Workshop details
The KAUST Supercomputing Core Laboratory (KSL) in collaboration with AMD are pleased to invite you to the "KAUST-AMD GPU Computing Workshop". This event provides an opportunity for attendees to learn about the latest advances by AMD GPUs in Scientific Computing and Artificial Intelligence as well as showcasing some use cases of relevant KAUST applications.
The workshop is planned on May 25th-27th , 2021 from 11AM to 6PM and delivered online virtually by AMD experts. The agenda of the workshop and the details for registration and speakers are detailed below.
Registration
Please fill up the registration form for this main event of the workshop on May 25th. Please use these seperate registration links for the Hands-On sessions on May 26th for HPC and 27th for ML, since those are are restricted to KAUST only and limited to 20 participants only.
You will receive a confirmation email automatically from Zoom along with the link to join the webinar.
We are looking forward to meeting you!
Agenda
Main Event: May 25th, 2021 from 11AM to 4:30PM
11:00AM: Welcome notes by organizers
11:00AM Session 1: AMD GPUs (~2 hours)
- CDNA Overview
- CDNA Memory Hierarchy
- Interlude on GPU Compute Terminology
- CDNA Compute Units
1:00-1:15 PM - Break
1:15 PM Session 2: AMD HIP (~2 hours)
- Introduction to HIP
- Core HIP API
- Memory Management in HIP
- Asynchronous Computing with HIP
- Tips & Tricks
3:15-3:30 PM - Break
3:30PM Session 3: AMD ROCm (~1 hours)
- Introduction to ROCm
- Porting CUDA & OpenACC codes to HIP
- ROCm Libraries
- Profiling tools
HPC Hands-On and Hackathon: May 26th, 2021 from 11AM to 6PM - Limited to 20 participants from KAUST Only
11:00AM Session 1: CUDA to HIP (~3 hours)
- SPECFEM3D
- KBLAS
2:00PM Session 2: OpenACC to HIP (~2 hours)
- Example: vector add
4:00 PM Session 3: Profiling HW counters (~2 hours)
- RocProfiler
- RocTracer
- ISA code
ML Hands-On: May 27th, 2021 from 11AM to 3PM - Limited to 20 participants from KAUST Only
11:00AM: Session 1: HPC Hands-On (~2 hours)
- Any follow up on May 26th session
1:00PM: Session 2: ML Hands-On (~2 hours)
- Training with TF & Pytorch
- Object detection, NLP & image synthesis with GAN
- Inference MIVision
Organizers
Saber Feki and Mohsin Shaikh - KAUST Supercomputing Core Lab
Mohammed Jaghoub, AMD
Trainers Biography
JC Baratault is EMEA Senior Market Development Manager, GPU Datacenter Business Unit at AMD, he joined AMD in 2013. JC initiated and managed the project which won the Green500 competition in November 2014 with the University of Frankfurt, Germany. Prior to joining AMD, JC enjoyed a 12-year career with NVIDIA in Europe. During this time JC led the EMEA Workstation Oil & Gas team and launched the GPU Computing activity in EMEA.”
Mazda Sabony is a senior manager of HPC/ML CSG (Client Solution Group). CSG is responsible for E2E optimization of HPC apps e.g. Gromacs, NAMD, LAMMPS, ZFP, RTM, … etc. on AMD GPUs. In his previous live, Mazda was leading version ASICs design at Infineon Technology. He holds BSc degree in physics from the university of Mosul.
Adil Lashab is a software system design engineer in the HPC/ML CSG (Client Solution Group). Adil has Extensive experience in Enablement and Optimization of the overall performances of the HPC and Machine learning Applications, starting from the application as high-level down to the low-level ISA code. He holds Master degree in Informatic from the Frankfurt University .
Dominic E. Charrier is a Software Development Engineer in AMD’s HPC/ML Client Solution Group (CSG). In the past, he has worked on the HIP C++ ports of SPECFEM3D_globe and AMBER but focuses now predominantly on porting accelerated Fortran HPC codes to Fortran & HIP C++ (or OpenMP). To make his work easier, he develops a translation tool called GPUFORT. Dominic holds a M.Sc. in Computational Engineering from TU Darmstadt, Germany, and a PhD in Computer Science from Durham University, UK.
Contact details
Follow us on Twitter @KAUST_HPC and please contact us at training@hpc.kaust.edu.sa if you need further information.
Our KSL training recordings are now available for you to browse on-demand in our KSL Youtube channel , Subscribe and hit the notifiation button to stay tuned with our latest material.