Dawn - Intel GPU (PVC) Nodes¶
These new nodes entered Early Access service in January 2024
Dawn Documentation¶
AIRR¶
Welcome to Cambridge Research Computing Access Portal
The Cambridge Research Computing Access Portal is a web-based platform developed by the University of Cambridge to provide secure, streamlined access to DAWN, one of the UK’s national AI Research Resource (AIRR) systems. It operates under the broader UKRI-supported AI Research Resource Federation (AIRRFED) initiative. A parallel facility, Isambard-AI, is hosted at the University of Bristol, forming the other half of the national AIRR infrastructure.
The portal enables researchers to:
- Request access to DAWN’s compute resources,
- Manage project membership and roles, and
- View and report on usage statistics.
This documentation provides a comprehensive guide to navigating the system, understanding its features, and utilizing its full capabilities.
Overview of AIRRFED
AIRRFED aims to provide secure, federated access to national AI and HPC infrastructures. This project is a collaborative effort involving multiple institutions, with Dawn at the University of Cambridge and Isambard-AI at the University of Bristol serving as key sites.
Authentication is federated via MyAccessID, allowing users to log in using their institutional credentials. Authorization and resource allocations are managed centrally through Waldur, which is integrated with the Atlassian Service Desk (JIRA) to automate access provisioning and operational workflows.
Principal Investigators (PIs) affiliated with UK universities can submit compute access requests to UKRI. Upon approval, projects are allocated to one of the two national AIRR sites, Cambridge (Dawn) or Bristol (Isambard-AI), based on suitability and availability.
Getting Access
Access to DAWN are allocated by bodies external. Prospective users will need to apply via these routes:
Hardware¶
The Dawn (PVC) nodes are:
- 256 Dell PowerEdge XE9640 servers
each consisting of:
- 2x Intel(R) Xeon(R) Platinum 8468 (formerly codenamed Sapphire Rapids) (96 cores in total)
- 1024 GiB RAM
- 4x Intel(R) Data Center GPU Max 1550 GPUs (formerly codenamed Ponte Vecchio) (128 GiB GPU RAM each)
- Xe-Link 4-way GPU interconnect within the node
- Quad-rail NVIDIA (Mellanox) HDR200 InfiniBand interconnect
and each PVC GPU contains two stacks (previously known as tiles) and 1024 compute units.