Höchstleistungsrechenzentrum Stuttgart: Machine Learning with AMD Instinct™ GPUs and ROCm™ Software

Prerequisites and content levels

Prerequisites:

Familiarity with Linux operating systems, including Linux shell (training will use Ubuntu)
Access to an SSH client to enable remote access for interactive portions of the training
Working proficiency in English (all training will be conducted in English)
Basic understanding of machine learning/deep learning concepts

Familiarity with TensorFlow and Pytorch will be a plus. In suggested prereading (videos and resources below) you will find more AMD material.

Please make sure to have a functioning working environment and remote access for interactive portions of the training prior to the course. In case of questions, please contact the course organizer (see below).

Content levels:

Intermediate: 6 hours
Advanced: 5 hours

Learn more about course curricula and content levels.

YouTube videos

Resources

Instructors

Adil Lashab, Essam Morsi, Philipp Samfass and Jens Stapelfeldt (AMD).

Agenda (subject to update)

All times are CEST.

Day 1

9:30 - 10:00 Drop in to Zoom

10:00 - 16:00 Machine Learning with AMD GPUs and ROCm Software

10:00 - 10:10 Welcome - HLRS, AMD
10:10 - 10:30 AMD GPU portfolio
10:30 - 11:15 ROCm Tensorflow (TF) (MNIST for image classification example)
11:30 - 12:30 ROCm PyTorch (PY) (LSTM for NLP example)
12:30 - 13:30 Lunch break
13:30 - 14:15 Multi-GPU
14:25 - 15:25 Debugging and profiling
15:30 - 16:00 QA session

Day 2

9:00 - 9:30 Concepts in ML:
- training, inference, multi-GPU/multi-node
9:30 - 10:15 Hands-on ML: Training
- Single GPU training with TF and PY
- multi-GPU training with TF and PY
- distributed multi-Node training with TF and PY
10:30 - 11:15 Hands-on ML: Inference
- MLPerf MIGraphX & TVM Backend, Object detection
- NLP & image synthesis with GAN Inference MIVisionX
11:30 - 12:15 AMD AI performance
12:30 - 13:30 Lunch break
13:30 - 14:45 FPGA for AI
15:00 - 15:45 Natural language processing with AMD
15:45 - 16:00 QA session

Registration information

This course is already fully booked.

Fees

This course is free of charge.

PRACE PATC and bwHPC

HLRS is part of the Gauss Centre for Supercomputing (GCS), which is one of the six PRACE Advanced Training Centres (PATCs) that started in Feb. 2012.

HLRS is also member of the Baden-Württemberg initiative bwHPC.

This course is also provided within the framework of the bwHPC training program. This course is not part of the PATC curriculum and is not sponsored by the PATC program.

Contact

Khatuna Kakhiani phone 0711 685 65796, kakhiani(at)hlrs.de
Lorenzo Zanon phone 0711 685 63824, zanon(at)hlrs.de

Official course URL(s)

http://www.hlrs.de/training/2022/DL-AMD

Further courses

See the training overview and the Supercomputing Academy pages.