Management structure and procedures
Quality of the management of the project
The quality of the project management is assured by taking into account that the participating partners into the project „GPU in GRID for space science and technology”, acronym (GUIDE), have already signed a Partnership Agreement, which indicates that Institute of Space Sciences (ISS) is the Project Coordinator and S.C. TRANS SPED SRL is the partner 1 (P1).
The Coordinating Partner has considerable experience in managing projects of the size and complexity of GUIDE. The management structure proposed for GUIDE is built on the basis of the experience gained by the partners in similar national or international projects and in particular in the Partnership series of projects, with the appropriate corrections from the experience gained already. The structure has been designed to deal with the typical project management challenges and problems associated with a diverse collection of scientific participants and all nine management areas as defined by the Project Management Institute (integration, scope, time, cost, quality, human resources, communications, risk and procurement). The main focus is the need to ensure that the management of GUIDE provides the environment to deliver dependable quality production services. This will be essential for the effective exploitation of the results throughout the different scientific user communities interested in this project and its overall success. In this respect the management structure of the project will be in accordance with the management plan described below, while respect the articles of the grant agreement to be signed with the ROSA and the consortium agreement that has been signed among the partners.
Detailed organisational structure and decision-making mechanisms
The aims of GUIDE project requires a central management organisation flexible enough to smoothly integrate with the pre-existing management infrastructures, while on the other hand it will require enough authority to lead, implement and control the overall development and the provision of the planned deliverables. The overall Coordination of the GUIDE Project will be split in five categories (see figure 1): Scientific Coordination, Financial Coordination (which will be nominated by the Coordinator – ISS), Technical Coordination, Dissemination/Communication Coordination and Quality Assurance Coordination (which will be nominated by the project Steering Committee (SC)) Also, the Task Leaders will assist the Coordinators above.
The Steering Committee (SC) is the ultimate project authority, with collective responsibility for the project as a whole and for long-term strategic decisions within the project. The SC main management goals are: to manage and direct the project; to confirm that the overall resources used are consistent with the work performed; to ensure that the progress of the work adheres to the Partnership Agreement; to control the progress of all Activities; to establish the right basis for the successful implementation of results; to ensure that the agreements within the consortium related to the Contract with the ROSA conform to the obligations of the Contract with the ROSA and maintain copies of the agreements for inspection by the ROSA. The SC main technical goals and functions are: to assume the responsibility of the technical direction of the project; to coordinate the internal review of ROSA deliverables; to maintain the technical coherence of the project; to solve any technical conflicts between tasks; to make all decisions related to technical matters; to design a quality control discipline; to control the progress of all technical activities.
GUIDE Management Structure
The SC is chaired by the Scientific Manager and is made up of one representative from each partner. Each project partner has one voting right for matter regarding project coordination and management. Tasks Leaders can participate, by invitation, without voting rights. Procedures and rules for attendees, representatives, quorum, absence, voting procedures, conventions and agendas, meeting minutes, changes in formal representation, frequency and places of meetings will be discussed at the kick-off meeting.
The Project Scientific Manager (PSM) has operational responsibility for the project and its day-to-day management. He is the focal point for all management activities in the project and it will be the person to whom the consortium will delegate the responsibility of being the only official interface between the project and the ROSA. The responsibilities of the PSM will be: to coordinate all activities within the project; to ensure that the project work program is maintained; to detect main deviations in time and budget; to resolve any deviation through regular contacts with the participant organizations; to be the channel for submitting all documents and for general liaison with the ROSA for all project matters, both administrative and technical, assisted by the members of the project he considers necessary; to coordinate the preparation, maintenance and distribution of control reports, including records of costs, resources and time scales associated with the Project; to organise management meetings, and internal and external reviews of the project. The Project Financial Manager (PFM) is working in close collaboration with the PSM. The main responsibility of the PFM is to ensure that the financial flow within the project is carried out efficiently. The main responsibilities are: to help PSM detect main deviations in budget and take care of the proper utilization of project resources; to advice all partners of the project for financial matters and check the eligibility of costs; to be the channel for submitting all documents and for general liaison with the ROSA for all project financial matters; to carry out the production, maintenance and distribution of control reports, including records of costs, resources and time scales associated with the Project. The PSM & PFM will also receive assistance from the Legal Advisor during the project negotiation and execution.
