joint usage/research center for interdisciplinary large ... · 2018 call for proposal of joint...

2017/12/06

1

Joint Usage/Research Center for Interdisciplinary Large-scale Information Infrastructures

2018 Call for Proposal of Joint Research Projects

The Joint Usage/Research Center for Interdisciplinary Large-scale Information Infrastructures

(JHPCN) calls for joint research projects for fiscal year 2018. JHPCN is a "Network-Type" Joint

Usage/Research Center, certified by the Minister of Education, Culture, Sports, Science and

Technology, and comprises eight super-computer equipped centers affiliated with Hokkaido

University, Tohoku University, The University of Tokyo, Tokyo Institute of Technology, Nagoya

University, Kyoto University, Osaka University and Kyushu University, among which Information

Technology Center at the University of Tokyo is functioning as the core institution of JHPCN.

Each JHPCN member institution will provide its research resources for joint research projects.

Researchers of an adopted joint research project can use provided research resources within

granted amount with free of charge. In addition, expenses for publication or presentation of

research results such as overseas travel expenses may be supported. Every eligible project

should use research resources provided by JHPCN member institutions or its research team

should include researchers affiliated with JHPCN member institutions. The available research

resources are computers, storages, visualization systems and so on. Some of them can be

connected to facilities inside and outside of JHPCN member institutions via high speed

networks for the purposes of exchange, accumulation and processing of data. It is also possible

to connect research resources of JHPCN member institutions with SINET5 L2VPN.

Part of High Performance Computing Infrastructure (HPCI) system, referred to as the HPCI-

JHPCN system, are also available for joint research projects. Applications for research

projects that use the HPCI-JHPCN System are invited via the HPCI online application system

and will be selected in line with our reviewing policy (see Section 6,“Research Project

Reviews”).

The aim of JHPCN is to contribute to the advancement and permanent development of

academic and research infrastructure of Japan by implementing joint research projects that

require large scale information infrastructures and that address Grand Challenge-type

problems, thus far considered extremely difficult to solve, in the following fields: Earth

environment, energy, materials, genomic information, web data, academic information, time-

series data from the sensor networks, video data, program analysis, application of large

capacity network and other fields in information technology. Since the JHPCN member

institutions have enrolled leading researchers, acceleration of joint research projects is

anticipated through the collaboration with these researchers. These joint research projects (for

the fiscal year 2018) will be implemented from April 2018 to March 2019.

2017/12/06

2

1. Joint Research Areas

This call for joint research projects will adopt interdisciplinary research projects in the four areas:

very large-scale numerical computation, very large-scale data processing, very large capacity

network technology, and very large-scale information systems. Approximately 60† joint research

projects will be adopted.

†Total number of JHPCN member institutions used for the adopted research projects.

(1) Very large-scale numerical computation

This includes scientific and technological simulations in scientific/engineering fields such as

Earth environment, energy, and materials, as well as modeling, numerical analysis algorithms,

visualization techniques, information infrastructure, etc., to support these simulations.

(2) Very large-scale data processing

This includes processing genomic information, web data (including from Wikipedia, as well as

news sites and blogs), academic information contents, time-series data from the sensor

networks, high-level multimedia information processing needed for streaming data for video

footage, etc., program analysis, access and search, information extraction, statistical and

semantic analysis, data-mining, machine learning, etc.

(3) Very large capacity network technology

This includes control and assurance of network quality for very large-scale data sharing,

monitoring and management required for construction and operation of very large capacity

networks, assessment and maintenance of the safety of such networks, a large-scale stream

processing framework taking advantage of very large capacity networks as well as the

development of various technologies for the support of research. Research dealing with large-

scale data processing using the large capacity network is classified in this area.

(4) Very large-scale information systems

This combines each of the above-mentioned areas, entailing exascale computer architectures,

software for high-performance computing infrastructures, grids computing, virtualization

technology, cloud computing, etc.

Please pay special attention to the following points when applying.

① We will only be accepting interdisciplinary joint research proposals that will involve

cooperation among researchers from a wide range of disciplines. For example, we

presume that a research team in the “very large-scale numerical computation” area will

2017/12/06

3

consist of researchers from computer and computational sciences who work together in a

cooperative and complimentary manner. In this area, therefore, we invite joint research

projects with researchers who will be solving problems in application fields using computers

and those who will be conducting research in computer science, such as on algorithms,

modeling, and parallel processing. It is not mandatory to use the HPCI-JHPCN system or

other research resources provided by JHPCN member institutions. In other words, a joint

research project with no use of provided research resources can be acceptable.

② We particularly appreciate the following categories of joint research projects.

● Research projects in close cooperation with multiple JHPCN member institutions

Taking advantage of the “network” of JHPCN, a project in this category should use

research resources and/or involve researchers of multiple JHPCN member institutions. For

example, relevant research topics may include but not limited to large-scale and

geographically distributed information systems and multi-platform implementations of

applications using research resources provided by multiple JHPCN member institutions.

● Research projects using both large-scale data and large capacity networks

The available research resources include those that can be directly connected to a very

wide bandwidth network provided by SINET5, including L2VPN, in cooperation with the

National Institute of Informatics. Therefore, research can be conducted that depends upon

a very wide bandwidth network. A project in this category should require massive data

transfer, using the very wide bandwidth network, between research facilities located at

involved researchers’ working places and at JHPCN member institutions or between those

at JHPCN member institutions. Please refer to Attachment 2 for possible examples of

research topics in this category.

2. Types of Joint Research Projects

Under the premise of the interdisciplinary joint research project structure mentioned above, joint

research projects to be invited are as follows.

(1) General Joint Research Projects (approximately 80% of the total number of accepted

projects will be in this type)

(2) International Joint Research Projects (approximately 10% of the total number of

accepted projects will be in this type)

International joint research projects are conducted in conjunction with foreign researchers to

address challenging problems that may not be resolved or clarified with only the help of

researchers within Japan. For such research projects, there will be a certain amount of

2017/12/06

4

subsidies paid to cover travel expenses incurred for holding meetings with foreign joint

researchers during the fiscal year of commencement. For details, please contact our office

once your research project has been accepted.

(3) Industrial Joint Research Project (approximately 10% of the total number of accepted

projects will be in this category)

Industrial joint research projects are interdisciplinary projects focused on industrial

applications.

A research proposal submitted to (2) International Joint Research Project or (3) Industrial Joint

Research Project might be adopted as a (1) General Joint Research Project. Further, we invite

applications in each of these three types for “research projects that use the HPCI-JHPCN

System” and “research projects that do not use the HPCI-JHPCN System.”

If you require the assistance and cooperation of researchers and research groups that are

affiliated with JHPCN member institutions in the computer science field, please complete

Section 7 of the application form describing the specific requirements in detail. We will arrange

for as much assistance and cooperation as possible.

During the application, please designate the university (JHPCN member institution) with which

you seek collaboration. You may also name several institutions. If this is difficult for you to

designate, it is possible for us to decide corresponding research center(s) on your behalf after

considering the research proposal. Please be sure to discuss this with our contact person (listed

on Section 10) in advance.

3. Application Requirements

Applications must be made by a Project Representative. Furthermore, the Project

Representative and the Deputy Representative(s), as well as any other joint researchers of a

project, must fulfill the following conditions.

