Initialising ...

Initialising ...

Initialising ...

Initialising ...

Initialising ...

Initialising ...

Initialising ...

Kushida, Noriyuki

PLOS ONE (Internet), 10(3), p.e0122331_1 - e0122331_16, 2015/03

Times Cited Count：7 Percentile：45.03(Multidisciplinary Sciences)The present paper introduces a condition number estimation method for preconditioned matrices. The newly developed method provides reasonable results, while the conventional method which is based on the Lanczos connection gives meaningless results. The Lanczos connection based method provides the condition numbers of coefficient matrices of systems of linear equations with information obtained through the preconditioned conjugate gradient method. Estimating the condition number of preconditioned matrices is sometimes important when describing the effectiveness of new preconditionerers or selecting adequate preconditioners. Operating a preconditioner on a coefficient matrix is the simplest method of estimation. However, this is not possible for large-scale computing, especially if computation is performed on distributed memory parallel computers. This is because, the preconditioned matrices become dense, even if the original matrices are sparse. Although the Lanczos connection method can be used to calculate the condition number of preconditioned matrices, it is not considered to be applicable to large-scale problems because of its weakness with respect to numerical errors. Therefore, we have developed a robust and parallelizable method based on Hager's method. The feasibility studies are curried out for the diagonal scaling preconditioner and the SSOR preconditioner with a diagonal matrix, a tri-daigonal matrix and Pei's matrix. As a result, the Lanczos connection method contains around 10% error in the results even with a simple problem. On the other hand, the new method contains negligible errors. In addition, the newly developed method returns reasonable solutions when the Lanczos connection method fails with Pei's matrix, and matrices generated with the finite element method.

Kushida, Noriyuki; Takemiya, Hiroshi

World Academy of Science, Engineering and Technology, 64, p.319 - 327, 2012/04

We developed a high-speed eigenvalue solver and a matrix generation code that are essentials of a plasma stability analysis system for nuclear fusion reactors on a Cell cluster system. In order to achieve sustainable operation of the reactors, we must evaluate the state of plasma within the characteristic confinement time. Therefore, we introduced a Cell that has high computational power and high performance/cost. Furthermore, we developed a novel eigenvalue solver, which consumes most of the plasma evaluation time. The eigensolver is based on the CG method and was designed by considering hierarchical parallelism of Cell. Moreover, we developed a new CG acceleration method, called locally complete LU that has both better acceleration and performance than the formers. Finally, we succeeded in obtaining our target performance: we were able to create and solve a block tri-diagonal Hermitian containing 1024 diagonal blocks, where the size of each block was 128128, within two sec.

Kushida, Noriyuki

Proceedings of 20th Euromicro International Conference on Parallel, Distributed and Network-Based Computing (PDP 2012), p.7 - 8, 2012/02

Infrasound wave propagation simulation and radioactive transfer simulation were accelerated with GPGPU and multicore processors. These simulation codes supports CTBTO mission that detects nuclear test. Since these applications have been carried out on an isolated workstation, high-performance computing units, like GPGPU and multicore processors, are helpful for more accurate simulation. As a result, we achieve 18.3 speedup for infrasound that enables us to run more reliable method than the other simpler method one the same calculation speed, and we achieve the Northern Hemisphere radioactive transfer simulation that have been virtually impossible so far.

Tatekawa, Takayuki; Teshima, Naoya*; Kushida, Noriyuki; Miyamura, Hiroko; Kim, G.; Takemiya, Hiroshi

High Performance Computing on Vector Systems 2011, p.107 - 117, 2012/01

no abstracts in English

Nishida, Akemi; Araya, Fumimasa; Kushida, Noriyuki; Kondo, Makoto; Sakai, Michiya*; Shiogama, Yuzo*

Progress in Nuclear Science and Technology (Internet), 2, p.576 - 581, 2011/10

Design evaluation of nuclear facilities would be facilitated by a numerical evaluation system that can evaluate both global and local behaviors under severe seismic loading. A critical part of such a system is the numerical model describing the dynamic physical interactions among component connections, called the elastic-plastic connection model. Here we propose such a model and use it to simulate dynamic interactions using real earthquake and plant data from the High Temperature Engineering Test Reactor (HTTR) of the JAEA. We focus on joints connecting the component supports and the building walls, which generally involve fixed/pinned boundary conditions. The results confirmed a reduction in the vibration response and a change in the natural frequencies of individual components under large virtual earthquake loading, which are considered to have resulted from dynamic interactions between the joints connecting the component supports and the building walls.

