Project

Back to overview

UP-IPAZIA: “UPgrade and full deployment of the Empa/Eawag computational cluster IPAZIA: towards an interdisciplinary on-site resource for computational sciences”

English title UP-IPAZIA: “UPgrade and full deployment of the Empa/Eawag computational cluster IPAZIA: towards an interdisciplinary on-site resource for computational sciences”
Applicant Bucher Christoph
Number 128754
Funding scheme R'EQUIP
Research institution Eidg. Materialprüfungs- und Forschungsanstalt (EMPA)
Institution of higher education Swiss Federal Laboratories for Materials Science and Technology - EMPA
Main discipline Material Sciences
Start/End 01.04.2010 - 31.03.2011
Approved amount 200'000.00
Show all

Keywords (8)

computational sciences; high performance computing; environment; nanotechnology; Computational science; materials science; environmental science; nanoscience

Lay Summary (English)

Lead
Lay summary
The Swiss federal laboratories for materials testing and research (Empa) have recently steered their main scope from a pure service company (as it was in the past) to a modern institution blending precious well-established skills with novel applied research directions, including nanotechnology, functional materials, effective energy storage solutions, interface and surface science, biocompatible materials. In parallel with the application and development of always more precise experimental tools, the role of computational modeling in all fields of research has grown enormously in the last years, and Empa's direction has acknowledged these needs by the creation of a theory and computational science group, with the aim of coordinating the set up of a local parallel computer cluster.The blending of serial, mildly parallel and fast interconnected parallel jobs on the same machine, with different needs corresponding to different modeling techniques has proven to be efficiently feasible within a frame of the size of Empa/Eawag.After this successful testing and early production phase, it is our intent to enhance the role of our computational cluster as an interdepartmental structure for the promotion of computational science at Empa and Eawag.With this proposal the applicants thus ask for support to expand and update the present cluster according to the increased local request. With this investment, we plan to reach a final size of about 130 nodes, half of which connected through a fast IB infrastructure.The need for internal resources, the documented support of the direction and of the involved laboratories makes the expansion plan to the final size and the continuous update of the existing hardware reasonable and realistic, and will allow to:1. Provide efficient computing power to research groups where modeling is not the central focus. Such groups would otherwise use local desktop resources with much less efficient results; 2. Enhance the productivity of groups which are already involved in large user projects in CSCS, relieving the Manno supercomputing center from the non optimal middle-size calculations and making free place for massively parallel computing; 3. Cross fertilize software tools, ideas, programming resources, "tricks of the trade" among different disciplines including atomistic simulation, density functional theory, multiphysics modeling, real world problems (as water and air pollution, train noise etc.), finite element modeling etc.4. Establish an ideal playground for future local interdisciplinary collaborations within the common mainstreams of Empa and Eawag.
Direct link to Lay Summary Last update: 21.02.2013

Responsible applicant and co-applicants

Associated projects

Number Title Start Funding scheme
153661 Predicting properties of realistic zigzag like graphene nanoribbons, a DFT challenge 01.07.2014 Project funding (Div. I-III)
136287 Understanding nanofriction and dissipation across phase transitions 01.09.2011 Sinergia
135435 Organic nanowires with unusual transport properties: a theoretical study 01.07.2011 Project funding (Div. I-III)
122479 Global agricultural green and blue water consumptive uses and virtual water flows in the context of water scarcity and climate change 01.02.2009 Interdisciplinary projects
132375 Molecular self-assemblies, their metallic substrates, and quasicrystal surfaces: a computational approach 01.10.2010 Project funding (Div. I-III)
116073 Atomistic simulations of surface-supported molecular nanostructures 01.10.2007 Project funding (Div. I-III)
120005 Understanding the adsorption behaviour of large organic molecules: A step towards the realisation of molecular electronics 01.05.2008 Project funding (Div. I-III)
120130 Understanding and controlling supramolecular network formation (CONE) 01.05.2009 Project funding (Div. I-III)
116130 Computational study of the surfaces of quasicrystals 01.02.2008 Project funding (Div. I-III)
135270 Relationships between 3D topology and oxygen reduction kinetics in mixed conducting LSC cathodes for SOFC 01.12.2011 Project funding (Div. I-III)
135492 Vibration-induced unjamming of sheared granular media: investigation by 3D Discrete Element Method modeling and simulation 01.05.2011 Project funding (Div. I-III)

