Build a Performance Model for a Molecular-Dynamics Job
Overview
What this challenge is about.
Build an analytical performance model covering: compute time per step (function of atom count + cutoff + interaction type), inter-rank communication cost (function of decomposition + halo size), parallel I/O cost. Calibrate with a benchmark sweep: 4 system sizes x 5 rank counts x 2 node configurations (24 runs). Validate model predictions against held-out runs. Build a small CLI tool (Python) that takes simulation parameters and recommends rank + node count. Deliver source code, the calibration dataset (CSV), the CLI tool, and a 10-page writeup.
The Brief
What you'll do, and what you'll demonstrate.
Build a calibrated analytical performance model for GROMACS MD runs and a CLI tool that recommends optimal rank + node count for a given simulation.
Earning criteria — what you'll demonstrate
- Derive an analytical performance model from first principles
- Design a controlled benchmark sweep that supports model calibration
- Validate model predictions against held-out experimental runs
- Translate a performance model into an end-user CLI tool
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Career mappings coming soon.