① The Project Representative must be affiliated with an institution in Japan (University,

National Laboratory, Private Enterprise, etc), and must be in a position to be able to

obtain the approval of his/her institution (or representative head).

② At least one Deputy Representative must be a researcher in a different academic field

from that of the Project Representative. There may be two or more Deputy

Representatives, and one of them can be in charge of HPCI face-to-face identity vetting.

③ A graduate student can participate in the project as a joint researcher, but an

undergraduate cannot. A Graduate student cannot become the Representative or the

Deputy Representative of the project. If a non-resident member, defined by the Foreign

2017/12/06

5

Exchange and Foreign Trade Act, is going to use computers, a researcher affiliated with

the JHPCN member institutions equipped with them must participate as a joint

researcher.

International joint research projects must, in addition to the above-mentioned ①–③, fulfill the

following conditions (④ and ⑤).

④ At least one researcher affiliated with a research institution outside Japan must be

named as a Deputy Representative. Furthermore, an application must be made using

the English Application Form.

⑤ A researcher affiliated with the JHPCN member institutions that you designate “Desired

University for Joint Research” must participate as a joint researcher.

Industry joint research projects must, in addition to the above-mentioned ①–③, fulfill the

following conditions (⑥ and ⑦).

⑥ The Project Representative must be affiliated with a private company, excluding

universities and national laboratories.

⑦ At least one researcher affiliated with the JHPCN member institutions that you designate

“Desired University for Joint Research” must be named as a Deputy Project

Representative.

4. Joint Research Period

April 1, 2018 to March 31, 2019.

Depending on conditions for preparing computer accounts, the commencement of computer

use may be delayed.

5. Facility Use Fees

The research resources listed in Attachment 1 can be used within granted amount with free of

charge.

6. Research Project Reviews

Reviews will be conducted by the Joint Research Project Screening Committee, which

comprises faculty members affiliated with JHPCN member institutions as well as external

members, and the HPCI Project Screening Committee, which comprises industry, academic,

and government experts. Research project proposals will be reviewed in both a general and

technical sense for their scientific and technological validity, their facility/equipment

requirements and the validity of these requirements, and their potential for development. The

feasibility of resource requirements at and cooperation/collaboration with the JHPCN member

institutions that you designate “Desired University for Joint Research” will also be subject to

2017/12/06

6

review. In addition, how relevant a proposal is to its type of joint research project will be

considered.

Furthermore, for projects continuing from the previous fiscal year and projects determined to

have substantial continuity, an assessment of the previous year’s interim report and previous

usage of computer resources may be considered during the screening process.

7. Notification of Adoption

We expect to notify the review results by mid-March 2018.

8. Application Process

1. Please note that the following application procedures differ for “Research projects that use

the HPCI-JHPCN System” and “Research projects that do not use the HPCI-JHPCN System”

(The HPCI-JHPCN system is described in Attachment 1 (1)).

2. In particular, for “Research projects that use the HPCI-JHPCN System,” the Project

Representative (and the Deputy Representative who will submit the proposal or who will be in

charge of HPCI face-to-face identity vetting on behalf of the Project Representative) and all

joint researchers that will use the HPCI-JHPCN system must have obtained their HPCI-ID

prior to the application.

3. For international joint research projects, an English application form must be completed.

(1) Application Procedure

After completing the Research Project Proposal Application Form obtained from the

JHPCN website (https://jhpcn-kyoten.itc.u-tokyo.ac.jp/en/) and completing the online

submission of the electronic application, please print and press seals onto the application

form and mail it to the Information Technology Center, The University of Tokyo (address

provided on Section 10). For details on the application process, please consult the JHPCN

website and the “HPCI Quick Start Guide.”

The summary of the application process is shown below.

I: For “Research Projects that use the HPCI-JHPCN System”

① Download the Research Project Proposal Application Form (MS Word format) from

the JHPCN website and complete it. In parallel, the Project Representative (and the

Deputy Representative who will submit the proposal or who will be in charge of HPCI

face-to-face identity vetting on behalf of the Project Representative) and all joint

researchers who will use the HPCI-JHPCN System must obtain their HPCI-IDs

https://jhpcn-kyoten.itc.u-tokyo.ac.jp/en/

2017/12/06

7

unless already have them. For online entries to be made as per the requirements of

step ②, it is necessary to register the HPCI-IDs of the Project Representative, the

Deputy Representative acting on behalf of Project Representative, and all joint

researchers who will use the HPCI-JHPCN System. Joint researchers who are not

registered at this stage will not be issued with computer user accounts (HPCI

account, local account) and thus, will not be able to use the HPCI-JHPCN system.

② Access the “Research Projects that use the HPCI-JHPCN System” on the Project

Application page of the above-mentioned JHPCN website. You will be transferred to

the HPCI Application Assistance System where after filling out the necessary items,

you should upload a PDF file of the Research Project Proposal Application Form

(without seal) created in ①.

※ For “Research Projects that use the HPCI-JHPCN System” the project

application system on the JHPCN website will not be used.

③ Print and press seals onto the Research Project Proposal Application Form created

in ①, and mail it to the address provided on Section 10. Contact Information

◆ If your project is adopted, please follow the “Procedures after the awarding of the

proposal” stipulated by the HPCI (https://www.hpci-office.jp/pages/e_taimen.html). In

particular, face-to-face identity vetting must be provided responsibly by the Project

Representative or a Deputy Representative, who may have to bring copies of

photographic identification of all joint researchers who will use the HPCI-JHPCN system.

(Note that the contact persons such as secretaries are ineligible for a face-to-face identity

vetting).

II: For “Research Projects that do not use the HPCI-JHPCN System”

① Download the Research Project Proposal Application Form (MS Word format) from

the JHPCN website and complete it.

② Access the “Research Projects that do not use the HPCI-JHPCN System” on the

project application page of the above-mentioned JHPCN website. After filling out

the necessary items, you should upload a PDF file of the Research Project

Proposal Application Form (without seal) created in ①. An e-mail “Notification of

Receipt” will be sent to the e-mail address that you registered on the Research

Project Application page.

※ Applications for “Research Projects that do not use the HPCI-JHPCN System” do

not require the HPCI Application Assistance System. Therefore, it is not

necessary to obtain HPCI-IDs.

https://www.hpci-office.jp/pages/e_taimen.html

2017/12/06

8

③ Print and press seals onto the Research Project Proposal Application Form

created in ①, and mail it to the address provided on Section 10. Contact

Information.

(2) Application Deadline

Online Registration (submission of PDF application form): January 9, 2018 (Tue.)

17:00 [Mandatory]

Submission of printed application form: January 15, 2018(Mon.) 17:00 [Mandatory]

The above deadlines must be followed for submission of the Research Project Proposal

Application Forms. However, if submission of printed documents created in (1)Ⅰ③ and (1)Ⅱ

③ above is delayed, please contact the JHPCN office (details on Section 10) in advance.

(3) Points to remember when creating the Research Project Proposal Application Form

① Applicants must comply with rules and regulations of the JHPCN member institutions

whose computer resources will be used in the joint research project.

② Research resources must be only used for the purpose of the adopted research project.

