Submit and manage SLURM-based MDClaw workloads on HPC clusters.
npx skills add https://github.com/matsunagalab/mdclaw --skill hpc-run
Please help me install this Agent Skill. Skill: HPC Run Source: https://github.com/matsunagalab/mdclaw/tree/main/skills/hpc-run Command: npx skills add https://github.com/matsunagalab/mdclaw --skill hpc-run
Coordinating efficient SLURM submissions for the MDClaw workflow on HPC clusters, handling login-node interactions, array jobs, and checkpoint extensions.
Create an eq node and submit a corresponding prod job on the target cluster using mdclaw.
High-intent search queries and answers about installing and using this skill.
Submitting MDClaw workloads to a SLURM cluster requires SLURM tooling like sbatch and a prepared MDClaw job directory with topo inputs and optional container configuration to manage single-node runs and job arrays.
Yes, Singularity container integration is supported to streamline GPU-enabled SLURM job submissions, allowing you to execute MDClaw workflows with proper environment isolation and policy validation on HPC clusters.
Managing job arrays and checkpoint restarts for MDClaw simulations involves tracking DAG state and job metadata, enabling per-entry failure isolation and chained workflows across multiple systems for robust HPC campaign execution.
Running MDClaw workflows on HPC requires SLURM tooling including sinfo for cluster discovery, sbatch for job submission, squeue for tracking, and sacct for accounting to validate policies and manage job state.
Chaining eq and prod nodes in a SLURM-based MDClaw workflow is achieved by creating an eq node and submitting a corresponding prod job on the target cluster, tracking DAG state to ensure proper execution order.
Yes, multi-system campaign submissions are supported, providing cluster discovery, policy validation, and per-entry failure isolation to coordinate efficient SLURM submissions across different HPC environments for MDClaw.
yale-som-hpcManage Slurm jobs on the Yale SOM HPC cluster with sbatch, squeue, sacct, scancel, and scontrol.
KwongFukMonitor Slurm sbatch jobs via sacct and squeue, apply minimal repairs, and resubmit failures.
dongzhuoyaoConfigure Slurm and ROCm environments for PyTorch jobs on LUMI.
DicklesworthstoneExecute SSH commands, transfers, and tunneling for remote server administration.