The Project Technical Manager (PTM) is nominated among the members of the Steering Committee and his responsibilities will be: to coordinate the technical work of the project; to keep up the technical objectives; to control the progress of all technical activities; to coordinate the technical issues of ROSA reviews; to coordinate the production of project deliverables. The Project Dissemination/Communication Manager (PCM) is nominated among the members of the SC.; the responsibilities of the PCM will be: to coordinate the dissemination and communication work of the project; to promote widespread use of the GUIDE infrastructure and user services. The Project Quality Assurance Manager (PQAM)is nominated among the members of the SC; the responsibilities of the PQAM will be: to assure the quality of the work of the project; to ensure the correct progress of the work so that the results of the project adhere to the contract; to implement and perform the appropriate quality control on project deliverables.
The Phase Leaders (PL) will have the following responsibilities: to manage all activities in the Phases; to ensure that the project schedule is maintained, notifying the PSM/PTM of any discrepancy; to ensure that the objectives and results of the activities within his/hers Phases are achieved, and that the deliverables for the ROSA are available according to the programme; to arrange technical reviews of specific milestones as required by the SC or the ROSA; to ensure technical liaison for the work in his/her PL with the PTM; to supply the PSM/PFM (in a specific section within its own periodic control reports) with all the information related to the PL, outlining progress within such PL and highlighting all problems and actual or expected delays; to supply the information required for the reports for the ROSA.
Information flow.The main tool that will be used to maintain the project management information system will be the Project Notebook; a Microsoft Excel file that maintains in worksheets all the project’s management information: project definition, staff coordinates, mailing lists members, Gantt, WBS, documents archive, planned effort/budget, actual effort/budget, activity input/objectives/output, etc. GUIDE will also make full use of a comprehensive project management portal to ensure a constant, high-quality information flow to all participants. This portal will act as a repository for all relevant documents, distribution lists, electronic bulletin boards and meeting information. Electronic Mail will be used for transmitting word-processor files, spreadsheet files, binary and source code with: internal control reports, contributions to periodic review reports, small software modules, working papers, informal communications. Final decision on the application software for interchange of files will be taken in the kick-off meeting.
Meetings.The SC shall meet half of year at the request of its chairman (unless otherwise stated, these meetings are ordinary meetings of the SC and discuss the full range of project management issues) or at any other time when necessary (extraordinary meetings). A calendar with the dates for the ordinary meetings shall be agreed yearly at the kick off meeting for the first year or at the last ordinary meeting of a project year. The SC will select the location and will be responsible of all organizational items. Meetings shall be convened by the chairman with at least fifteen (15) calendar days prior notice with an agenda. The Agenda of meetings will always include the approval of the minutes of the last meetings held since it. Minutes of the meetings of the SC shall be transmitted in two weeks time after each meeting is held. The minutes shall be considered as accepted by the other Parties if, within fifteen (15) calendar days from receipt, no one has objected in a traceable form to the Coordinator.
Deliverables. The Phase Leader is responsible to see that the deliverable is drafted according to the agreed procedure. Two working days before the final deadline, the Deliverable Editor should send the final version in electronic format to the Project Coordinator and the Phase Leader.
Individual participants
Beneficiaries
The GUIDE consortium consists of 4 beneficiaries (partners), divided into one Public Research Institution, and one private companies (SME) specialized in PKI certification.
