Skip to content

Feat/forecast inference dataset creation from superjuice#18

Merged
khintz merged 32 commits intodmidk:mainfrom
leifdenby:feat/forecast-inference-dataset-creation-from-superjuice
Jan 27, 2026
Merged

Feat/forecast inference dataset creation from superjuice#18
khintz merged 32 commits intodmidk:mainfrom
leifdenby:feat/forecast-inference-dataset-creation-from-superjuice

Conversation

@leifdenby
Copy link
Copy Markdown
Contributor

Add files I forgot to commit on superjuice

leifdenby and others added 30 commits May 14, 2025 12:35
…feat/forecast-inference-dataset-creation
…feat/forecast-inference-dataset-creation
- expose container build program with env var so we can use docker on
  DGX Spark
- use nvidia docker registry for ARM nvidia image on ARM platforms
- install with pip directly into system python site-packages (with
  overwrite) while setting torch version constraint (otherwise
  dependencies install overwrites the pytorch install)
- optionally use pip directly inside container (rather than via uv) to
  ensure we use system python without venv
Use pytorch base image with CUDA>=12.8 (we'll need the same for DGX Spark), update README with instructions to handle missing podman configuration for CUDA devices pass-through and default to not using `uv` in container.
…feat/forecast-inference-dataset-creation-from-superjuice
Copy link
Copy Markdown
Contributor Author

@leifdenby leifdenby left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I've read through this all, LGTM of course :)

@khintz over to you :)


## Building image and running inference

Currently building the image and running inference is only supported on the "superjuice" machine (`27sj894.dmi.dk`).
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Would you not expect it work on one of the sparks?


In a rootless Podman setup without system-level NVIDIA integration:

- --gpus all is a no-op
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

no-op ?
Does this meaning it really does nothing ?

@khintz
Copy link
Copy Markdown
Contributor

khintz commented Jan 27, 2026

I think this looks good. Still a lot of work needed, but let's iterate on that as we proceed.
I'll suggest merging as we are working with non-production code currently.

@khintz khintz merged commit ba7a52a into dmidk:main Jan 27, 2026
4 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants