EoI 1-Day Introduction to AMD GPUs (AMD Instinct Architecture and ROCm)
9:30am - 4:00pm
This is a hands-on, online event. Places are limited so register only if you plan to attend.
Join AMD Specialists to learn about AMD Instinct GPUs and ROCm toolchain, including HIP basics. The training focuses on both HPC and AI ML using containers from both AMD’s Github ROCm source repository and the New AMD Infinity Hub.
- Overview of MI200 (AMD Instinct Architecture and Introduction to ROCm)
- Course Part 1 (2Hrs) + 1 Hr Extended Lab time. – AI / ML FOCUS
- System Setup and ROCm installation
- Basic ROCm tools including ROCm Debugger
- TensorFlow installation and benchmarking
- PyTorch installation and benchmarking
- Multi-GPU scaling
- Machine Learning models deployable and benchmarks
- ROCm HPC and ML Docker Containers
- HPC Job Scheduler and Monitoring
- Math Libraries
- Communication Libraries for Multi-GPU Scale-out
- CUDA to HIP – Tools and Examples
- GPU Performance Profiling
- HPC Benchmarking with Examples
This training includes exercises / labs for those who want to experiment with the AMD Accelerator Cloud.
Attendee limit: 8 individuals. Each student will have a dedicated AMD GPU (MI100) during the training.
Please register only if you plan to attend.