The EPICURE Hackathon, held from October 28-31, 2024, brought together developers, researchers, and enthusiasts from across Europe to dive deep into the world of GPU programming and high-performance computing (HPC). This event was hosted in collaboration with CINECA, one of the leading HPC centers in Europe, renowned for its contributions to scientific computing and advanced research.
The hackathon was meticulously structured, featuring both hands-on labs and theoretical sessions, designed to empower participants with the latest skills in CUDA programming, OpenACC, and GPU optimization techniques.
Day 1: Getting Started with GPU Computing
The first day served as an introduction to GPU computing, aimed at setting a strong foundation. Topics included:
- Introduction to GPU Computing: A look into the architecture and advantages of using GPUs for high-performance tasks.
- CUDA Basics: Practical insights into CUDA, NVIDIA’s parallel computing platform, providing a gateway to harnessing the power of GPU cores.
The agenda for Day 1 was as follows:
| Time | Session |
|---|---|
| 9:00 - 9:15 | Welcome and Registration |
| 9:15 - 9:30 | EPICURE HPC Application Support for EuroHPC Users |
| 9:30 - 10:30 | Introduction to GPU Computing |
| 10:30 - 11:00 | Coffee Break |
| 11:00 - 12:30 | Introduction to CUDA |
| 12:30 - 14:00 | Lunch |
| 14:00 - 15:30 | OpenACC Programming and Profiling with Nsight |
| 15:30 - 16:00 | Coffee Break |
| 16:00 - 17:30 | OpenACC Programming and Profiling with Nsight |
Images from Day 1:
Day 2: Advanced GPU Optimization and Profiling
Day 2 took things to the next level, focusing on advanced GPU optimization and profiling techniques. Participants were guided through various optimization methods to fine-tune GPU applications, reducing computation time and maximizing efficiency.
| Time | Session |
|---|---|
| 9:00 - 10:30 | Advanced GPU Optimization and Profiling |
| 10:30 - 11:00 | Coffee Break |
| 11:00 - 12:30 | Advanced GPU Optimization and Profiling (Continued) |
| 12:30 - 14:00 | Lunch |
| 14:00 - 17:30 | Lab Session (with your code or a provided code) |
The hands-on lab in the afternoon provided an opportunity to implement what we learned using real-world examples, supported by the CINECA team.
Day 3-4: Intensive Lab Sessions
The last two days were dedicated to intensive lab sessions. Participants worked on challenging GPU computing projects, with mentors available to provide guidance and feedback. These sessions enabled us to:
- Test and debug our code using real-time feedback from experts.
- Engage in peer-to-peer learning, exchanging ideas and optimizing our implementations.
- Utilize Nsight and other profiling tools to analyze performance and identify bottlenecks.
| Date | Morning (9:00 - 12:30) | Afternoon (14:00 - 17:30) |
|---|---|---|
| October 30 | Lab Session | Lab Session |
| October 31 | Lab Session | Lab Session |
Key Takeaways
This hackathon was more than just a technical event; it was a community-building experience. Here are some of the main takeaways:
- Networking: I had the opportunity to connect with like-minded individuals passionate about GPU programming and HPC.
- Learning New Tools: From CUDA basics to advanced profiling with Nsight, I gained hands-on experience with cutting-edge tools.
- Real-World Problem Solving: The projects and examples we tackled were grounded in real-world applications, highlighting the impact of GPU acceleration in domains like scientific research, machine learning, and data analysis.
Final Thoughts
Reflecting on these four days, I feel a renewed enthusiasm for GPU computing and HPC. The skills I gained, combined with the amazing people I met, made this hackathon an unforgettable experience.
Interested in learning more? Check out the links below to access more details on EPICURE events, a recap video, and the GitHub repository with sample codes used during the hackathon.
Links
- EPICURE HPC Event Page
- Event Recap Video
- GitHub Repository of Sample Codes
- GitLab GPU Repository
- HackMD Hackathon Document
- My HackMD Project Document
This hackathon has certainly sparked a deeper interest in exploring the potential of GPUs for high-performance tasks. I’m looking forward to applying these skills in future projects and sharing my insights with the community.
YouTube Vlogs
I’ve made some mini vlogs about my experience in Bologna during this event and this is the first of the playlist