Speaker
Description
As part of CERN-GEANT-PRACE-SKA collaboration and in the context of EGI-ACE (Advanced Computing for the European Open Science Cloud ) collaborators are working towards enabling
efficient HPC use for Big Data sciences. Approaching HPC site with High Throughput
Computing (HTC) workloads presents unique challenges in areas concerning data
ingress/egress, use of shared storage systems, and efficient use of an increasing variety of heterogeneous computing resources.
We present work enabling the benchmarking of HEP workloads on heterogeneous resources found at HPC centers (in collaboration with the HEPiX Benchmarking working group), as well as initial investigations into the topic of data access. We share the development and testing of a benchmark for HPC shared storage systems which may be used as a metric to guide job
scheduling policy.
| Speaker time zone | Compatible with Europe |
|---|
