search

E4S-Project / e4s

E4S for Spack
https://e4s.readthedocs.io
MIT License
28 stars 13 forks source link

Missing 22.05 environment YAML for general x86_64 #69

Closed chuckatkins closed 2 years ago

chuckatkins commented 2 years ago

The published environments for v22.05 in environments/22.05 are for GPU builds with CUDA, ROCm, and DPC++. Is a baseline environment YAML available for the CPU-only installations? Furthermore, is a docker image even available without any of the GPU stacks enabled?

eugeneswalker commented 2 years ago

Is a baseline environment YAML available for the CPU-only installations?

There is not. The CPU only install would be the same as the CUDA or ROCm environment, minus the ROCm specs, which are clearly labeled in the environment. You would just delete those lines. The package preferences and compiler definitions would remain the same.

Furthermore, is a docker image even available without any of the GPU stacks enabled?

Yes, the Docker container image we used for our containerized builds was:

This image has GCC 9.4.0, has no GPU stack pre-installed, and is what we use for our containerized builds.

eugeneswalker commented 2 years ago

The published environments for v22.05 in environments/22.05 are for GPU builds with CUDA, ROCm, and DPC++. Is a baseline environment YAML available for the CPU-only installations? Furthermore, is a docker image even available without any of the GPU stacks enabled?

Is a baseline environment YAML available for the CPU-only installations?

There is not. The CPU only install would be the same as the CUDA or ROCm environment, minus the ROCm specs, which are clearly labeled in the environment. You would just delete those lines. The package preferences and compiler definitions would remain the same.

For instance, taking 22.05/cuda-x86_64.spack.yaml:

eugeneswalker commented 2 years ago

Closing this per follow up with @chuckatkins via Slack.