③ The proposal must be for a project with peaceful purposes.

④ The proposal should consider human rights and profit protection.

⑤ The proposal must comply with the Ministry of Education, Culture, Sports, Science and

Technology’s “Approach to Bioethics and Safety”.

⑥ The proposal must comply with the Ministry of Health, Labour and Welfare’s “Guidelines

on Medical and Health research.”

⑦ The proposal must comply with the Ministry of Economy, Trade and Industry’s

“Concerning Security Trade Management.”

⑧ Applicants must complete a course in Responsible Conduct of Research (RCR).

Programs on RCR, including lectures, e-learning, and training workshops, are conducted

at each of the affiliated organizations (including CITI Japan e-learning). If your affiliated

organization does not provide any RCR programs, please contact our office. Researchers

who are eligible to apply for the Japanese Grant-in-Aid for Scientific Research awarded by

the Ministry of Education, Culture, Sports, Science and Technology and the Japan Society

for the Promotion of Science will be considered as eligible if they note their “Kakenhi”

Researcher Code in the application form. In addition, researchers who were recently

awarded research grants that required RCR training will be also considered as eligible if

they present evidence.

⑨ After being accepted, applicants must comply with the deadlines for submitting a

research introduction poster and various reports.

2017/12/06

9

9. Other Important Notices

(1) Submission of a written oath

Research groups whose research projects have been adopted will be expected to submit a

written oath pledging adherence to the contents of the above-mentioned “(3) Points to

remember when creating the Research Project Proposal Application Form” of “8. Application

Process.” The specific process of submission will be provided post-adoption, but a sample of

it is provided on the website for those applicants who wish to familiarize themselves with the

requirements of the oath.

(2) Regulations for use of facilities

While using the facilities, you are expected to follow the regulations for use pertaining to

research resources as stipulated by the JHPCN member institutions with which you will

work.

(3) Types of reports

The submission of research introduction posters, interim reports, and final reports are

expected. Each report must be submitted by the designated deadline. Research introduction

posters as well as the final reports, in particular, will be expected to be suitable for

disclosure (see the JHPCN website for details). For International Joint Research Projects,

these reports should be submitted in English in principle. Furthermore, you will be asked to

introduce the research through poster presentation at the JHPCN symposium (where the

results of research projects from the previous year will be presented orally) that is expected

to be held during the summer.

(4) Disclaimer

Each JHPCN member institution assumes no responsibility at all for any inconveniences that

affect applicants as a consequence of joint research projects.

(5) Handling of intellectual property rights

In principle, any intellectual property that arises as a result of a research project will belong

to each of the research groups involved. However, it is presumed that any recognition of

inventors will be in accordance with each institution’s policy concerning intellectual property

rights. Please contact each JHPCN member institution for details and handling of other

exceptional matters.

(6) RCR training

If a participant in an accepted project (excluding students) cannot be confirmed to have

completed a program pertaining to RCR or equivalent (for example, eligibility for the

Japanese Grant-in-Aid for Scientific Research that is awarded by the Ministry of Education,

Culture, Sports, Science and Technology, or the Japan Society for the Promotion of

2017/12/06

10

Science), this individual’s use of computers may be suspended until the completion of such

a program can be confirmed.

(7) Acknowledgement

Upon publication of the results of an accepted project, the author(s) should indicate in

Acknowledgements that the project was supported by JHPCN (see the JHPCN website for

an example sentence)..

(8) Other

① Personal information provided by this proposal shall only be used for screening research

projects and providing system access.

② After the adoption of a research project, however, the project name and the

name/affiliation of the Project Representative will be disclosed

③ After the adoption of the research project, changes cannot be made to the JHPCN

member institutions to work with or the computers to be used.

④ If you wish to discuss your application, please contact us on the e-mail address listed on

Section 10 (please note in advance that we are not able to respond to telephone-based

inquiries).

10. Contact information (for inquiries about application, etc.)

● For inquiries about application

Joint Usage/Research Center for Interdisciplinary Large-scale Information Infrastructures Office

E-mail address: jhpcn.adm＠gs.mail.u-tokyo.ac.jp

● For available resources, how to use resources, details of eligibility, faculty members who can

participate in joint research projects, and the handling of intellectual property of each institution,

please feel free to directly contact the following.

JHPCN member institutions e-mail address

Information Initiative Center, Hokkaido University [email protected]

Cyberscience Center, Tohoku University [email protected]

Information Technology Center, The University of Tokyo [email protected]

Global Scientific Information and Computing Center,

Tokyo Institute of Technology [email protected]

Information Technology Center, Nagoya University [email protected]

Academic Center for Computing and Media Studies,

Kyoto University [email protected]

mailto:[email protected]







2017/12/06

11

Cybermedia Center, Osaka University [email protected]

Research Institute for Information Technology,

Kyushu University [email protected]

◆ Mailing address for the Research Project Proposal Application Form

2-11-16 Yayoi, Bunkyo-ku, Tokyo 113-8658

Information Technology Center, The University of Tokyo

Joint Usage/Research Center for Interdisciplinary Large-scale Information Infrastructures Office

(You may use the abbreviations, “Information Technology Center Office,” or the “JHPCN Office”)



2017/12/06

12

Attachment 1

List of research resources available at the JHPCN member institutions for the Joint Research

Projects

The research resources that can be directly connected via SINET5 L2VPN provided by National

Institute of Informatics is annotated as “L2VPN ready.”

（１）List of the HPCI-JHPCN system resources available for Joint Research Projects

JHPCN Institution

Computational Resources, Type of Use

Estimated number of Projects adopted

Information Initiative Center, Hokkaido University

① Supercomputer HITACHI SR16000/M1

(Max. 4 node ･four months per 1 project)

2018-04 – 2018-07 (due to system renewal)

《Hardware resources》

168 nodes，5,376 physical cores，Total main memory capacity 21TB，164.72 TFLOPS (Shared with general user, MPI parallel processing of up to a maximum of 128node per 1 job is possible) 1) Calculation time: 27,000,000 seconds: file 2.5 TB 2) Calculation time: 2,700,000 seconds: file 0.5 TB (Assumed amount of resources:166,666,667s and 48 TB when combining 1) and 2))

《Software resources》

【Compilers】Optimized FORTRAN90 (Hitachi),

XL Fortran (IBM)，XL C/C++ (IBM)

【Libraries】MPI-2 (without dynamic process creation function)，

MATRIX/MPP，MATRIX/MPP/SSS，MSL2，NAG SMP Library，

ESSL，Parallel ESSL， BLAS，LAPACK，ScaLAPACK，ARPACK，FFTW2.1.5/3.2.2

【Application software】Gaussian

② Cloud-System Blade Symphony BS2000 2018/04 - 2018/10 (due to system renewal)


Virtual server（”S”/”M”/”L” server）

2 nodes，80 cores

Equal to 8 units of “L” server (May also use “S” server and “M” server)

(Max. 4 units･seven months per 1 project)

Physical server（”XL” server）

1 unit (40 cores，Memory 128GB，DISK 2TB)

《Usage》

L2VPN Ready: Only for ” XL” server. (VPN service is available for “S”/”M”/”L” servers by “Cloud Stack”)

