-
Notifications
You must be signed in to change notification settings - Fork 0
Azure meeting Feb 18 2022
Kenneth Hoste edited this page Feb 18, 2022
·
4 revisions
- Overview of spent credits
- EESSI hackathon January 2022
- EUM'22
- Monitoring
- Repository for hosting datasets (e.g. for WRF)
- EESSI paper published!
- Azure support for CitC
- Bob Dröge (RUG)
- Kenneth Hoste (HPC-UGent)
- Davide Vanzo (MS Azure)
- Alan O'Cais (CECAM)
- January 2022: ~€1195
- temporary Magic Castle cluster EESSI hackathon (Jan'22): ~€610
- AMD Milan build node: ~€350
- Stratum-1 in us-east: ~€200
- February 2022 (partial): ~€112
- basically only Stratum-1 in us-east
- https://github.com/EESSI/meetings/wiki/EESSI-hackathon-Jan'22
- show & tell meeting to wrap up the hackathon
- good progress on:
- installing software on top of EESSI (experiments with LAMMPS)
- support running software on NVIDIA GPUs
- archiving EESSI software installations into a container image
- https://easybuild.io/eum22
- incl. talk by Hugo & Davide on running WRF on Azure via EESSI
- see https://easybuild.io/eum22/#azure-eessi-wrf
- good progess, work done by Terje @ Univ. of Oslo
- https://monitoring.eessi-infra.org
- see Hugo's proposal at https://github.com/EESSI/filesystem-layer/pull/112
- EESSI: A cross-platform ready-to-use optimised scientific software stack
- https://doi.org/10.1002/spe.3075
- Open access
- "Software: Practice and Experience" special issue on "New Trends in HPC"
- work done by Hugo
- resulted in PRs to CitC:
- Matt Williams (main developer of CitC) has been granted access to EESSI Azure credits for testing CitC on Azure
- discussion with Davide
- would also be relevant for Azure, business cases are popping up where GPU support in EESSI would be very useful!
- technical issues are basically figured out during EESSI hackathons (effort led by Alan)
- not being able to ship (full) CUDA toolkit is annoying, but not a blocker
- shipping CUDA runtime would not be a problem
- script is available to install what's needed: CUDA + compat libs
- currently only works on CentOS
- doesn't take into account all possible corner cases yet
- but good enough to be included in EESSI for testing?
- split up script in "install" and "test" phase
- we also want to hide GPU-aware software installations until script has been run through an Lmod plugin
- would be interesting to include CUDA samples in EESSI as a module