GUIDE beneficiaries are firmly committed to the GUIDE vision. The participants-beneficiaries carry diverse technological background, so as to ensure achievement of the project objectives, transfer of technical know-how and guarantee overall success. In more detailed consortium participant roles are as follows:
CO-Institute of Space Sciences -ISS (http://www.spacescience.ro/)
ISS is a legal subsidiary of the National Institute for Lasers, Plasma and Radiation Physics, operating under the auspices of the Romanian Ministry for Education, Research and Innovation. Its mission is to participate to the world effort of knowledge and use of the cosmic space for increasing the life quality on the Earth, and also to have an important contribution for developing and maintaining the Romanian expertise and capabilities in the cosmic space area. The Grid Computing Research Group from ISS consists of experienced and young talented researchers in the distributed computing and also in some application domains of the Grid computing and HPC. The group is experienced in data mining techniques, geometric modeling, Monte Carlo simulations, benchmarking studies, parallel computing algorithms, etc. The IT team at ISS administrates a High-performance computing system, which is dedicated to work in a standalone classical distributed system and also in two brands of Grids – Alien (http://alien.spacescience.ro) and LCG. Also, ISS developed a system based on an HPC cluster counting about 320 logical CPUs with 2GB/CPU, used to perform advanced distributed and parallel computations for space sciences for ESA missions in RoISS (PLANK http://planckgrid.spacescience.ro, CLUSTER2, VENUS-EXPRESS http://rospacegrid.spacescience.ro) and also with GPUs. The ISS HPC cluster has a very good availability, and a performant automatic management system. The Institute of Space Sciences currently operates a high performance distributed computing cluster of moret han 650 CPUs and 140 Terabytes storage, which provides support for LCG, and also for the Alien and Plank VOs. ISS became member of the EGEE project in 2006. In 2012 ISS will develop a GPU cluster powered by Nvidia devices. The total processing capacity will be around 80 TFLOPS, 20000 Core-CPU. The current ISS effort for developing a High Performance Computing system is supported by national and European funding.
Role in the project:
Institute for Space Scienceswill coordinate, monitoring and controlling this project and will provides the Project Management ( with ERP solution)
ISS will design and implement the GRID infrastructure, certification of GRID sites, the Monte-Carlo simulation tools module, GPU specific applications for High Energy Physics, data analysis software simulated data, provide dissemination tools for computing infrastructure(user guides, software scripts, organization of kick-off meeting workshop etc.)
ISS will implement the CUDA APIs in distributed environment. ISS will support the computing resources( GRID and GPU hardware/software infrastructure) for the partners.
ISS will start the procedures of technologic transfer of the methods, algorithms and GPU software to the partner 1 TRANS SPED LTD.
P1. TRANS SPED (http://www.transsped.ro)
(TRANSSPED) is one of the most important Certificate Service Providers at national levels, with the authorization to provide qualified digital certificates that can be used not only for creation of electronic signatures with legal value but also for safe authentication of users, encrypting and signing of data etc. Also TRANSSPED has an rich experience in designing and implementing of PKI complex solutions based on the most advanced technologies in the field
Role in the project:
The main contribution in the project activities proposed is the realization of a GRID workbench for production of results from data analysis, in extreme conditions of security and authentication using the two types of academical and commercial certificates provided by ROSA and TRANSPED. The creation of a specific documentation (Certification Policy and Practice and Procedure Code), GRID certificate application on the non academical zone. The existence of these documents is essential in order to define the hierarchical authentication structures. The realization of case study for implementation of non-academical GRID structures.
Consortium as a whole
The consortium participants have already demonstrated their experience and skills and have a track record of participation in national and European efforts in previous programmes and projects. Furthermore, all participants have stated their commitment to allocate the adequate human resources to guarantee the successful implementation of the project. The reliability of the partnership in delivering high-quality results is supported by related projects, included in project proposal, interested in the development of the region and eInfrastructures and specifically HPC.
