Ultra-Large-Scale HPC Challenge
Abstract
The “Ultra-Large-Scale HPC Challenge" is an open call project that provides an environment for executing large-scale parallel computing on the MASAMUNE-II. Following each scheduled maintenance period, participants are permitted to run one large-scale job on the Large-Scale Parallel Computing Server or the Accelerator Server.
Supercomputer resources
The following queues are available.
Large-Scale Parallel Computing Server（HPE Cray XD220v）
| Queue name | Maximum number of nodes |
Memory limit [GiB] |
Walltime limit [hours] |
Maximum number of parallel processes |
|---|---|---|---|---|
| HP_127 | 127 | 68,080 | 24 | 14,224 |
Accelerator Server（HPE Cray XD670）
| Queue name | Maximum number of nodes |
Memory limit [GiB] |
Walltime limit [hours] |
Maximum number of GPUs |
|---|---|---|---|---|
| HA_011 | 11 | 10,120 | 24 | 88 |
User Guidelines
Applications to this program must be submitted in response to the open call.
In each call, only one application will be selected for each of
the Large-Scale Parallel Computing Server and the Accelerator Server.
Only proposals that use in-house (self-developed) programs are eligible;
those relying on commercial or existing software are not accepted.
Approved jobs will be executed following each maintenance period.
Each application is limited to a single calculation. However, during
the first two hours, users are allowed to submit multiple jobs for
testing purposes. A final report must be submitted upon completion
of the project. The service is provided free of charge, and the
node-time consumed during execution will be returned to the user’s
allocated time after the job has finished.
Eligibility and Selection Process
・Eligibility: Applicants must be registered users of the Supercomputing
system under a project already accepted for the GIMRT in the current fiscal year.
・Selection process: Applications will be reviewed by the Committee of the Center
for Computational Materials Science. As part of the review process, the number of
times an applicant has been selected during the current fiscal year may also be
taken into consideration.
How to Apply
Please fill out the required information in the application form and send it as an email attachment to the Application Submission Office.
Schedule
For the fiscal year 2026, the application periods for the "Ultra-Large-Scale HPC Challenge" are scheduled as follows:
| Execution Period (Tentative) | Application Deadline | Notification of Selection |
|---|---|---|
| After August 2026 Scheduled Maintenance (Early August 2026) |
June 5, 2026 | Early July 2026 |
| After November 2026 Scheduled Maintenance (Late November 2026) |
September 25, 2026 | Late October 2026 |
| After February 2027 Scheduled Maintenance (Late February 2027) |
December 25, 2026 | Late January 2027 |
Workflow from Application to Execution
Detailed information regarding job submission and other
procedures will be provided to the selected applicants.
Application Submission Office
2-1-1 Katahira, Aoba-ku, Sendai 980-8577, Japan
Center for Computational Materials Science, IMR, Tohoku University
E-mail: ccms-info.imr__at__grp.tohoku.ac.jp (Please replace "__at__" with "@")