③ Data Science Cloud System HITACHI HA8000


1) Physical server 2 units (20 cores, Memory 80GB, Hard-Disk 2TB)

①＋②：10

③

Max.2

④＋⑤：8

⑥：4

2017/12/06

13

《Usage》

L2VPN Ready

④ New Supercomputer A 2019/01 - 2019/03 (due to system renewal)


(Max. 32 node years per 1 project)

About 900 nodes，36,000 physical cores，Total main memory capacity

337TB，2.5PFLOPS

(Shared with general user, MPI parallel processing of up to a maximum of 128node per 1 job is possible) 1) Calculation time: 15,000,000 seconds: file 3TB (Assumed amount of resources: 3,000,000,000 s and 600TB for multiple sets of 1))


【Compilers】Fortran, C/C++

【Libraries】MPI, Numerical libraries

⑤ New Supercomputer B 2019/01 - 2019/03 (due to system renewal)


(Max. 16 node years per 1 project)

About 400 nodes，24,000 physical cores，Total main memory capacity

43TB，1.2PFLOPS

(Shared with general user, MPI parallel processing of up to a maximum of 64node per 1 job is possible) 1) Calculation time: 15,000,000 seconds: file 3TB (Assumed amount of resources: 1,500,000,000 s and 600TB for multiple sets of 1))


【Compilers】Fortran, C/C++

【Libraries】MPI, Numerical libraries

⑥ New Cloud System 2018/12 – 2019/3 (due to system renewal)


1) Physical server 5 nodes (40+ cores, Memory 256+ GB, DISK 2+ TB)

2) Intercloud package 1 sets (Physical servers each of which is installed at Hokkaido university, University of Tokyo, Osaka university, and Kyushu university, connected via SINET VPN)

3) Virtual server (L server) 8 nodes (10core Memory 60+ GB, DISK 500+GB)

《Usage》

L2VPN Ready

Cyberscience Center, Tohoku University

① Supercomputer SX-ACE(2,560nodes)


48 node years / project

About 707TFLOPS，Main memory 160TB，Maximum number of nodes

1,024, Shared use


【Compilers】FORTRAN90/SX, C++/SX, NEC Fortran2003

【Libraries】MPI/SX, ASL, ASLSTAT, MathKeisan

10

2017/12/06

14

② Supercomputer LX406Re-2(68nodes)


12 node years / project

About 31.3TFLOPS，Main memory 8.5TB，Maximum number of nodes

24，Shared use


【Compilers】Intel Compiler(Fortran, C/C++)

【Libraries】MPI, IPP, TBB, MKL, NumericFactory

【Application software】Gaussian16

③ Storage 4PB


20TB / project(possible to add)

Information Technology Center, the University of Tokyo

① Reedbush-U (Intel Broadwell-EP cluster, High-speed file cache system (DDN IME))


Maximum tokens for each project: 16 node-year, Storage 16TB (138,240 node-hour) Options: node occupied service, customized login nodes, L2VPN Ready (negotiable)


【Compilers】Intel Fortran, C, C++

【 Libraries 】 MPI, BLAS, LAPACK/ScaLAPACK, FFTW, PETSc,

METIS/ParMETIS, SuperLU/SuperLU_DIST

【Application software】OpenFOAM, ABINIT-MP, PHASE, FrontFlow/Blue,

FrontFlow/Red, REVOCAP, ppOpen-HPC, BioPerl, BioRuby, BWA, GATK, SAMtools, K MapReduce, Spark

② Reedbush-H (Intel Broadwell-EP+NVIDIA P100 (Pascal) cluster (2 GPU per node), High-speed file cache system (DDN IME))


Maximum tokens for each project: 8 node-year, Storage 32TB (69,120 node-hour)


【Compilers】Intel Fortran, C, C++, PGI Fortran, C, C++ (with Accelerator

support), CUDA Fortran, CUDA C

【Libraries】GPUDirect for RDMA: Open MPI, MVAPICH2-GDR, cuBLAS,

cuSPARSE, cuFFT, MAGMA, OpenCV, ITK, Theano, Anaconda, ROOT, TensorFlow

【Application software】Torch, Caffe, Chainer, GEANT4

③ Reedbush-L (Intel Broadwell-EP+NVIDIA P100 (Pascal) cluster (4 GPU per node), High-speed file cache system (DDN IME))


Maximum tokens for each project: 4 node-year, Storage 16TB (34,560 node-hour) Options: node occupied service, customized login nodes, L2VPN Ready (negotiable)


Same as Reedbush-H

④ Oakforest-PACS (Intel Xeon Phi 7250 (Knights Landing))


10

2017/12/06

15

Maximum tokens for each project: 64 node-year, Storage 16TB (552,960 node-hour)


【Compilers】Intel Fortran, C, C++

【 Libraries 】 MPI, BLAS, LAPACK/ScaLAPACK, FFTW, PETSc,

METIS/ParMETIS,, SuperLU/SuperLU_DIST

【 Application software (planned) 】 OpenFOAM, ABINIT-MP, PHASE,

FrontFlow/Blue, FrontFlow/Red, REVOCAP, ppOpen-HPC

Global Scientific Information and Computing Center, Tokyo Institute of Technology

① Cloudy, Big-Data and Green Super Computer “TSUBABE3.0”


TSUBAME3.0 system includes 540 compute nodes, which provides 12.15PF performance (CPU 15,120 cores, 0.70PF + GPU 2,160 slots, 11.45PF) in total. Maximum system available at a time is 50% of full-system. (Shared use) Total provided resource is 230 units (= 230,000 node-hour, 1 unit = 1,000 node-hour). 27 units (= 3.125 node-year) are maximum resources for each project. Please specify not only total amount of resources but also quarterly amounts of resources. Maximum storage is 300TB for each project. Ensuring 1TB of storage for one year requires 120 node-hour of resource. Resources should be requested accordingly.


【OS】SUSE Linux Enterprise Server 12 SP2

【Language Compiler:】Intel Compiler (C/C++/Fortran), PGI Compiler

(C/C++/Fortran, OpenACC, CUDA Fortran), Allinea FORGE, GNU C, GNU Fortran, CUDA, Python, Java SDK, R

【Libraries】OpenMP, MPI (Intel MPI, OpenMPI, SGI MPT)，BLAS，

LAPACK，CuDNN, NCCL, PETSc, fftw, PAPI

【Application software】Gaussian, Gauss View, AMBER (only for academic

users) , CST MW-STUDIO (only for Industrial Joint Research Projects), Caffe, Chainer, TensorFlow, Apache Hadoop, ParaView, POV-Ray, VisIt, GAMESS, CP2K, GROMACS, LAMMPS, NAMD, Tinker, OpenFOAM, ABINIT-MP

14

Information Technology Center, Nagoya

University

(1) Fujitsu PRIMEHPC FX100


3.2PF (2,880 nodes, 92,160 cores, 92TiB memory)


【OS】FX100 dedicates OS (Red Hat Enterprise Linux for login node)

【Compilers】 Fortran，C，C++，XPFortran

【Libraries】 MPI, BLAS, LAPACK, ScaLAPACK, SSLII, C-SSL II,

SSL,II/MPI, NUMPAC, FFTW, HDF5

【Application software】Gaussian, LS-DYNA, AVS/Express,

AVS/Express PCE(parallel version ), EnSight Gold, IDL, ENVI, ParaView, Fujitsu Technical Computing Suite (HPC Portal)

