MetOffice / MetOffice/data_science_cop

create instructions and exmaple on using ECMWF GPUs

Open
#44 3 comments 0 reactions 0 assignees View on GitHub
Dominant language
Jupyter Notebook
Stars
26
Forks
10
Avg merge
7d 16m
Merged PRs (30d)
1

Description

Met Office has an allocation of compute on ECWMF HPC, which includes GPU nodes. It would be useful to explore running MLint stuff on there, specifically doing experiments using ECWMF as initial conditions (e.g. IFS HRES analysis fields)

More info on ecwmf confluence:

To get access to the ECMWF compute, you need to create an account. This is done through the Met Office representative, who is Steve Grant at email comp-rep@metoffice.gov.uk

### more info
- https://confluence.ecmwf.int/display/UDOC/HPC2020+User+Guide
- https://europeanweather.cloud/get-access
- https://www.ecmwf.int/en/about/contact-us/computing-representatives

### Definiition of done
- [ ] Get acccount on ECMWF HPC
- [ ] Get [ML Weather tutorial](https://github.com/informatics-lab/ml_weather_tutorial/) running on ECMWF HPC
- [ ] Create a "Getting started" guide for MO staff to gain access and use ECMWF HPC for training an ML model and running inference

Contributor guide

No contributing guide indexed for this repository

Research direction

Start with the ECMWF HPC2020 User Guide and the linked access pages, then request an account through the Met Office representative. Run the linked ML Weather tutorial on ECMWF HPC, including training and inference with ECMWF initial conditions. Done means a Getting Started guide explains access and those workflows for Met Office staff.

Written by the indexing model from the issue text.

Assessment

Tech stack
hpc, jupyter-notebook, machine-learning
Domain
documentation, hpc, machine-learning
Issue type
Documentation
Difficulty
4/5
Estimated time
3-5 days
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
42/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.