Cargando…
Profiling CPU-bound workloads on Intel Haswell-EP platforms
With the increasing adoption of public and private cloud resources to support the demands in terms of computing capacity of the WLCG, the HEP community has begun studying several benchmarking applications aimed at continuously assessing the performance of virtual machines procured from commercial pr...
Autores principales: | , , |
---|---|
Lenguaje: | eng |
Publicado: |
2017
|
Materias: | |
Acceso en línea: | http://cds.cern.ch/record/2257973 |
_version_ | 1780953820199649280 |
---|---|
author | Guerri, Marco Giordano, Domenico Cristovao, Cordeiro |
author_facet | Guerri, Marco Giordano, Domenico Cristovao, Cordeiro |
author_sort | Guerri, Marco |
collection | CERN |
description | With the increasing adoption of public and private cloud resources to support the demands in terms of computing capacity of the WLCG, the HEP community has begun studying several benchmarking applications aimed at continuously assessing the performance of virtual machines procured from commercial providers. In order to characterise the behaviour of these benchmarks, in-depth profiling activities have been carried out. In this document we outline our experience in profiling one specific application, the ATLAS Kit Validation, in an attempt to explain an unexpected distribution in the performance samples obtained on systems based on Intel Haswell-EP processors. |
id | cern-2257973 |
institution | Organización Europea para la Investigación Nuclear |
language | eng |
publishDate | 2017 |
record_format | invenio |
spelling | cern-22579732019-09-30T06:29:59Zhttp://cds.cern.ch/record/2257973engGuerri, MarcoGiordano, DomenicoCristovao, CordeiroProfiling CPU-bound workloads on Intel Haswell-EP platformsComputing and ComputersWith the increasing adoption of public and private cloud resources to support the demands in terms of computing capacity of the WLCG, the HEP community has begun studying several benchmarking applications aimed at continuously assessing the performance of virtual machines procured from commercial providers. In order to characterise the behaviour of these benchmarks, in-depth profiling activities have been carried out. In this document we outline our experience in profiling one specific application, the ATLAS Kit Validation, in an attempt to explain an unexpected distribution in the performance samples obtained on systems based on Intel Haswell-EP processors.CERN-IT-Note-2017-002oai:cds.cern.ch:22579732017-04-03 |
spellingShingle | Computing and Computers Guerri, Marco Giordano, Domenico Cristovao, Cordeiro Profiling CPU-bound workloads on Intel Haswell-EP platforms |
title | Profiling CPU-bound workloads on Intel Haswell-EP platforms |
title_full | Profiling CPU-bound workloads on Intel Haswell-EP platforms |
title_fullStr | Profiling CPU-bound workloads on Intel Haswell-EP platforms |
title_full_unstemmed | Profiling CPU-bound workloads on Intel Haswell-EP platforms |
title_short | Profiling CPU-bound workloads on Intel Haswell-EP platforms |
title_sort | profiling cpu-bound workloads on intel haswell-ep platforms |
topic | Computing and Computers |
url | http://cds.cern.ch/record/2257973 |
work_keys_str_mv | AT guerrimarco profilingcpuboundworkloadsonintelhaswellepplatforms AT giordanodomenico profilingcpuboundworkloadsonintelhaswellepplatforms AT cristovaocordeiro profilingcpuboundworkloadsonintelhaswellepplatforms |