• Production system
• In service
• Regional system for Wales
• Funded by ERDF, Welsh Government
• 6 nodes, with 8 NVIDIA A100 40GB accelerators per node
• Manufactured by Atos
• Scheduler: Slurm
• Interconnects: Infiniband HDR, NVLink

AccelerateAI

AccelerateAI is a GPU cluster hosted at Swansea University. Comprising 6 x Atos BullSequana nodes, each with 8 x NVIDIA A100 GPUs, it serves the machine learning needs of Wales. It forms part of the Supercomputing Wales infrastructure.

GPUs are linked by NVLink in pairs. Using more than 2 GPUs in parallel requires data transfer over the system bus.

Documentation

• https://sa2c.github.io/AccelerateAI
• https://portal.supercomputing.wales

Gaining access

Access is restricted to members of the Supercomputing Wales partner institutions and their external collaborators. Access is initially via a user account and project request to the MySCW Service, where access to AccelerateAI should be mentioned explicitly.

Restrictions

By default, jobs may use up to 4 GPUs. Runs with more than 4 GPUs may be enabled when evidence of scalability of a project’s work to 4 GPUs is demonstrated, since the system is not designed for scalability to large GPU counts, and many problems will not scale well.

Jobs may run for up to 48 hours.