In collaboration with our partner AMD, we delivered an introductory training focused on optimising AMD’s cutting-edge MI300 series GPUs.
AMD experts delivered this training remotely over the afternoons of Tuesday 11th March, Wednesday 12th March and Thursday 13th March, from 1.00 pm to 5.00 pm. The sessions were suitable for all users to gain knowledge and skills on GPUs, and to explore AMD’s cutting-edge MI300 series GPUs, including the new MI300A Accelerated Processing Unit (APU).
These sessions were designed to help you:
-
Grasp new concepts for enhanced memory management, with the potential of totally removing data transfer times.
-
Learn techniques that unlock the full potential of these powerful GPUs.
-
Take advantage of AMD’s latest innovations to supercharge your research and development workflows.
DiRAC teamed up with Intel to deliver 3 afternoons of introductory training on the Intel oneAPI Base Toolkit – a core set of tools and libraries for developing high-performance, data-centric applications across diverse GPU architectures.
After the announced in 2024, by the UK government to investment into GPU-based compute for AI and simulation, and a more general move towards accelerator based systems, the successful refactoring of CPU based codes onto GPU architectures can be critical to future performance and portability.
Delivered by experts from Intel and their training partners Codeplay, this 3 afternoon, lecture-based, virtual course covered some of the essential concepts and introduced basic skills, including:
- Intel Advisor Tool
- C++ to SYCL
- SYCL Essentials training
- SYCL ibraries – oneMKL, oneDPL and oneDNN
- VTune
