The University of Osaka’s D3 Center is pleased to announce the launch of new GPU nodes equipped with the latest high-performance NVIDIA H200 GPUs on the mdx II virtual server-based computing platform, operated as part of the Data-Driven Social Transformation Platform Consortium.

Like commercial cloud services, mdx II allows users to access virtual servers and storage resources with performance levels tailored to their needs, for only as long as required, and at affordable rates. These new GPU nodes provide a flexible and powerful computing environment ideal for AI, machine learning, large-scale simulations, and other research requiring intensive computational processing.

Overview of mdx II GPU Nodes

Total: 7 nodes (Total theoretical performance: 982.1 TFLOPS)

Specifications per 1 node
CPU	Intel Xeon Gold 6530（32 core x 2）
GPU	NVIDIA H200 x 4
Memory	1TB（1,024GB）
Theoretical performance	140.3 TFLOPS

 

Key Features of mdx II (Compared to Traditional Supercomputers)

• - Virtual server-based platform: Users can freely build customized virtual environments as needed
• - Granular GPU usage: GPUs can be used on a per-unit basis, offering flexible resource allocation
• - Monthly exclusive use: Avoid job queueing and use resources on a predictable schedule, similar to commercial cloud platforms

Inquiries and Access

If you are interested in using mdx II or have any questions, please feel free to contact us via this form.
 

More Information

•   System Configuration
•   Pricing
•   Process from Application to Start of Use
•   Brochure (PDF)

We hope mdx II will be a valuable resource in supporting your research activities. Thank you for your continued interest and support.