Kushida, Noriyuki

Progress in Nuclear Science and Technology (Internet), 2, p.663 - 669, 2011/10

Many nuclear applications require more computational power than the current computers can provide. Furthermore, some of them require dedicated machines, because they must run constantly or no delay is allowed. To satisfy these requirements, we introduce accelerators which can provide higher computational power with lower prices than the current processors. But the feasibility of accelerators had not well investigated. Thus, we applied the Cell and GPGPU to plasma stability monitoring and infrasound propagation analysis, respectively. In the plasma monitoring, we could achieve sufficient performance. On the basis of these results, we showed the potential of plasma monitoring by using our Cell cluster system. In infrasound propagation, we accelerated two-dimensional parabolic equation (PE) method by using GPGPU. By applying several optimization, we obtained 18.3 speedup on GPU from CPU. Our achieved computing speed could be comparable to faster but more inaccurate method.

Kushida, Noriyuki

Proceedings of 19th Euromicro International Conference on Parallel, Distributed and Network-Based Computing (PDP 2011), p.401 - 408, 2011/02

I introduced a new implementation of the finite element method (FEM) that is suitable for the cell processors. Since the cell processors have a far greater performance and low byte per flop (B/F) rate than traditional scalar processors, I reduced the amount of memory transfer and employed memory access times hiding technique. The amount of memory transfer was reduced by accepting additional floating-point operations by not storing data if it was required repeatedly. In this study, such memory access reduction was applied to conjugate gradient method (CG). In order to achieve memory access reduction in CG, element-wise computation was employed for avoiding global coefficient matrices that causes frequent memory access. Moreover, all data transfer times were concealed behind the calculation time. As a result, my new implementation performed 10 times better than a traditional implementation that run on a PPU.

Yamada, Tomonori; Kushida, Noriyuki; Araya, Fumimasa; Nishida, Akemi; Nakajima, Norihiro

Key Engineering Materials, 452-453, p.701 - 704, 2011/01

The paper describes a component-wise meshing approach and bonding strategy on the interface of components. In order to assemble component-wise meshes, the penalty method is introduced not only to constrain the displacements, but also to introduce classical spring connection on the joint interface. The convergence performance of an iterative solver with penalty method is investigated and the detailed component-wise distributed computation scheme is described with numerical examples.

Kushida, Noriyuki; Fujibayashi, Kenichi; Takemiya, Hiroshi

Procedia Computer Science, 4, p.898 - 907, 2011/00

Times Cited Count：0 Percentile：0.00(Computer Science, Theory & Methods)A high speed eigenvalue solver that is an essential part of a plasma stability analysis system for fusion reactors on a Cell cluster system is described. For the purpose of continuous operation of fusion reactors, we must evaluate the state of plasma within the characteristic confinement time of the plasma density and temperature in fusion reactors. In order to resolve the problem, we introduced a Cell cluster system and developed a novel eigenvalue solver, which usually consumes most of the plasma evaluation time, to achieve high performance of the system. As a result, we succeeded in obtaining our target performance: we were able to solve a block tri-diagonal Hermitian matrix containing 1024 diagonal blocks, where the size of each block was 128128, within a second. Therefore, we have found a suitable candidate for achieving a satisfactory monitoring system.

Nishida, Akemi; Araya, Fumimasa; Kushida, Noriyuki; Kondo, Makoto; Sakai, Michiya*; Shiogama, Yuzo*

Proceedings of Joint International Conference of 7th Supercomputing in Nuclear Application and 3rd Monte Carlo (SNA + MC 2010) (USB Flash Drive), 6 Pages, 2010/10

The objective of this research is to contribute to the seismic design evaluation of nuclear facilities through the construction of a numerical evaluation system which is able to evaluate both global and local behaviors of facilities under severe seismic events. As one of the technology components to realize this objective, we are developing a physical model describing the dynamic interaction characteristics of component connections, called as the elastic-plastic connection model. We focused on the joints of the support structures of the component and the building in nuclear plants which generally designed as fixed/pinned boundaries, and tried to consider their dynamic interaction effects. In this paper, we show the proposal of the elastic-plastic connection model and the application of the model to a numerical simulation using a real plant data. The precision of the model was optimized by adjusting its parameters using the data obtained in the experiment.

