizens in Europe that could be connected."
"EGEE has a goal to make as many resources as possible available to European researchers,"
says Bob Jones, EGEE Project Director. "Connecting with volunteer grid systems is a key part of our strategy."
The Desktop Grids in EDGeS connect computers from citizens at home and from companies and other organisations. Once joint, citizens can donate otherwise unused computing time to science. Not only do they help science with this, it also contributes to Green IT, because there is no need to build new large computer centres in several cases. Computers at home can be very useful for a number of scientific application areas. Especially those that do large parameter study based simulation or analysis.
The EDGeS project is a European project with 9 partners and 7 subcontractors led by MTA SZTAKI.
EDGeS is supported by a Grant from the European Commission's FP7 1ST Capacities Programme under grant agreement RI-211727.
More information is available on the EDGeS website at http://edges-grid.eu
Commodity data storage
Vitéz Gábor <vitezg@avaxio.hu>
Avaxio Informatikai Kft.
During the last couple of years there are rising sounds about those data storage facilities that are built from ordinary elements such as IDE or, more recently, SATA disks. These facilities can handle terabytes or even dozens of terabytes of data at reasonable access speed and sufficient redundancy level.
In the beginning the low-cost systems were furbished manually, as the official storage solutions like EMC's fiber channel products or the other possibilities distributed by the large vendors were too costly to be purchased by small, low-income organizations.
Typical low-cost storage solutions were the 3ware controller engined, SATA backplane, self- assembled servers that were operated at minimal support level, i.e. when meeting performance bottlenecks and reliability problems, the server operators could only find help on the different mailing lists.
Meanwhile the world of data storage has developed fairly quickly as the different components have become faster and more reliable than their predecessors. The performance of the RAID controllers and the processors improved significantly, while the gigabit Ethernet has also became accessible. Inspired by these changes the market re-shuffled, new promising technologies and standards like iSCSI and AoE have emerged. Ethernet based storage networks have become reality.
The paramount of these changes have been the appearance of the low-end Ethernet-compliant storage solutions. The old fashioned self-made assembly then has gradually been replaced by market-ready solutions allowing storage operators to access the usual guarantee and support conditions at affordable cost.
Exploiting the recent low-end data storage possibilities the innovation-minded can build up high-quality, highly-available and well-performing data storage systems.
ETICS - the Software Engineering Service for the Grid
Takács Eva <eva.takacs@4dsoft.hu>
4D SOFT Kft
ETICS - the International Software Engineering Service for the Grid - stands for
"elnfrastructure for Testing, Integration and Configuration of Software". The ETICS project is a consortium of eight partners (CERN, INFN, Engineering Ingegneria Informatica, 4D Soft, the University of Wisconsin-Madison, MTA SZTAKI, Forschungszentrum Juelich GmbH, VEGA IT GmbH). It provides a framework to help software developers, managers and users to better manage complexity and improve the quality of their software. ETICS service allows users to fully automate the way their software is built and tested. In other words, ETICS provides software professionals with an "out-of-the-box" build and test system, powered with a build and test product repository, with advanced multinode testing capabilities and having integrated several testing tools. Additionally, ETICS system allows taking into account complex dependencies among applications and middleware components and provides a rich environment to perform static and dynamic analysis of the software and execute deployment, system and interoperability tests.
As a system is multi-platform and open source consisting of a build and test job execution system based on different grid middlewares and an integrated set of web services and software engineering tools to design, maintain and control build and test scenarios.
This presentation gives an overview of the system architecture and functionality set and then describes a web testing distributed testing scenario demonstrating the capabilities of the system. Lastly, the Diligent/D4Science/D4Science2 FP6/FP7 project as a user project of ETICS is presented.
Service-oriented Grid research in the S-Cube project
Juhász Zoltán Dr. <juhasz@virt.uni-pannon.hu>
Pannon Egyetem
Németh Zsolt Dr. <zsnemeth(a)sztaki.hu>
MTA SZTAKI
Kecskeméti Gábor <kecskemeti(a)sztaki.hu>
MTA SZTAKI
Kertész Attila <keratt@infu-szeged.hu>
MTA SZTAKI
The goal of the Software Services and Systems Network (S-CUBE) project is to create a unified, multidisciplinary research community that fosters the research and development of service-oriented technologies. The project's primary focus is integration by unifying and coordinating the research agendas of the key research groups of the field, creating Europe wide teaching and training programmes as well as a long term research vision. The project concentrates on the design and implementation methodology aspects of software services that will enable the development, operation and control of complex, high-level service-based systems. The SZTAKI LPDS (using its expertise in grid research) works on the research and implementation of reflexive service infrastructures. The operation os a service-based infrastructure necessitates some level of autonomy typically relying on so-called self-*
properties (serlrf-configuration, self-healing, self-adaptation, self-defence). The paper will provide an overview ot hte S-Cube project activities and describe some interesting aspects of the work performed by the SZTAKI-LPDS.
Securing BOINC through virtualization
Németh Dénes <nemeth.denes(a)iit.bme.hu>
BMEIK
Szeberényi Imre Dr. <szebi@iit.bme.hu>
BMEIK
Deák Szabolcs <szdeak@mail2.ik.bme.hu>
BMEIK
In our presentation we will demonstrate the software environment "BoincInVbox" developed by BUTE. The main aim of the project is to widen the spectrum of applications executable on the BOINC platform and to minimize the work of porting regular applications to the platform.
The BOINC platform has two main problems: heterogeneity and the risks associated with running applications with unknown origins.
One of the main advantages of heterogeneity are that the BOINC platform is available on more than 70 different architectures. This is also its main problem, since in order to be able to obtain resources, a BOINC application must support a large subset of the possible platforms
that BOINC is capable of running on. While this is not a problem for projects such as Seti@HOME, it may prove to be prohibitive for smaller scale projects.
One of the main disadvantages of the BOINC platform itself it that the applications themselves have a rather high level of access to the system they are running on. In order to minimize the associated security risks, one needs to chose from a smaller set of trusted applications. This in turn means that a new project has to pass a lengthy evaluation to be added to the trusted set.
We shall present a solution to the above mentioned problems in the form of virtualization. In our solution the BOINC environment is contained by an operation system that itself is run on a virtual machine, thereby completely separating the BOINC environment from the host system. This also provides a solution for the problem of heterogeneity, since the same operating system can be used in a virtual machine across a range of hosts.
We shall present furthermore how easily the implemented solution can be installed, configured and maintained on a Windows host as opposed to configuring grid-type networks.
To infinity... and beyond!
Sepp Norbert <norbert_sepp@hu.ibm.com>
IBM Magyarország
IBM has a holistic approach in creating future trends of IT. This approach is also followed in
IBM has a holistic approach in creating future trends of IT. This approach is also followed in