Personal tools
University Computing Service

Scientific Computing

CamGrid

CamGrid is a distributed computing resource based on the Condor middleware, and provides a powerful computational tool for participating university members. These pages describe the facilities available and how Cantabrigians can gain access to them.

CamGrid was conceived by a number of departments and groups within the university who federated their computational facilities, and now encompasses these groups. The resources could be desktops or dedicated machines, possibly behind different firewalls, and using public or private IP addresses (though Condor daemons are forced to utilise CUDN-only routeable IP addresses). CamGrid mainly caters for serial (single-threaded) applications, but can also handle "small" parallel/MPI jobs. This model has been running since January 2005, with each institution maintaining its own Condor pool, so the grid consists of flocked pools. CamGrid is coordinated by the University Computing Service.

Documentation and Tools

  • Technical details of CamGrid can be found here. These include an architectural overview, examples of machine configuration and various tutorials for using different aspects of the grid.
  • University members interested in using CamGrid can invesitgate gaining access by approaching the right contact.
  • Early experiences of setting up CamGrid using a VPN are described in this paper. Note that CamGrid no longer uses a VPN, but RFC 1918 (known locally as CUDN-only routeable) addresses as described in the technical details section.
  • Manual for Condor v7.6. This is a great place to start learning about Condor. A local copy of the manual (with an added search facility), accessible only from within Cambridge University, can be found here.

 

Colossal

Example of work carried out on CamGrid

Electronic structure calculations, carried out on CamGrid and involving sweeps of several MPI calculations, have helped explain why the framework material Ag3[Co(CN)6], shown in the figure, exhibits colossal positive and negative thermal expansion. The low temperature structure is on the left, with the high temperature structure on the right.


These findings have been published in Science, 319, p794-797, 8 February (2008).