Suzuki, Yoshio; Kushida, Noriyuki; Tatekawa, Takayuki; Teshima, Naoya; Caniou, Y.*; Guivarch, R.*; Dayde, M.*; Ramet, P.*

Proceedings of Joint International Conference of 7th Supercomputing in Nuclear Application and 3rd Monte Carlo (SNA + MC 2010) (USB Flash Drive), 6 Pages, 2010/10

The "Research and Development of International Matrix-Solver Prediction System (REDIMPS)" project, which is founded by the Strategic Japanese-French Cooperative Program on "Information and Communications Technology including Computer Science" with CNRS and JST, aims at improving the "Test for Large System of Equations (TLSE)" sparse linear algebra expert system by establishing an international grid computing environment between Japan and France. Here, we have established the interoperable environment between French and Japanese grid middleware (DIET and AEGIS), and have confirmed that TLSE can rely on this French and Japanese interoperable environment for researchers to select a matrix-solver suitable to their each application program. By this study, we proposed to the French and Japanese researchers the environment in which they can obtain useful information for the improvement of their application program.

Kushida, Noriyuki

Proceedings of Joint International Conference of 7th Supercomputing in Nuclear Application and 3rd Monte Carlo (SNA + MC 2010) (USB Flash Drive), 6 Pages, 2010/10

Several applications in nuclear energy field require dedicated supercomputing environment (DSE). One of the greatest common aspects of DES applications is that they have to complete tasks within deadlines. High-end supercomputers are not suitable for DSE applications, because it is publically shared facilities and allotted time for a person is quite limited. Moreover, owning a high end supercomputer is not realistic because of the high price of supercomputers. On the other hand, many core processors have good cost performance and therefore they are suitable for DSE. In the present paper, the author describes the feasibility of many-core processors for such purpose by using two examples: (1) Fusion reactor monitoring and (2) Infrasound propagation analysis.

Kushida, Noriyuki; Takemiya, Hiroshi

Nihon Keisan Kogakkai Rombunshu (Internet), 2010, 10 Pages, 2010/03

In this study, we introduce a novel implementation of FEM targeting the Cell which has a software controlled local (SCLM) memory. In recent days, developers of scientific or engineering numerical simulation code suffer from the big latency of data transfer between a processor and main memory, and power consumption. To overcome these problems, several researchers have proposed new computer architecture, which combines SCLM and SIMD processing unit. FEM is the one of the most famous numerical simulation method especially for the engineering. Such new computer architecture is, however, less effective for FEM than other applications. The Cell is well known as the accelerator of the Roadrunner which is the fastest computer in the Top500 supercomputer list, and is only existing processor which has software controlled local memory. In this study we developed new FEM implementation method which provides faster computation than conventional method by reducing main memory access.

Kushida, Noriyuki; Takemiya, Hiroshi; Tokuda, Shinji*

Proceedings of 18th Euromicro International Conference on Parallel, Distributed and Network-Based Computing (PDP 2010), p.482 - 488, 2010/02

In this study, we developed a high speed eigenvalue solver that is the necessity of plasma stability analysis system for International Thermo-nuclear Experimental Reactor (ITER) on Cell cluster system. According to our estimation, the most time consuming part of analysis system is eigensolver. However, current supercomputer is not applicable for such instantaneous calculation, because the overhead of network communication becomes dominant. Therefore, we employ Cell cluster system, whose processor has higher performance than current supercomputer, because we can obtain sufficient processing power with small number of processors. Furthermore, we developed novel eigenvalue solver with the consideration of hierarchical architecture of Cell cluster Finally, we succeeded to solve the block tridiagonal Hermitian matrix, which had 1024 diagonal blocks and the size of each block was 128 128 within a second.

Kushida, Noriyuki; Takemiya, Hiroshi; Tokuda, Shinji

Proceedings of International Conference for High Performance Computing, Networking, Storage and Analysis (SC '09) (USB Flash Drive), 2 Pages, 2009/11

In this study, we developed a high speed eigenvalue solver that is the necessity of plasma stability analysis system for International Thermo-nuclear Experimental Reactor (ITER) on Cell cluster system. Our stability analysis system is developed in order to prevent the damage of the ITER from the plasma disruption. The stability analysis system requires solving the eigenvalue of matrices whose dimension is hundred thousand in a few seconds. However, current massively parallel processor (MPP) type supercomputer is not applicable for such short-term calculation, because of the network overhead become dominant. Therefore, we employ Cell cluster system, because we can obtain sufficient processing power with small number of processors. Finally, we succeeded to solve the block tri-diagonal Hermitian matrix, which had 1024 diagonal blocks and each block size was 128128, within one second.