(2) Fujitsu PRIMERGY CX400/2550


447TF (384 nodes, 10,752 cores, 48TiB memory)


【OS】 Red Hat Enterprise Linux

【Programming languages】 Fujitsu Fortran/C/C++/ XPFortran，Intel

Fortran /C/C++, Python (2.7, 3.3)

【Application software】 MPI(Fujitsu, Intel), BLAS, LAPACK,

(1) +(2)＋

(3)+(4):10

2017/12/06

16

ScaLAPACK,SSLII(Fujitsu), C-SSL II, SSL II/MPI, MKL(Intel), NUMPAC, FFTW, HDF5, IPP(Intel)

【Application software】 LS-DYNA, ABAQUS, ADF, AMBER,GAMESS,

Gaussian, Gromacs, LAMMPS, NAMD, HyperWorks,OpenFOAM, STAR-CCM+, Poynting, AVS/Express Dev/PCE,EnSightGold, IDL, ENVI, ParaView, Fujitsu Technical Computing Suite (HPC Portal) (3) Fujitsu PRIMERGY CX400/270


279TF (184 nodes + 184 Xeon Phi, 4,416 + 10,488 cores, 23TiB memory)


Same to software resources of (2) Fujitsu PRIMERGY CX400/2550 (4) SGI UV2000


24TF (1,280 cores, 20TiB shared memory, 10TB memory base file system) 3D visualization system:

High resolution 8K display system (185in size by 16 panel tile display), Full HD circularly polarized light binocular vision system (150in screen, projector, circularly polarized light eyeglasses, etc.),

head mount display system (MixedReality, VICON, etc.), dome type display system


【OS】 SuSE Linux Enterprise Server

【Programming languages】Intel Fortran C C++, Python (2.6, 3.5)

【Libraries】 MPT(SGI),MKL(Intel), PBS Professional, FFTW,

HDF5,NetCDF, scikit-learn, TensorFlow

【Application software】 AVS/Express Dev/PCE, EnSightDR,

IDL,ENVI(SARscape), ParaView, POV-Ray, NICE DCV(SGI), ffmpeg, ffplay, osgviewer, vmd(Visual Molecular Dynamics) Maximum resource allocation to one project: 15 unit or 86 node year 1 unit becomes 50 thousand node hour (equivalent to 180 thousand

yen charge) You can use large scale storage by converting 5000 node hour to

10TB. All resources are shared with general users. When using 3D visualization system, it is 100 thousand yen per

research projects.(Equivalent to additional contribution)

Academic Center for Computing and Media Studies, Kyoto University

Cray XC40(Camphor 2:Xeon Phi KNL/node)


① 128 nodes, 8,704 cores, 390.4TFLOPS×throughout the year (Each

project is assigned up to 48 nodes×throughout the year)

② 128 nodes, 8,704 cores, 390.4TFLOPS×20 weeks (Each project is

assigned up to 128 nodes×4 weeks. )

Available amount of computing resources are adjusted by the projects.)


【Compilers】Fortran2003/C99/C++03(Cray, Intel)

【Libraries】MPI, BLAS, LAPACK, ScaLAPACK

【Application Software】Gaussian09

《Usage》

L2VPN ready (negotiable).

① ②：5

Cybermedia Center,

(i) Super Computer SX-ACE

《Hardware resources》 (i)(ii)(iii): 10

http://www.hpc.cmc.osaka-u.ac.jp/en/sx-ace-sys/

2017/12/06

17

Osaka University

- Resource per project: up to 22 node years

- Computational node: 512 nodes (141 TFlops) will be provided up to 280,000 node hours in shared use.

- Storages per project: 20 TByte


- 【Compilers】C++/SX, FORTRAN90/SX, NEC Fortran 2003 compiler

- 【Libraries】ASL, ASLQUAD, ASLSTAT, MathKeisan(including BLAS,

BLACS, LAPACK, ScaLAPACK), NetCDF (ii) PC Cluster for large scale visualization (possible to link with the large-

scale visualization system)


- Resource per project: up to 6 node years

- Computational node: 69 nodes (CPU: up to 31.1 TFlops, GPU: up to 69.03 TFlops) will be provided up to 53,000 node hours in shared use. Arbitrary number of GPU resources can be attached to each node based on user requirement. Simultaneous use of this system and the large-scale visualization system is possible based on user requirement.

- Storages per project: 20 TByte


- 【Compilers】Intel Compiler (C/C++, Fortran)

- 【Libraries】Intel MKL, Intel MPI(Ver.4.1)

- 【 Application software 】 OpenFOAM, GROMACS (for CPU/for

GPU),LAMMPS (for CPU/for GPU), Gaussian09, AVS/Express PCE,AVS/Express MPE (Ver.8.2)

(iii) OCTOPUS


- Resource per project: General purpose CPU nodes: up to 35 node years GPU nodes: up to 5 node years Xeon Phi nodes: up to 6 node years Large-scale shared-memory nodes: up to 0.3 node years

- Computational node: General purpose CPU nodes: 236 nodes (CPU: 471.1 TFLOPS) will be provided up to 310,000 node hours in shared use. Because processors on these nodes are the same as that of GPU nodes, users can use 273 nodes for a job. GPU nodes: 37 nodes (CPU: 73.9 TFLOPS, GPU: 784.4 TFLOPS) will be provided up to 49,000 node hours in shared use. Xeon Phi nodes: 44 nodes (CPU: 117.1 TFLOPS) will be provided up to 58,000 node hours in shared use. Large-scale shared-memory nodes: 2 nodes (CPU: 16.4 TFLOPS, memory: 12TB) will be provided up to 2,600 node hours in shared use. Storages per project: 20 TByte


[Compilers] Intel Compiler (FORTRAN, C, C++), GNU Compiler (FORTRAN, C, C++), PGI Compiler (FORTRAN, C, C++), Python 2.7/3.5, R 3.3, Julia, Octave, CUDA, XcalableMP, Gnuplot

[Library] IntelMPI, OpenMPI, MVAPICH2, BLAS, LAPACK, FFTW, GNU Scientific Library, NetCDF 4.4.1, Parallel netcdf 1.8.1, HDF5 1.10.0

[Application software]

http://www.hpc.cmc.osaka-u.ac.jp/en/vcc-sys/

http://www.hpc.cmc.osaka-u.ac.jp/en/vcc-sys/

http://www.hpc.cmc.osaka-u.ac.jp/en/octopus/

2017/12/06

18

NICE Desktop Cloud Visualization, Gaussian16, GROMACS, OpenFOAM, LAMMPS, Caffe, Theano, Chainer, TensorFlow, Torch, GAMESS, ABINIT-MP, Relion

Research Institute for Information Technology, Kyushu University

① ITO Subsystem A (Fujitsu PRIMERGY)

《Hardware Resources》

1) (Nearly dedicated-use) The Maximum resources allocated for 1 project is 32 nodes for a year. Most of resources are dedicated to the project. 32 nodes(1,152 cores), 110.59 TFLOPS

2) (Shared-use) Up to 64 nodes can be used at the same time per project. It is shared with general users. 64 nodes(2,304 cores), 221.18 TFLOPS

《Software Resources》

【Compilers】Intel Cluster Studio XE(Fortran, C, C++), Fujitsu Compiler

② ITO Subsystem B (Fujitsu PRIMERGY)


(Nearly dedicated-use) The maximum resource allocation per task is 16 nodes for a year. Most of resources are dedicated to the project.

