Reference Translation: Charging Rules of TSUBAME4.0 Supercomputer

This page contains an unofficial English translation of the original Japanese text for reference purposes only.
This translation may not reflect the most up-to-date version of the original.
In the event of any inconsistency between this translation and the original Japanese version, the original Japanese text shall prevail.
The original text is available here.


Charging Rules of TSUBAME4.0 Supercomputer, Institute of Science Tokyo

October 1, 2024

Article 1 (Purpose)  
These rules prescribe the matters necessary for charging usage fees and related matters for the TSUBAME4.0 supercomputer (hereinafter referred to as “TSUBAME4.0”) managed by the Center for Information Infrastructure, Institute of Science Tokyo, pursuant to Article 13, paragraph 3 of the Detailed Terms of Use for Computer Systems (hereinafter referred to as the “Detailed Rules”).

Article 2 (Charging of Usage Fees)

1. Usage fees for TSUBAME4.0 shall be charged at the time of purchase of TSUBAME4 points (hereinafter referred to as “points”) used for the use of TSUBAME4.0.
2. The price of points shall be as specified in Appended Table 1.
3. Points may be purchased on a group basis. In such cases, when the group includes users with different eligibility categories, the highest unit price category applicable within the group shall be applied.

Article 3 (Use of Points)  
Points shall be used for computation nodes, group storage, and applications specified in Appended Tables 6 and 7.

Article 4 (Points for Use of Computation Nodes)

1. Points to be used for computation nodes shall be calculated according to the formulas specified for each category in Appended Table 2. Any fraction less than 1/10,000 of a point shall be rounded down.
2. Coefficients used in the calculation formulas shall be as specified in Appended Tables 3, 4, and 5 according to resource type, priority, and reservation timing, respectively.

Article 5 (Points for Use of Group Storage)

1. Points for group storage consisting of hard disk drives shall be 0.5 points per TB per month. The number of files (i-nodes) that can be stored shall be 2,000,000 per TB.
2. Points for group storage consisting of SSDs shall be 0.2 points per 100 GB per month. The number of files (i-nodes) that can be stored shall be 200,000 per 100 GB.

Article 6 (Points for Use of Applications)

1. Points for the use of applications specified in Appended Table 6 on TSUBAME4.0 (per user per month) shall be as specified in that table.
2. Points for the use of applications specified in Appended Table 7 using TSUBAME4.0 licenses on computers in laboratories on campus (per laboratory per fiscal year) shall be as specified in that table.

Article 7 (Validity Period of Points)  
Points shall be valid only during the fiscal year in which they are purchased.

Supplementary Provisions

These rules shall come into force on December 20, 2024, and shall apply retroactively from October 1, 2024.

Supplementary Provision (Partially revised on March 7, 2025)  
These rules shall come into force on April 1, 2025.

Supplementary Provision (Partially revised on March 24, 2026)  
These rules shall come into force on April 1, 2026.

Appended Table 1 (related to Article 2)

User Eligibility (as defined in Article 4 of the Detailed Rules)Unit PricePoints
Article 4, Paragraph 1, Item (1)1,000 JPY16
Article 4, Paragraph 1, Item (2)
Article 4, Paragraph 1, Item (3)
Article 4, Paragraph 1, Item (6)
Article 4, Paragraph 1, Item (4)1,000 JPY5
Article 4, Paragraph 1, Item (5)
Article 4, Paragraph 1, Item (7)100,000 JPY (tax excluded)400
Article 4, Paragraph 4 (Results disclosed)
Article 4, Paragraph 4 (Results undisclosed)400,000 JPY (tax excluded)400

Appended Table 2 (related to Article 4)

CategoryFormula
Pay-as-you-goPoints per job =
(Number of nodes × Resource type coefficient × Priority coefficient × (0.7 × max(actual runtime (sec), 300) + 0.1 × requested runtime (sec))) ÷ 3,600
Interactive Job QueuePoints per job =
(Number of nodes × 0.100 × actual runtime (sec)) ÷ 3,600
However, users defined in Article 4, Paragraph 1, Items (1) and (2) are exempt from charges.
Node ReservationPoints per reservation =
Number of nodes × Reservation timing coefficient × Resource type coefficient × Reserved time (hours)
Flat-rateMonthly points =
Number of nodes × 10.0 × 30 × 24

Appended Table 3 (related to Article 4)

Resource TypeProvided ResourcesCoefficient
node_f1 node (CPU 192 cores, 4 GPUs)1.000
node_h1/2 node (CPU 96 cores, 2 GPUs)0.500
node_q1/4 node (CPU 48 cores, 1 GPU)0.250
node_o1/8 node (CPU 24 cores, 1/2 GPU)0.125
gpu_1CPU 8 cores, 1 GPU0.200
gpu_hCPU 4 cores, 1/2 GPU0.100
cpu_160CPU 160 cores0.600
cpu_80CPU 80 cores0.300
cpu_40CPU 40 cores0.150
cpu_16CPU 16 cores0.060
cpu_8CPU 8 cores0.030
cpu_4CPU 4 cores0.015

Appended Table 4 (related to Article 4)

PriorityCoefficient
01.00
12.00
24.00

Appended Table 5 (related to Article 4)

Reservation TimingCoefficient
April–SeptemberOctober–March
Up to 14 × 24 hours before slot start2.505.00
After 14 × 24 hours before slot start and up to 24 hours before1.252.50
Less than 24 hours before slot start5.0010.00

Appended Table 6 (related to Article 6)

ApplicationPoints (per user per month)
ANSYS16.0
ABAQUS / ABAQUS CAE12.0
AMBER8.0
Materials Studio / Discovery Studio24.0
Mathematica16.0
COMSOL Multiphysics16.0
Gaussian / GaussView8.0
Schrödinger Small-Molecule Drug Discovery Suite24.0

Appended Table 7 (related to Article 6)

ApplicationPoints (per laboratory per fiscal year)
ANSYS800
Materials Studio / Discovery Studio1200
Gaussian / GaussView400