spack-site

Please fill in the stack_conf.json with:

stack_root: directory where the stack will be installed
compiler: compiler and version to use (spack syntax), e.g., gcc@11.3.0
buildcache: directory to be used as a build cache
spack_version: version of spack to use

Boostrap stack

source setup-env.sh
bash bootstrap.sh`

This script will install the compiler and its dependencies. It will also configure spack.

Environment installation

All the software stack is separated into different spack environments. Each environment has its own purpose but all them generate modules files in the same place.

list of environments

base: very basic environment with build tools. This is the first environment that should be installed
std: it contains all programming languages compiler and interpreters as well as imaging and math libs.
hpc: it provides parallel software and parallel runtimes.

creating an environment

For creating a new environment, please follow the following steps:

execute create_env_scrip.py $env_name
Add packages on the right section using spack add $package
Add configuration for modules if needed
Commit the new script

CI process

The CI is done by:

https://sc-jenkins.dcsr.unil.ch/

Debug process

If you need to debug a package installation, you can use the script spack_init.sh which can be executed with srun:

srun --constrain zen2 -t 6:00:00 --qos dcsr -c 8 --pty bash --rcfile spack_init.sh

This script will perform a spack chain based on the current stack. This means that the installation of the package is done using the current packages installed on the cluster.

TODO

be able to install several versions of R or be sure that devtools can be installed from scratch.

Scientific-Computing/spack-site

About

Resources

Stars

Watchers

Forks

Releases

Packages 0

Contributors 3

Languages

Packages