Speaker
Description
For over two decades, the LHC@home volunteer computing project has provided additional opportunistic computing capacity to support the scientific research conducted at CERN. With the retirement of the SixTrack application, the only natively executable one, there has been a significant reduction in job throughput. This paper highlights the difference in job throughput between the native and virtualized approaches, the VM Capacity Gap, which arises from the friction introduced by the additional steps required to support virtualization.
We describe two new developments aimed at closing this gap and restoring throughput: XTrack, a new native application designed to replace SixTrack, and a container-based solution from the BOINC project, which has been piloted with the Theory application. Results from the initial deployments of both approaches are presented, demonstrating their potential to close this gap. It is also shown how the container-based approach can simplify the adoption of volunteer computing for applications and scientists within HEP and beyond.
Together, these developments help revitalize LHC@home and place it on a solid path for the future.