Skip to main content

dagster-slurm

The Super developer experience for HPC

server_cluster

Unified Compute Interface

Develop on your laptop, deploy to an HPC cluster. The ComputeResource provides a single, consistent API for local and remote execution, so your code never has to change. Dependencies are consistently managed and distributed automatically via pixi

HPC-Optimized Execution

Go beyond simple job submission. Use multi-assets to fuse multiple assets into a single Slurm allocation, minimizing queue times.

monitor

Extensible & Environment-Aware

Built-in Launchers for Ray and Spark (WIP) enable distributed workloads out of the box. Automatic environment packaging with pixi ensures that your remote execution environment is a perfect replica of your local one.