Supercomputing Division, Information Technology Center, The University Tokyo

Oakforest-PACS Supercomputer System
“Large-Scale HPC Challenge”


The Joint Center for Advanced High Performance Computing (JCAHPC) has been implementing the Large-Scale HPC Challenge for the Oakforest-PACS Supercomputer System since October 2017. The Large-Scale HPC Challenge is a public recruitment project that allows a research group to monopolize the use of computational resources for a maximum of 21 hours, for the largest number of compute nodes (8,192 nodes (557,056 cores)) possible for Oakforest-PACS Supercomputer System.

[Updated: Jan. 15, 2021]:The recruitment scheduled in FY2020 and FY2021 has been canceled for the time being due to COVID-19 pandemic. We will inform you again on this page when restarting.

[Updated: May 17, 2021]: We have resumed recruiting under some restricted conditions as below.
  • Time of Implementation: 8 hours(9 a.m.- 5 p.m. on the day of implementation)
  • The maximum number of nodes that users can monopolize is not 8,192. Users can use up to the limit of each setting: 4,200 nodes(flat mode) or 3,200 nodes(cache mode). A maximum of two cases can be accepted per month, one in flat mode and one in cache mode. However, it is possible to use both flat mode and cache mode in one group.
  • There will be no operational switchover before or after the implementation of the Large-scale HPC Challenge.

  • The schedule may be changed or the implementation will be cancelled without prior notice depending on the future situation.

    Available resources

    Oakforest-PACS Supercomputer System
    • Users can monopolize the use of 8,192 nodes (557,056 cores) for a maximum of 21 hours.
    • Users can monopolize the use of 4,200 nodes (flat mode) or 3,200 nodes (cache mode) or both for a maximum of 8 hours.

    User’s guide

    • This initiative allows users to monopolize the use of 8,192 nodes for a maximum of 21 hours once a month, in principle from 12:00 p.m. on Tuesday – 9:00 a.m. on Wednesday before the month-end processing.
      (With regard to the memory mode for Oakforest-PACS, about half of all compute nodes are set to flat mode, and the remaining half are set to cache mode during normal service for all users. However, on the day of the Large-Scale HPC Challenge, users under the Challenge initiative may change all compute nodes to the mode they prefer. This change is expected to take about 2 to 3 hours, so the start time for the Large-Scale HPC Challenge will be at about 12:00 p.m., and end at 9:00 a.m. the following day. Your kind understanding is much appreciated.)

      This initiative allows users to monopolize the use of 4,200 nodes(flat mode) or 3,200 nodes(cache mode) or both for a maximum of 8 hours once a month, in principle from 9:00 a.m. – 5:00 p.m. on Tuesday before the month-end processing.
    • The projects are recruited from the public, and a wide range of projects are accepted in addition to submissions from existing users. While applications are accepted by both individuals and groups, one group is selected each month in principle.
    • The results produced through the use of this initiative will be released to the public. In the disclosure of the results of the project, selected projects are required to state clearly that the project was implemented using the Oakforest-PACS Supercomputer System under the Large-Scale HPC Challenge initiative. Selected projects will also be requested to submit articles reporting on the results of the project to the PR magazine, etc., submit articles (brief reports) to peer-reviewed international conferences, etc.
    • There may be cases where the selected projects are invited to present at seminars and workshops organized or co-organized by JCAHPC (or the University of Tokyo or the University of Tsukuba).
    • Use of the supercomputers under this initiative is free of charge.

    Recruitment schedule

    The following is the provisional schedule for the recruitment of projects for the Large-Scale HPC Challenge.

    • The recruitment schedule is subject to changes depending on the maintenance schedule and other factors.
    • It is possible to apply more than once during the year. However, please note that there are cases where your requests cannot be fulfilled due to the application situation. It is possible to use the supercomputer systems only once with each application.

    FY2021 Schedule (Provisional)

    We have resumed recruiting under some restricted conditions.
    Implementation period Recruitment
    deadline
    Screening Notification
    of
    selection
    Tue., June 22, 2021, 9 a.m. – 5 p.m.
    Tue., July 27, 2021, 9 a.m. – 5 p.m.
    Mon.,
    May 24, 2021,
    5 p.m.
    [Closed]
    Early
    June
    2021
    Mid-
    June
    2021
    [Re-recruitment]
    Tue., July 27, 2021, 9 a.m. – 5 p.m.
    Mon.,
    June 7, 2021,
    5 p.m.
    [Closed]
    Mid-
    June
    2021
    End-
    June
    2021
    Tue., Aug. 24, 2021, 9 a.m. – 5 p.m.
    Wed., Sep. 22, 2021, 9 a.m. – 5 p.m.
    Tue., Oct. 26, 2021, 9 a.m. – 5 p.m.
    Wed., Nov. 24, 2021, 9 a.m. – 5 p.m.
    Mon.,
    June 28, 2021,
    5 p.m.
    [Closed]
    Early
    July
    2021
    Mid-
    July
    2021
    [Re-recruitment]
    Wed., Sep. 22, 2021, 9 a.m. – 5 p.m.
    Tue., Oct. 26, 2021, 9 a.m. – 5 p.m.
    Wed., Nov. 24, 2021, 9 a.m. – 5 p.m.
    Mon.,
    Aug. 2, 2021,
    5 p.m.
    [Closed]
    Early
    Aug.
    2021
    End-
    Aug.
    2021
    [Re-recruitment]
    Tue., Oct. 26, 2021, 9 a.m. – 5 p.m.
    Wed., Nov. 24, 2021, 9 a.m. – 5 p.m.
    Mon.,
    Sep. 6, 2021,
    5 p.m.
    [Closed]
    Mid-
    Sep.
    2021
    End-
    Sep.
    2021
    Tue., Dec. 14, 2021, 9 a.m. – 5 p.m.
    Tue., Jan. 25, 2022, 9 a.m. – 5 p.m.
    Mon., Feb. 21, 2022, 9 a.m. – 5 p.m.
    Wed., Mar. 30, 2022, 9 a.m. – 5 p.m.
    Mon.,
    Oct. 25, 2021,
    5 p.m.
    Early
    Nov.
    2021
    Mid-
    Nov.
    2021

    Application details

    Please click here to see the application details for the Large-Scale HPC Challenge.

    Selected projects

    FY2021: Selected projects (1st term ) new

    FY2019: Selected projects (1st term / 2nd term / 3rd term)

    FY2018: Selected projects (1st term / 3rd term)

    FY2017: Selected projects (1st term / 2nd term / 3rd term)

    English translation of acknowledgements when reporting the research results in academic papers, etc.

    Please use the English translations provided below when publishing English papers, etc. after the use of supercomputers under this initiative.

    English title of this initiative

    "Large-scale HPC Challenge" Project, Joint Center for Advanced High Performance Computing (JCAHPC).

    Example of English acknowledgement

    Computational resource of Fujitsu PRIMERGY CX600M1/CX1640M1(Oakforest-PACS) was awarded by "Large-scale HPC Challenge" Project, Joint Center for Advanced High Performance Computing (JCAHPC).

    FAQ for Large-Scale HPC Challenge