HIP Training Workshop
May 24-26, 2021
11:00 AM – 5:00 PM (ET)
- This meeting will be held as a virtual (online-only) event via Zoom. Please register using the form below. You will be emailed the Zoom coordinates for the meeting upon successful registration. The last day to register for this event is May 20, 2021.
About the Meeting
AMD will present a three-day hands-on workshop on using HIP to accelerate algorithms that are common in HPC workflows. This workshop is designed to help OLCF users for Frontier. Participants will receive an overview of HIP and run exercises to Hipify code on AMD’s accelerator cloud the first two days of the workshop. On the third day they will use what they have learned to accelerate a small application with HIP. This training will utilize the AMD accelerator cloud for the hands-on exercises.
HIP is a C++ runtime API that allows developers to write portable code to run on AMD and NVIDIA GPUs. It is an interface that uses the underlying Radeon Open Compute (ROCm) or CUDA platform that is installed on a system. The API is similar to CUDA so porting existing codes from CUDA to HIP should be fairly straightforward in most cases.
Participants are encouraged to familiarized themselves with HIP ahead of the event with material given here: https://developer.amd.com/resources/rocm-resources/rocm-learning-center/
The workshop is limited to 100 participants and will be given via Webinar. All OLCF users may participate on a first come first serve basis. We will form a wait list if capacity is reached.
Unfortunately this workshop has filled to capacity and registration is now closed. We will be hosting more events like this in the future.
[tw-toggle title=”Slides“]HIP Training Day 1: Introduction to GPU Programming
HIP Training Day 2: Converting CUDA Codes to HIP
HIP Training Day 3: ROCgdb and HIP Math Libraries
[/tw-toggle] [tw-toggle title=”Agenda“]Draft Agenda (ALL times given in EDT)
Day 1: May 2411:00 am Introduction to HIP and interactive exercises
- Learn How to accelerate codes with HIP
12:00 pm-5:00 pm Independent work on examples with help given on the event Slack channel
Day 2 May 25
11:00 am Accelerating, “hipifying”, and porting code examples.
- Learn how to use a script to “hipify” existing CUDA codes
- Learn how a code accelerated with HIP on NVIDIA GPUs can be ported to AMD GPUs
12:00- 5:00 pm Independent work on examples with help given on the event Slack channel
Day3 May 26
11:00 am Accelerating Applications with HIP
12:00 pm Use what you have learned to accelerate a mini application with HIP with help given on the event Slack channel
4:00 pm AMD Presenters share how they accelerated the mini application.
5:00 pm Close
[/tw-toggle]
