System Overview
Super computing System for computational Materials Science
The high-performance supercomputing system of CCMS specializes in materials science simulations and can satisfy diverse needs in computational materials science research. It consists mainly of HPE Cray XD220v and HPE Cray XD670 with GPU accelerator, and its total theoretical peak performance is 4 PFLOPS. This system provides the domestic and foreign researchers in the field of materials science with various application software for materials design.
Nickname and Front Panel Design of Supercomputing System
Nickname: MASAMUNE-II
MAterials science Supercomputing system for Advanced MUlti-scale simulations towards NExt-generation II
Nickname “MASAMUNE-II” has been given to the supercomputing system of CCMS, continuing the nickname of the previous system “MASAMUNE-IMR”, in honor of “Masamune Date”, the first lord of Sendai in the 17th century, who dispatched the sailing ship “San Juan Bautista” towards the world from Sendai in 1613. The nickname “MASAMUNE-II” expresses our hope that creative and innovative achievements of the advanced multi-scale simulations on materials science by our supercomputing system will give great impacts towards the next-generation from Sendai to the World. “Masamune Date” is drawn on the front panel of our supercomputer by Sumi-E Artist OKAZU with the brave letter “弐” meaning “second” in Japanese Kanji. This “Masamune Date” on the front panel is looking ahead to the future advances in the materials science of the world from Sendai.
System configuration
MASAMUNE-II consists of the Supercomputer (HPE Cray XD220v, HPE Scale-up Server 3200, and HPE Cray XD670). It is integrated by Hitachi, Ltd. The system specifications are shown below.
| System Name | Supercomputer | ||
|---|---|---|---|
| Model | HPE Cray XD220v | HPE Scale-up Server 3200 | HPE Cray XD670 |
| total nodes | General node : 120 |
2 nodes |
11 nodes |
| CPU |
|
Intel Xeon Platinum 8490H
|
Intel Xeon Platinum 8480+
|
| Accelerator | - | - |
NVIDIA H100 80GB SXM5
|
| memory size | 512 GiB/node 2.0 TiB/node |
4.0 TiB/node | 1.0 TiB/node |
| Theoretical peak performance | 4.052 PFLOPS (CPU : 1.104 PFLOPS, GPU : 2.948 PFLOPS) |
||
