Säntis stack for Spack v1.1#52
Conversation
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
1 similar comment
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
1 similar comment
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
1 similar comment
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
| @@ -0,0 +1,6 @@ | |||
| default: | |||
| features: [osu-micro-benchmarks, mpi, serial, openmp] | |||
There was a problem hiding this comment.
| features: [osu-micro-benchmarks, mpi, serial, openmp] | |
| features: [osu-micro-benchmarks, mpi, serial, openmp, nccl, nccl-tests, cuda, cray-mpich] |
probably best reformatted to
| features: [osu-micro-benchmarks, mpi, serial, openmp] | |
| features: | |
| - cuda | |
| - mpi | |
| - cray-mpich | |
| - nccl | |
| - nccl-tests | |
| - openmp | |
| - osu-micro-benchmarks | |
| - serial |
| cc: mpicc | ||
| cxx: mpic++ | ||
| ftn: mpifort | ||
| activation: /user-environment/env/default/activate.sh |
There was a problem hiding this comment.
Do you know if there's a particular need for this still? If not:
| activation: /user-environment/env/default/activate.sh | |
| views: | |
| - default |
There was a problem hiding this comment.
No, I don't know honestly. @leclairm Do you know more?
| @@ -0,0 +1,4 @@ | |||
| gcc: | |||
| version: "13" | |||
There was a problem hiding this comment.
Did you already try GCC 14? Not essential, just curiuous if it broke anything (quite a few packages didn't build with it when it was first released).
There was a problem hiding this comment.
No, I did not try, but I can give it a quick try. Let's see.
| commit: releases/v1.1 | ||
| packages: | ||
| repo: https://github.com/spack/spack-packages.git | ||
| commit: develop |
There was a problem hiding this comment.
Just a reminder to pin this before finalizing the uenv. We typically do
| commit: develop | |
| commit: <sha> # develop on <date> |
| mpi: cray-mpich@9.1.0 +cuda %c,cxx,fortran=nvhpc | ||
| unify: when_possible | ||
| specs: | ||
| - python@3.12 %c,cxx=gcc ^libffi %c,cxx=gcc ^expat %c=gcc ^libmd %c=gcc ^libxcrypt %c=gcc ^gettext %c,cxx=gcc |
There was a problem hiding this comment.
| - python@3.12 %c,cxx=gcc ^libffi %c,cxx=gcc ^expat %c=gcc ^libmd %c=gcc ^libxcrypt %c=gcc ^gettext %c,cxx=gcc | |
| - python@3.13 %c,cxx=gcc ^libffi %c,cxx=gcc ^expat %c=gcc ^libmd %c=gcc ^libxcrypt %c=gcc ^gettext %c,cxx=gcc |
if there's no particular reason to stick with 3.12?
There was a problem hiding this comment.
My only reason was that 3.13 it's still in the "bugfix" phase, but I have nothing against updating. I simply wanted to stay on the safe side, since we anyway had Python 3.10 previously, so 3.12 is already an upgrade.
There was a problem hiding this comment.
Yeah, that's a fair point. And to be honest, at least for icon4py integration we might need to test a few versions just to be on the safe side that python 3.13 (or 12 or 11) doesn't actually slow things down compared to 3.10.
But I guess in this uenv python is probably only needed for some build-time scripts, not actually at runtime like with the icon4py dycore?
| # everything needed for nccl on SS11 | ||
| - nccl | ||
| - nccl-tests |
There was a problem hiding this comment.
| # everything needed for nccl on SS11 | |
| - nccl | |
| - nccl-tests | |
| # everything needed for nccl on SS11 | |
| - nccl | |
| - nccl-tests | |
| - aws-ofi-nccl |
There was a problem hiding this comment.
The versions for nccl and aws-ofi-nccl are a bit finicky, but if you add the nccl and nccl-tests features to reframe.yaml, our nccl tests will run after the uenv is built which would tell us if we need to pin the versions or not.
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
1 similar comment
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
|
cscs-ci run alps;system=santis;uarch=gh200;uenv=icon:26.2 |
No description provided.