Getting started Betzy
This page is a quick intro for the projects invited to participate in the pilot phase and relevant information will be eventually merged into main documentation.
Access during pilot
To be added.
Please use the regular support channel to submit requests and mark the case with Betzy.
Betzy uses the SLURM queuing system. For more information the queue system on Betzy, please check out job types page.
NOTE: During the pilot phase there will be no CPU core limitations imposed.
For generic information about job submission, please check out our documentation here.
Betzy uses a DDN powered 2.5 PB Lustre parallel file system with 51 GB/s bandwidth and 500k+ metadata operations per second.
The Lustre file system is mounted under
Betzy: NO BACKUP
Please note that NO BACKUP is taken during the pilot phase.
The `/cluster` file system might be reformatted at the end of the pilot phase, before machine being placed into production.
NIRD project file systems -
/trd-project[1-4] - will be mounted on the Betzy
login nodes, similarly to Fram and Saga HPC clusters.
As on Fram and Saga, scientific software on Betzy will be installed using the EasyBuild system, and the Lmod modules tool will be used for changing environment setup via modulefiles.
The two common toolchains
intel will be installed on Betzy.
- GCC compilers (
- Open MPI library
- OpenBLAS (including LAPACK) + ScaLAPACK
- FFTW library
- Intel compilers (
- Intel MPI library
- Intel MKL library (including BLAS, LAPACK, ScaLAPACK, FFT)