Access
Durham’s hardware testbeds are integrated into different clusters on site, and these clusters differ in the way how you access them. This is due to funding constraints, access constraints, and prototyping considerations – a lot of these nodes are bought as evaluation prototypes.
For each and every system access request, we will evaluate access requests on a case-by-case base. As a rule of thumb, we grant access to Durham members, UK academics, PostDocs, PhDs and RSEs, and we try to accommodate collaborations with international research partners.
The systems can be used free of charge. However, we ask people to refrain from running very long production jobs, unless they have explicitly asked for permission to do so. Different clusters are acquired through different funding. We usually give members of the underlying research consortia higher priority.
COSMA
All testbeds plugged into COSMA use the same login nodes, the same file system, and the same software environment. To acquire access, you have to register for a COSMA account following the instructions here. You will require a specific project code to access specific hardware:
| System | Project | Remark |
| Grace Hopper | do016 | |
| Ponte Vecchio | do017 | |
| AMD MI200 | do018 |
Once you have a valid SAFE account, you can sign up to different projects. Ensure you pick the right project when you submit your SLURM script.
Documentation and Access
Access to COSMA is realised through ssh. Use the cosma8 login nodes by default. This is an AMD architecture. Depending on the testbed node that you want to use, you have to log into this type of architecture first before you compile the code. Otherwise, the compiler might produce executables for the wrong kind of machinery. An extensive documentation of the system and its usage can be found online.
Funding and acknowledgements
The acknowledgement text to be used in publications is here.
NCC
NCC is managed by the Departmen of Computer Science, and generally available to Durham users only. The registration is informal via email and details are provided on their webpage. Please consult the acknowledgment section which should be mentioned in any paper or presentation to document the support through the machine.
Bede
Preview/test nodes in Bede can only be accessed if you are eligible for a Bede account. The system is funded by the N8 CIR.
Documentation and Access
Documentation of the system is collected on a dedicated webpage. Test nodes are available through dedicated SLURM queues.
Funding and acknowledgement
This work made use of the facilities of the N8 Centre of Excellence in Computationally Intensive Research (N8 CIR) provided and funded by the N8 research partnership and EPSRC (Grant No. EP/T022167/1). The Centre is co-ordinated by the Universities of Durham, Manchester and York.