16 nodes(576 cores), CPU 42.39TFLOPS + GPU 339.2TFLOPS，

including SSD


【Compilers】Intel Cluster Studio XE(Fortran, C, C++), Fujitsu Compiler,

CUDA

③ ITO Frontend (Virtual server / Physical server)


Standard Frontend：

1 nodes(36 cores), CPU 2.64TFLOPS，Memory 384GiB，GPU

（NVIDIA Quadro P4000）

Large Frontend：

1 nodes(352 cores), CPU 12.39TFLOPS，Shared memory 12GiB，

GPU（NVIDIA Quadro M4000）


【Compilers】Intel Cluster Studio XE(Fortran, C, C++), CUDA

Storages per project: 10 TByte, (possible to add)

Regarding ①/②, a server for pre/post-processing or visualization is

available (Standard Frontend)

①1):2

2):8

②:3

③:5

2017/12/06

19

（２）Other facilities/resources available for Joint Research Projects

The following facilities/resources, despite not being part of the HPCI-JHPCN system, are available

for Joint Research Projects.

JHPCN Institution

Computational Resources, Type of Use

Estimated number of Projects adopted



(1) Large-format printer


《Usage》

12



(1) Large-format printer (2) 3D visualization system(12 screens large stereoscopic display

system(1920×1080(FULL HD) 50 inch projection module × 12

screens), visualization server × 4, 3D visualization glasses)


AVS/Express MPE(3D visualization software)

《Usage》

・ L2VPN Ready

・ On-demand L2VPN

10



(1) FENNEL (Real-time Data Analysis Nodes) x 4 At maximum four VMs or one bare metal server are provided to each group. The provided VMs or a bare metal server are dedicated to the group.

(2) GPGPU (Nvidia Tesla M60) is available on request. 10Gbps Network Access

(3) etwork Attached Storage (NAS) 150 TB 10Gbps Network Access


Analysis nodes

【OS】Ubuntu 16.04, Ubuntu 17.04, CentOS 7.4

【Programming Language】Python, Java, R

【Application Software】Apache Hadoop, Apache Spark, Apache Hive,

Facebook Presto, Elastic Search, Chainer, Tensor Flow

《Usage》

L2VPN ready (1) Analysis nodes

You can directly login to the nodes via remote network with SSH. You can also mount NAS as native file system.

(2) NAS You can mount with iSCSI, NFS and/or CIFS protocol via SINET L2VPN. Y

《Other Options》

Housing services 20U Rack-Space (Arm-Type Racking)

3 or 4

2017/12/06

20

Network Connectivity (Network Bandwidth: 10Gbps, (consultation

is necessary), Power: AC100V/15A）

Global Scientific Information and Computing Center, Tokyo Institute of Technology


(1) Remote GUI environment: the VDI (Virtual Desktop Infrastructure)

system

If you are planning to use the VDI system, please contact us in advance.


《Usage》

Information Technology Center, Nagoya University


(1) Visualization system Remote visualization system, on-site use(for data translation)

(2) Network connection up to 40GBASE Note: Nagoya University only provides SFP+/QSFP port so that you have to prepare optical modules including the module for network switch of Nagoya University.


(1) Fujitsu PRIMERGY CX400/2550 Python + scikit-learn + TensorFlow Other software which is executable on Linux

(2) Visualization system AVS/Express Dev/PCE, EnsightHPC, 3DAVSPlayer, IDL, ENVI (SARscape),ParaView, osgviewer, EasyVR, POV-Ray, 3dsMAX, 3- matic,VMD, LS-prepost

《Usage》

L2VPN Ready You can use L2VPN network connection when you connect own system to the network via hardware resource (2). We configure SINET L2VPN connection to outer university and internal university VLAN (with connection between them)



(1) Virtual Server Hosting Standard Spec: CPU 2 Cores, Memory 8GB, Disk capacity 1TB


(1) 【Hypervisor】VMware

【OS】CentOS7

【Application】Basic software prepared at the center and support for

introducing various open source software.

《Usage》

L2VPN Ready Dedicated VLAN (on-campus) or SINET L2VPN(off-campus) Connection

Cybermedia Center, Osaka University


1. 24-screen Flat Stereo Visualization System

・ Install location: Umekita Office of the Cybermedia Center, Osaka University

・ 50-inch FHD (1920 x 1080) stereo projection module: 24 displays Image Processing PC Cluster: 7 computing nodes

・ * Prioritize the visualization by users of “PC cluster for the large-scale visualization”.

2. 15-screen Cylindrical Stereo Visualization System

・ Install location: the main building of the Cybermedia Center, Suita Campus, Osaka University

・ 46-inch WXGA (1366 x 768) LCD: 15 displays

2017/12/06

21

・ Image Processing PC Cluster: 6 computing nodes * Prioritize the visualization by users of “PC cluster for the large-scale visualization”.


・ AVS Express/MPE, IDL

・ AVS Express/MPE, IDL

《Usage》

L2VPN Ready



(1) Visualization environment Visualization server (SGI Asterism), 4K projector, and 4K monitor

《Usage》

L2VPN Ready

2017/12/06

22

Attachment 2

The following are possible examples of “Research projects using both large-scale data and large

capacity networks” described in Attention ② of "1. Joint Research Areas" in the call for

proposals. The purposes of this attachment are to present, by example, how research resources

provided by JHPCN member institutions may be used. We appreciate proposals of joint research

projects using both large-scale data and large capacity networks, not limited to those examples.


Distributed systems (virtual private cloud systems) can be deployed using our cloud system / data

science cloud system federated with computational resources of the other universities or the

applicant’s laboratory, connected via SINET L2VPN or with software VPN solutions.

Available resources

Supercomputer system, Cloud system, data science cloud system (c.f. Attachment 1.)

How to use

Dedicated systems can be developed for the collaborative research projects employing physical

and virtual servers as virtual private clouds. Distributed systems can also be developed by

connecting our cloud resources with servers and storages in the users’ laboratories as hybrid

cloud systems. The users can access the systems not only via ssh/scp but also with virtual

console, which provided by Cloud Middleware,through web browsers.

Email address for inquiring about resource usage and joint research

[email protected]

Details of anticipated projects

- Experiment data analysis platform in the intercloud environment: constructing a data store,

analysis, and sharing infrastructure employing virtual/real machines and storages of the

cloud system / data science cloud system of Hokkaido university connected to computational

resources of the other universities via SINET L2VPN.

- Development of a large-scale pre/post-processing environment federating supercomputers

and intercloud systems: developing a large-scale distributed processing environment such

as performing analysis of big data generated by supercomputers using Hadoop clusters to

visualize at the other universities’ remote systems.

