Using HEP experiment workflows for the benchmarking and accounting of WLCG computing resources

Abstract

Benchmarking of CPU resources in WLCG has been based on the HEP-SPEC06 (HS06) suite for over a decade. It has recently become clear that HS06, which is based on real applications from non-HEP domains, no longer describes typical HEP workloads. The aim of the HEP-Benchmarks project is to develop a new benchmark suite for WLCG compute resources, based on real applications from the LHC experiments. By construction, these new benchmarks are thus guaranteed to have a score highly correlated to the throughputs of HEP applications, and a CPU usage pattern similar to theirs. Linux containers and the CernVM-FS filesystem are the two main technologies enabling this approach, which had been considered impossible in the past. In this paper, we review the motivation, implementation and outlook of the new benchmark suite.

Design and engineering of a simplified workflow execution for the MG5aMC event generator on GPUs and vector CPUs

“…Finally, we believe that our implementations of ME calculations in CUDA (and eventually other languages) for GPUs and in C++ for vector CPUs represent very useful software workloads for the benchmarking of computing resources used by the HEP experiments, in an increasingly heterogeneous environment. In this context, we are collaborating with the HEPiX benchmarking working group, with the goal of eventually providing a MG5aMC-based containerized standalone application to be integrated in the HEP-Benchmarks suite [67,68]. We also plan to port our code to new architectures, such as AMD and Intel GPUs but also non-x86 CPUs such as ARM and Power9, also in view of the use of the software at HPC facilities.…”

Design and engineering of a simplified workflow execution for the MG5aMC event generator on GPUs and vector CPUs

“…Finally, we believe that our implementations of ME calculations in CUDA (and eventually other languages) for GPUs and in C+ for vector CPUs represent very useful software workloads for the benchmarking of computing resources used by the HEP experiments, in an increasingly heterogeneous environment. In this context, we are collaborating with the HEPiX benchmarking working group, with the goal of eventually providing a MG5aMC-based containerised standalone application to be integrated in the HEP-Benchmarks suite [66,67]. We also plan to port our code to new architectures, such as AMD and Intel GPUs but also non-x86 CPUs such as ARM and Power9, also in view of the use of the software at HPC facilities.…”

Celeritas: GPU-accelerated particle transport for detector simulation in High Energy Physics experiments

“…The Summit supercomputer at the OLCF alone has a total of 27 648 GPUs that, using the aforementioned factor of 160, represents the equivalent of 4 423 680 CPU cores. For comparison purposes, in 2017 the Worldwide LHC Computing Grid (WLCG) encompassed approximately 500 000 CPU cores [38]. Incorporating DOE's network of LCFs will bring the total HEP computing resources to an unprecedented level and significantly alleviate current CPU bottlenecks.…”