supercomputing

A lightweight (serverless) native python parallel processing framework based on simple decorators and call graphs.

The LUMI AI Guide is designed to assist users in migrating their machine learning applications from smaller-scale computing environments to the LUMI supercomputer.

qtop (pronounced queue-top) is a tool written in order to monitor the state of Queueing Systems, along with related information relevant on HPC & related clusters. It supports **SLURM, PBS, SGE & OAR** families, up to 10000s of nodes. It is the evolution of the historic shell version of the tool, born out of CERN's LHC project needs:

A science inference cluster — every model, from protein folding to LLMs, behind one OpenAI/Anthropic-compatible endpoint and a single key. Built to run beside HPC so agentic AI researchers can drive simulations and model inference together. Self-deploying on RKE2 with KServe/Knative + HAMi GPU.

An HPC abstraction over MPI with built-in parallel compute primitives

A scheduler for resource-aware parallel external computing on clusters

🧠 HPC job scheduling simulation & queue optimization framework. Compares FCFS and Backfill efficiency using 2D Gauss and Poisson workload models.

Command-line tool for benchmarking HPC hardware performance

DVFS techniques

A command line password cracking tool which works on a cluster(HPC).

Run MPI (sbatch) tasks remotely on NSG and SSH with ease.

Xùn ☴

Ansible playbooks & scripts for the installation & maintenance of a Slurm based cluster at MGHPCC hosted by UMass Amherst RC

SixRackUnits (6RU) delivers short form news and educational content around AI/HPC hardware.

A physical distribution architecture for the Distributed Exact Generalized Grover Algorithm (DEGGA) using the CUNQA emulator and virtual quantum communication.

Repo from graduate data science course on HPC with Lustre/SLURM

Metadata and scripts accompanying the Year in the Life of a Parallel File System paper

SPMD (HPC) collectives library for Python implemented using pyzmq

University project where a command line software called Ale was developed. Ale runs molecular dynamics simulations, at lowest effort. It was designed to be user friendly with config-files specifying simulations. The simulations can be run on e.g. supercomputers in parallel to achieve high throughput when simulating material properties.

Fotireps is a user-friendly conglomerate of different data processing tools used by the MWA EoR group. Fotireps is made to be highly modularized and robust with each step being runnable in isolation but can also perform standard end to end EoR analysis on MWA observations.