Nishida, Akemi; Araya, Fumimasa; Yamada, Tomonori; Kushida, Noriyuki; Takemiya, Hiroshi; Nakajima, Norihiro

Anzen Kenkyu Fuoramu 2009 Shiryoshu, p.25 - 29, 2009/02

The development of a vibration simulator has been planned for used in a full-scale nuclear power station; this simulator will implement the latest computational technologies and will allow comparisons to be made among the conventional spring-mass models and the actually observed data obtained from experiment and practice. This simulator will be used for evaluating the conservativeness of conventional models and the fragility of components through a seismic PSA (probabilistic safety assessment). This simulator will be used for analyzing the seismic responses of the entire nuclear facility by modeling each component independently. The numerical capability of the simulator has already been confirmed; presently, the accuracy of the dynamic response results is being investigated by means of a comparison with the recorded data of an actual plant system.

Kino, Chiaki; Suzuki, Yoshio; Kushida, Noriyuki; Nishida, Akemi; Hayashi, Sachiko; Nakajima, Norihiro

High Performance Computing on Vector Systems 2008, p.89 - 97, 2009/00

In a field of research for nuclear power station, large and complex data analysis is an important issue. To overcome this issue, we have been developing Cognitive methodology based Data Analysis System (CDAS) in order to support researchers to analyze large and complex data. In the present study, we cleared up a structure of data analysis consisting of analysis target, evaluation index and judgment criteria. Additionally, we proposed some computational technologies to actualize CDAS. We have applied the system to the virtual plant vibration simulator and confirmed the implementability of this system.

Suzuki, Yoshio; Kushida, Noriyuki; Teshima, Naoya; Nakajima, Kohei; Nishida, Akemi; Nakajima, Norihiro

High Performance Computing on Vector Systems 2008, p.65 - 77, 2009/00

Center for computational science and e-systems, Japan Atomic Energy Agency (CCSE/JAEA) has been carrying out R&Ds of grid computing technology since FY1995 aiming at establishing an infrastructure for the computational science research. One of them is R&D of IT-Based Laboratory (ITBL) infrastructure. The ITBL project is a national project placed as one of the e-Japan Priority Policy Program to realize the e-Japan Strategy. By succeeding the technology of ITBL infrastructure, CCSE is carrying out R&D of Atomic Energy Grid Infrastructure (AEGIS) to construct an intelligent infrastructure for the atomic energy research. CCSE has been carrying out international cooperation researches in various fields to advance computer science and to expand AEGIS grid computing environment. To achieve this, we have developed the system to construct interoperable environments between AEGIS and other grid middleware.

Suzuki, Yoshio; Nishida, Akemi; Araya, Fumimasa; Kushida, Noriyuki; Akutsu, Taku; Teshima, Naoya; Nakajima, Kohei; Kondo, Makoto; Hayashi, Sachiko; Aoyagi, Tetsuo; et al.

Journal of Power and Energy Systems (Internet), 3(1), p.60 - 71, 2009/00

Center for computational science and e-systems of Japan Atomic Energy Agency is carrying out R&D in the area of extra large-scale simulation technologies for solving nuclear plant structures in its entirety. Specifically, we focus on establishing a virtual plant vibration simulator on inter-connected supercomputers intended for seismic response analysis of a whole nuclear plant. The simulation of a whole plant is a very difficult task because an extremely large dataset must be processed. To overcome this difficulty, we have proposed and implemented a necessary simulation framework and computing platform. The computing platform enables an extra large-scale whole nuclear plant simulation to be carried out on a grid computing platform ITBL-IS and AEGIS. The simulation framework based on the computing platform has been applied to a linear elastic analysis of the reactor pressure vessel and cooling systems of the nuclear research facility, HTTR.

Nakajima, Norihiro; Araya, Fumimasa; Nishida, Akemi; Suzuki, Yoshio; Ida, Masato; Yamada, Tomonori; Kushida, Noriyuki; Kim, G.; Kino, Chiaki; Takemiya, Hiroshi

Proceedings of International Symposium on Structures under Earthquake, Impact, and Blast Loading 2008, p.119 - 123, 2008/10

Japan is so said an energy consumption country of the fourth place world, but the energy resources such as petroleum, the natural gas are poor and depend on import for the most, and stable supply becomes a big problem. For the greenhouse gas restraint, the promotion of the energy saving is featured. A nuclear power plant for commerce in Japan has been started in 1966. The supply occupies about 30 percent of the now Japanese electricity generating. Due to the nature of Japan, earthquake proof is an important subject for social infrastructure operation. To encourage its proofing, many approaches have been applied into many infrastructures, not only computational approach. A computational science approach for earthquake proof is suggested with FIESTA(Finite Element Structural analysis for Assembly), a large scale simulation. A methodology is discussed from the point of view of impact and blast loadings. Examples of loadings in the nuclear engineering are introduced.