- An always-on platform to support network-oriented research projects: development of a


2017/12/06

23

nation-wide distributed high-speed networking platform employing the cloud system / data

science cloud system of Hokkaido university and private clouds of the other universities

connected via SINTE5 L2VPN.


Cyberscience center provides vector parallel and scalar parallel supercomputers, a distributed

data sharing environment through on-demand L2VPN, and a large-scale remote visualization

environment with other centers. These environments allow users to share and analyze vast

amounts of observed data obtained by sensors or IoT devices. We strongly invite proposals that

try to exploit the potential of these environments. For example, joint research regarding the real-

time analytics using supercomputers, and storage/network architectures for a large-scale

distributed data sharing.

Available resources


・Storage (500TB / project)

・3D visualization system

・Supercomputers（SX-ACE, LX406 Re-2）

・On-demand L2VPN

2017/12/06

24


【OS】Redhat Enterprise Linux 6

【Programing languages】

LX406 Re-2: Fortran,C, C++, Ruby, Python, java, etc.

SX-ACE : Fortran, C, C++

【Application software】

Basic applications provided by Cyberscinece Center and original codes developed by

users. We also support installing/migrating required software to our system. (please

contact us in advance.)

How to use

Supercomputers (LX406 Re-2, SX-ACE)

• log in to the compute nodes using ssh

• transfer files to the node using the scp / sftp

Network

• Possible to build L2VPN on SINET5

Storage

• Possible to use remote mount by NFS through L2VPN


[email protected]

Details of anticipated projects (in japanese)

http://www.ss.cc.tohoku.ac.jp/collabo/net/


http://www.ss.cc.tohoku.ac.jp/collabo/net/

2017/12/06

25


Available resources and Connectivity to SINET5

(1) Reedbush System

We would like to offer Reedbush-U, H, and L systems for Big-Data analyses using Burst

Buffer (DDN IME) which incorporates high-speed and high-capacity SSD storages. In

addition, we will offer the direct connection via SINET5 L2VPN for login-node of

Reedbush-U and L (Please ask us for detailed configuration).

Available Resources


Reedbush-U, H, L System


Refer to description of Reedbush-U, H, L System in Attachment 1.

How to use

You can directly login to the nodes via remote network with SSH.

You can transfer data via remote network with SCP / SFTP.

You can directly access to login node via SINET5 L2VPN (negotiable for

Reedbush-U and L).

(2) FENNEL and Network Access Storage

We, Information Technology Center in the University of Tokyo, would like to offer a Big-Data

analysis infrastructure in which a researcher can take sole usage of its resources. The key design

concept of the infrastructure is to facilitate real-time analysis and accumulating intermittently

generated data. Further, we will also offer an environment which can connect both your own

2017/12/06

26

resources and our resources with SINET L2VPN service. The anticipated issues for this resources

are IoT Big-Data analysis, SNS message analysis, anomaly detection of network traffic, and etc.

Overview of FENNEL

Available Resources


FENNEL (Real-Time Dana Analysis Nodes) x 4 nodes

At maximum four VMs or one bare metal server are provided to each group.

The provided VMs or a bare metal server are dedicated to the group.

GPGPU (Nvidia Tesla M60) is available on request.

Network Access Storage (150TB)

10GbE Network Access

Rack Space (with our Housing Service: You can bring your own devices into our

datacenter and connect to the network)


【OS】Ubuntu 16.04, Ubuntu 17.04, CentOS 7.4

【Programming languages】Python, Java, R

【Application software】Apache Hadoop, Apache Spark, Apache Hive, Apache Impala,

Presto, Elastic Search, Chainer, Tensor Flow

2017/12/06

27

How to use

FENNEL (Dedicated and Real-Time Data Analysis Nodes)

You can directly login to the nodes via remote network with SSH.

You can transfer data via remote network with SCP / SFTP.

You can use GPGPU on VM if you want.

Network Access Storage (NAS)

You can mount with iSCSI, NFS and/or CIFS protocol via SINET5 L2VPN.

You can use data proxy to store your own data that you obtain form your own

sensor devices.

The volumes provided by the storage can be mounted on VMs as native filesystem.


[email protected]


https://www.itc.u-tokyo.ac.jp/iiccs/supplement2018-en

GSIC, Tokyo Institute of Technology

Machine learning jobs, especially in deep learning which recently attracts great attention, require

both storage resources for storing large scale I/O data and high performance computation

resources. For these jobs, we provide environment for large-scale high-performance machine

learning by using lots of GPUs (>2,000 in the whole system) and large storage (up to 300TB per

user group) equipped by the TSUBAME3.0 supercomputer. By using pre-installed frameworks

that harnesses GPUs, acceleration of research projects of large scale machine learning is

expected.


https://www.itc.u-tokyo.ac.jp/iiccs/supplement2018-en

2017/12/06

28

Available Resources


Refer to description of TSUBAME3.0 in Appendix 1. Especially, 4 Tesla P100 GPUs per node

are available.


Refer to description of TSUBAME3.0 in Appendix 1. The followings are highly related items to

this page:

- OS: SUSE Linux Enterprise Server 12 SP2

- Programming Languages: Python, Java SDK, R

Application software: Caffe, Chainer, TensorFlow

How to use

TSUBAME3.0

Same as regular usage.


[email protected]


http://www.gsic.titech.ac.jp/en/jhpcn/dl-en

Information Technology Center, Nagoya University

We provide SGI UV2000 subsystem for very large scale data processing research field which is

a large scale shared memory system equipping various storage system. UV2000 subsystem

equips 10TB memory based file system by utilizing large scale shared memory, SSD file system

which is superior in random access performance, and 2.9PB HDD array which is connected with

DAS as storage systems. You have to include planned quota in proposal. UV2000 subsystem

equips scikit-learn and TensorFlow libraries for machine learning and deep learning usage.

Additionally, Fujitsu PRIMERGY CX400 subsystem equips many CPU cores including Intel Xeon

Phi (Knights Corner) so that it is suitable for data processing which requires much CPU power.

We are just supposing them for very large scale data processing or real-time large scale

information collection and summarizing from the Internet, as research candidates for this

subsystem.

We provide up to 40GBASE network connection for very large bandwidth network technology

research field. You can connect your own system to given network port (with own optical modules)

to create very large bandwidth network experiment environment by creating L2 flat network via


http://www.gsic.titech.ac.jp/en/jhpcn/dl-en

2017/12/06

29

SINET L2VPN for external university and internal university VLAN. We are just supposing it for

large scale parallel data processing with utilizing remote computer systems or speculative data

caching via large bandwidth network, as research candidates for this resource.

We provide visualization subsystem and visualization software suite which can utilize from

remote environment for very large scale information system research field. The subsystem is

connected to SGI UV2000 subsystem and we execute the visualization software suite on UV2000

subsystem. We are just supposing them for remote visualization processing, VR teleconferencing

system which accelerates brain storming, and high performance communication protocol to real-

time remote visualization, as research candidates for this subsystem.