Abstract

The Swiss federal laboratories for materials testing and research (Empa) have recently steered their main scope from a pure service company (as it was in the past) to a modern institution blending precious well-established skills with novel applied research directions, including nanotechnology, functional materials, effective energy storage solutions, interface and surface science, biocompatible materials. In parallel with the application and development of always more precise experimental tools, the role of computational modeling in all fields of research has grown enormously in the last years, and Empa’s direction has acknowledged these needs by the creation of a theory and computational science group, with the aim of coordinating the set up of a local parallel computer cluster. The latter should accommodate the needs of Empa research groups and of the neighboring federal laboratories for water research (Eawag). The central coordination of the purchase, administration, scientific software, machine usage and infrastructure handling would then allow an optimal use of the resources. The practical realization of the cluster has started about 2 years ago. The funding system for the cluster has been kept flexible and efficient. The direction of Empa itself has provided an initial investment, and approved the set up of the necessary infrastructure including water cooled racks. Eawag contributed with an effort corresponding to the 15% of the total costs. Some laboratories paid for the purchase of nodes, others have submitted projects and are using publicly available nodes. In the last two years, the cluster has grown to the size of 46 nodes, corresponding to 248 CPU cores and 768 GB RAM, only partly (16 nodes) connected with Infiniband. Ten to fifteen different laboratories at Empa and Eawag use the machine, and, as shown below, new laboratories are stepping in. In agreement with the Swiss federal guidelines for supercomputing, our high performance computing (HPC) cluster should not become a super-computing center; rather, it should provide resources for middle-size projects, complementary to the massively parallel computations run at CSCS, and for small-sized projects, that profit dramatically from the porting from desktop-like architectures to a dedicated high performance cluster. The blending of serial, mildly parallel and fast interconnected parallel jobs on the same machine, with different needs corresponding to different modeling techniques has proven to be efficiently feasible within a frame of the size of Empa/Eawag. After this successful testing and early production phase, it is our intent to enhance the role of our computational cluster as an interdepartmental structure for the promotion of computational science at Empa and Eawag. With this proposal the applicants thus ask for support to expand and update the present cluster according to the increased local request. With this investment, we plan to reach a final size of about 130 nodes, half of which connected through a fast infiniband infrastructure. The need for internal resources, the documented support of the direction and of the involved laboratories makes the expansion plan to the final size and the continuous update of the existing hardware reasonable and realistic, and will allow to: 1. Provide efficient computing power to research groups where modeling is not the central focus. Such groups would otherwise use local desktop resources with much less efficient results; 2. Enhance the productivity of groups which are already involved in large user projects in CSCS, relieving the Manno supercomputing center from the non optimal middle-size calculations and making free place for massively parallel computing; 3. Cross fertilize software tools, ideas, programming resources, "tricks of the trade" among different disciplines including atomistic simulation, density functional theory, multiphysics modeling, real world problems (as water and air pollution, train noise etc.), finite element modeling etc. 4. Establish an ideal playground for future local interdisciplinary collaborations within the common mainstreams of Empa and Eawag.Scientific projects from different disciplines (both in fundamental and applied science) can profit from the presence of local specialized and extended computational resources. Such benefits will appear evident from the following project list, which will serve also to determine the amount of resources requested. The use of a common local computational resource by different projects not only allows to optimize the computational resources, but realizes a common ground for discussion, giving the possibility to find common solutions to diverse problems. The organization of technical and topical meetings among the cluster users (“Ipazia circles”) has already proven to be an effective tool in this respect. The users can discuss about programming techniques, state-of-the-art simulation methods, visualization tools and administrative aspects like database, storage and backup tools. At variance with respect to a large supercomputing center or to facilities dedicated to large number of users that almost never meet each other, the community of users of Ipazia coincides in principle with the relatively small computational science community at Empa/Eawag. Therefore, the resources can directly be “fine tuned” at a local level following the requirements of the user groups. The main projects (coming from various Empa and Eawag research groups) that will profit from the proposed installation are:A: Railway noise calculation with sonRAIL. B: Effects of large-scale implementation of water harvesting on food production and the hydrological regime in Sub-Saharan Africa.. C: Global agricultural green and blue water consumptive uses and virtual water flows in the context of water scarcity and climate change. D: Investigation of the relationship between the performance of solid oxide fuel cells (SOFCs) and the electrodes microstructures by means of finite element modelling and 3D-microstructure analysis. E: Computational study of the surfaces of quasicrystals F: Atomistic simulations of surface supported molecular nanostructures G: Imaging disordered media using Time Reversal Processing. H: Discrete Element Method modeling of sheared granular layers. I: Atmospheric transport and chemistry modeling. J: Multiscale modeling of materials for energy applications. S1: Aircraft noise model calculations based on a complete set of radar data S2: Modelling reactive flows in urban water engineering. S3: Large organic molecules and supramolecular networks on metal surfaces: Understanding experiments with atomistic simulations S4: Molecular modeling of squaraine dyes.S5: Thermodynamic properties of complex hydrides
-