Schedule¶
| DAY 1 - Tuesday 04/02 | |
|
09:00 EET
08:00 CET |
Welcome and Introduction
Presenters: Jørn Dietze (LUST) |
|
09:15 EET
08:15 CET |
Introduction to LUMI
Presenter: Jørn Dietze (LUST) |
|
09:45 EET
08:45 CET |
Using the LUMI web interface
Presenters: Mats Sjöberg (CSC) and Lukas Prediger (CSC) |
|
10:05 EET
09:05 CET |
Hands-on: Run a simple PyTorch example notebook |
|
10:35 EET
09:35 CET |
Break (15 minutes) |
|
10:50 EET
09:50 CET |
Your first AI training job on LUMI
Presenters: Mats Sjöberg (CSC) and Lukas Prediger (CSC) |
|
11:20 EET
10:20 CET |
Hands-on: Run a simple single-GPU PyTorch AI training job |
|
12:05 EET
11:05 CET |
Lunch break (45 minutes) |
|
13:05 EET
12:05 CET |
Understanding GPU activity & checking jobs
Presenter: Luca Stanisic (AMD) |
|
13:25 EET
12:25 CET |
Hands-on: Checking GPU usage interactively using rocm-smi |
|
13:45 EET
12:45 CET |
Running containers on LUMI Presenter: Gregor Decristoforo (LUST) |
|
14:05 EET
13:05 CET |
Hands-on: Pull and run a container |
|
14:25 EET
13:25 CET |
Break (15 minutes) |
|
14:40 EET
13:40 CET |
Building containers from conda/pip environments Presenter: Jørn Dietze (LUST) |
|
15:00 EET
14:00 CET |
Hands-on: Creating a conda environment file and building a container using cotainr |
|
15:20 EET
14:20 CET |
Extending containers with virtual environments for faster testing Presenter: Gregor Decristoforo (LUST) |
|
15:40 EET
14:40 CET |
Hands-on: Getting started with your own project
Bring your own AI code, you want to run on LUMI, and spent some time applying what you have learned during the workshop - with on-site support from LUST/AMD. |
|
17:00 EET
16:00 CET |
End of the course day |
| DAY 2 - Wednesday 05/02 | |
|
09:00 EET
08:00 CET |
Scaling AI training to multiple GPUs
Presenters: Mats Sjöberg (CSC) and Lukas Prediger (CSC) |
|
09:40 EET
08:40 CET |
Hands-on: Converting the PyTorch single GPU AI training job to use all GPUs in a single node via DDP |
|
10:20 EET
09:20 CET |
Break (15 minutes) |
|
10:35 EET
09:35 CET |
Extreme scale AI
Presenter: Paul Bauer (AMD) |
|
11:20 EET
10:20 CET |
Demo/Hands-on: Using multiple nodes |
|
11:35 EET
10:35 CET |
Loading training data on LUMI
Presenter: Harvey Richardson (AMD) |
|
12:00 EET
11:00 CET |
Lunch break (45 minutes) |
|
13:00 EET
12:00 CET |
Coupling machine learning with HPC simulation Presenter: Harvey Richardson (HPE) |
|
13:30 EET
12:30 CET |
Hands-on: Advancing your own project and Q&A
Bring your own AI code, you want to run on LUMI, and spent some time applying what you have learned during the workshop - with on-site support from LUST/AMD. |
|
16:30 EET
15:30 CET |
End of the course day |