Furthermore, the project participants – which are major lead institutes in the region for computing aspects of eInfrastructures
Previous managerial experience of the project coordinator constitutes from the beginning a warranty for the good use of the material and financial resources. Envisioned are both the improvements of the existing resources so that it meets the requirements of the national and international projects and the financial motivation of the participating team as to increase its experience and competitiveness. The delivery schedule was established after careful planning with all the participants involved in this project. Moreover, the schedule and the allocation of financial resources follow closely the objectives of the National Research, Development and Innovation Plan, preparing the material infrastructure and the scientific personnel for research and development activities at European level.
The project studies are developed in the following directions: GRID middleware, GPU developing, benchmarking in essence, covered much of the directions followed by the international collaborations.
A complex interconnection is between the members of collaborations for developing, installing, testing, deployment and benchmarking.
The end product of the research is requested by the SME (TRANS SPED), which has major interests in PKI certification for different kind of application useful for public and enterprises fields(Industry, Stocks, Finance, Hospitals, Space Science, Material Science, etc).
Given the interest of TRANS SPED in obtaining a product which is to be multiplied, it will be determinant factor in all decisions about the final product, described in Business Case.
Resources to be committed
The GUIDE consortium is committed to amass all the necessary resources for the successful implementation of the project, in terms of human resources, material, facilities and finance. The participant’s institutes have been carefully selected to satisfy the need for both skilled personnel and related material such as equipment (specifically GRID and GPU installations) and facilities. In the previous sections, the consortium description, including profiles, CVs and the roles has been described in a vertical way (per participant) and the consortium complementarity has been demonstrated on a per Phase basis. In this section we present the resources in a horizontal way (per Phases), covering the human resources, travel and the necessary material including equipment and facilities.
Methodology and associated work plan:
The organization of the project will be centered around four Phases detailed in Work Plan (CLIC).
Currently, algorithms for specific applications exist to compute these parameters, but they run only on CPU. We believe that an implementation of these algorithms on GPU will most likely improve the performance and dramatically decrease the time required for the study. Also, there is no national or international effort in the implementation of these algorithms on GPU.
A first approach is to use the simple case possible in the implementation of the algorithms on GPU: the obtained data (simulated or experimental) is transferred into RAM (Random Access Memory), then copied into device RAM (DRAM), after this the CUDA kernel is launched to process it. After all data is processed, the final stage is to retrieve the results back in RAM.
This naive implementation may not bring a great increase in the performance of GPU algorithms and some optimization techniques will be considered.
The most important performance challenges for CUDA developers is the optimization of device memory accesses. One may consider three ways to optimize access to DRAM memory spaces: global, shared and page-locked memory. Different memory spaces have different bandwidths and latencies and this has impact on the design, development, optimization of the algorithms and their performance on GPU.
Present project proposes an implementations of GPUs on Grid in order to enforce the two computation architecture for researches and enterprises.
In order to have a fair evaluation of the obtained results, tests and benchmarking must be done.
The methods and algorithms will be tested against data obtained from:
- UrQMD (Ultra relativistic Quantum Molecular Dynamics)simulations for nuclear interactions in stars. In order to measure the performance on GPU we measure the execution time of the algorithms compared for different nuclear collisions (gold-gold and proton-proton) on different architectures: CPU vs GPU.
- Satelllite Images ISS (CO)
In the frame of this project, we propose an improvement of the modality to adapt SITS data to the input data format of the data mining pattern extraction algorithm. This improvement makes more efficient this preprocessing stage of our approach. In order to speed up this stage, we propose a parallel processing by dividing the satellite input images in several mini-images. We will use the parallel processing framework PROOF of the ROOT system developed for High Energy Physics applications and which is able to handle and analyze large amounts of data. Furthermore we intend to modify the preprocessing stage for changing the former input format of the data mining algorithm (an ASCII file that implied a time and memory consuming processing) by storing the sequence database of SITS directly in the memory.
Tests for the validity of the developed algorithms will include histograms and graphics of interested parameters.