Available Resources


1. SGI UV2000 subsystem (1280 cores, 24 TF, 20TiB shared memory) + memory base

file system + SSD file system + HDD array (2.9PB)

2. Fujitsu PRIMERGY CX400 subsystem

3. Visualization subsystem: HMD system (Canon HM-A1, nVisorST50, Sony HMZ-T2) ,

dome type display system, 8K display system

4. Network connection up to 40GBASE (with internal university VLAN and SINET

100G

10Gx4

40Gx2

SINET5

Border switch

super core SW

(10G)

40G

Experiment with very large

network bandwidth with

40GBASE(max) connection and

SINET L2VPN connection

vPC

10G

Fujitsu PRIMERGY CX400/2550

Fujitsu PRIMERGY CX400/270

10G

A subsystem for very large

scale data processing by

CPU core amount oriented

system (with Xeon Phi)

A visualization subsystem

for very large scale

information system

A subsystem for very large

scale data processing by data

supply speed oriented system

(with SSD file system)

HDD array storage

InfiniteStorage 17000

2.39PB

File system with SSD

512TB

Visualization system

with

file system controller

1280 cores (24TF/s),

20TiB shared memory

High resolution display system

7680 x 4320 resolution

4m width x 2.3m height (16 tiles) Dome type display system

Handy type

HMD systemStationary type

HMD system

2017/12/06

30

L2VPN configuration)


1. SGI UV2000 subsystem

【OS】SuSE Linux Enterprise Server

【Compilers】Fortran/C/C++, Python (2.6, 3.5)

【Libraries】MPT(SGI), MKL(Intel), PBS Professional, FFTW, HDF5, NetCDF, scikit-

learn, TensorFlow

【Application software】AVS/Express Dev/PCE, EnSightHPC, FieldviewParallel, IDL,

ENVI(SARscape), ParaView, 3DAVSplayer, POV-Ray, NICE DCV(SGI), ffmpeg,

ffplay, LS-prepost, osgviewer, vmd(Visual Molecular Dynamics)

2. Fujitsu PRIMERGY CX400 subsystem

【OS】Red Hat Enterprise Linux

【Compilers】Fujitsu Fortran/C/C++/ XPFortran，Intel Fortran /C/C++, Python (2.7, 3.3)

【Libraries】MPI(Fujitsu, Intel), BLAS, LAPACK, ScaLAPACK, SSLII(Fujitsu), C-SSL II,

SSL II/MPI, MKL(Intel), NUMPAC, FFTW, HDF5, IPP(Intel)

【Application software】 LS-DYNA, ABAQUS, ADF, AMBER, GAMESS, Gaussian,

Gromacs, LAMMPS, NAMD, HyperWorks, OpenFOAM, STAR-CCM+, Poynting, Fujitsu

Technical Computing Suite 4 (HPC Portal)

3. Visualization subsystem

【Visualization software suite】AVS/Express Dev/PCE, EnsightHPC, 3DAVSPlayer, IDL,

ENVI(SARscape), ParaView, osgviewer, EasyVR, POV-Ray, 3dsMAX, 3-matic, VMD,

ffmpeg, ffplay, LS-prepost

How to use

Remote login with ssh through login node.

You can use remote visualization by VisServer on UV2000 subsystem.

File transfer with scp / sftp through login node.

Job posting and job management through Fujitsu Technical Computing Suite (HPC Portal).


[email protected]

Very large scale numerical calculation field: Takahiro Katagiri (High Performance

Computing Division)

Very large scale data processing field: Takahiro Katagiri (High Performance

Computing Division)

Very large bandwidth network technology field: Hajime Shimada (Advanced


2017/12/06

31

Networking Division)

Very large scale information system field: Masao Ogino (High Performance

Computing Division)


http://www.icts.nagoya-u.ac.jp/en/center/jhpcn/suppl/


We will provide the infrastructure for collecting large scale data from laboratory equipment and

observation equipment possessed by researchers via large capacity network or internet such as

Kyoto University internal LAN (KUINS) or SINET5 L2VPN for 24 hours a day, 365 days and

analyzing them with a supercomputer system in real time or periodically then offering information

of the results on the Web.

Available resources


・Supercomputer system

Cray XC40 (Camphor 2: Xeon Phi KNL/node) Each project is assigned up to 48 nodes×

throughout the year.

DDN ExaScaler (SFA14K) Each project is assigned up to 192TB

・Mainframe System Virtual Server Hosting

Virtualized environment: VMWare

Standard Spec: CPU 2 Cores, Memory 8GB, Disk capacity 1TB



【Compilers】Fortran2003/C99/C++03(Cray, Intel)

【Libraries】MPI, BLAS, LAPACK, ScaLAPACK

・Mainframe System VM Hosting

Standard OS: CentOS7

How to use


Login with SSH (Key authentication)

・Mainframe System VM Hosting

Login with SSH (Granting Root authority)

Access by various service port such as HTTP (80/TCP) or HTTPS (443/TCP)

Multiple virtual domains are available

http://www.icts.nagoya-u.ac.jp/en/center/jhpcn/suppl/

2017/12/06

32

Kyoto University internal LAN (KUINS) VLAN and SINET5 L2VPN can be housed directly in VM


[email protected]

Cybermedia Center, Osaka University

Our center provides L2 VPN connection service between other site and Osaka University testbed

network through SINET. It aims to construct an environment of sharing visualization contents with

our systems, devices and storages. Please contact us for detail of our service, usage of our

resources or research collaboration.


[email protected]


http://www.hpc.cmc.osaka-u.ac.jp/en/for_jhpcn/

http://vis.cmc.osaka-u.ac.jp/en/


We provide a remote visualization and data analytic infrastructure that researchers can use from




http://www.hpc.cmc.osaka-u.ac.jp/en/for_jhpcn/

http://vis.cmc.osaka-u.ac.jp/en/

2017/12/06

33

remote sites. The provided system allows us to process generated large-scale data without

moving, thus efficient processing is possible. Besides, if available, L2VPN enables the combined

usage of resources between end users and bases. The provided resources are assumed to be

used for research subjects that visualize and analyze large-scale parallel simulation and/or

observation data. Available user scenarios are batch mode (use the back-end nodes), interactive

mode (use the front-end nodes), and in-situ mode (use both the front-end and the back-end nodes

simultaneously).

If the data you generated does not correspond to the data format of the provided software or the

supplied system does not have the analysis function you want, consultation is available.

Available resources

Hardware resources

・ Subsystem A, Subsystem B, Standard Frontend (c.f. Attachment 1.)

Software resources

・【OS】Linux

・【Programming languages】Python, R

・【Application software】Tensor Flow, OpenFOAM, HIVE(Visualization)

How to use

Batch environment

2017/12/06

34

Direct login is possible to the node using ssh via the network.

File transfer is possible to/from the node using scp/sftp via the network.

Conventional batch usage.

Interactive environment

Login is possible to the front-end node using ssh.

Real-time parallel visualization and data analysis are performed using visualization

application that runs on the front-end. In the situation where a job is running on the

back-end node, it provides an interactive visualization environment through the file

information between the back-end and the front-end. The interactive rate is

assumed to be on the order of several to 0.1 fps depending on the communication

bandwidth and the amount of transferred data.


[email protected]


https://www.cc.kyushu-u.ac.jp/scp/service/jhpcn/jhpcn.html (in Japanese)


https://www.cc.kyushu-u.ac.jp/scp/service/jhpcn/jhpcn.html

joint usage/research center for interdisciplinary large ... · 2018 call for proposal of joint...

Documents