HPC Training Series - Course 16 "Compute at Scale: Concepts and Spotlight on AI, IoT, Simulations"

Monday 29 Sep 2025, 10:00 → 14:00 Europe/Athens

EuroCC@Greece announces the 16th Course of HPC Training Series with the subject "Compute at Scale: Concepts and Spotlight on AI, IoT, Simulations".  

Date: September 29th, 2025, at 10:00 EET  

Location: Online via Zoom

Presentation Language: Greek

Audience: Suitable for students, researchers, and engineers interested in gaining an understanding of essential concepts of parallelism and a practical grasp of its role in data-intensive domains of HPC.

Course Description: This course provides an introduction to essential parallelism concepts that underpin modern HPC and reviews how these principles underlie some of the most popular real-world use cases of high-performance computing (HPC) in the domains of AI, streaming systems, and scientific computing. The goal of the course is to help the participants take a step towards becoming well-informed users of technologies and frameworks that exploit parallelism, and to be introduced to the aspects they should consider when  gauging which set-ups may be suitable for their needs, as well as what benefits they can expect. To bridge theory with practice, the webinar concludes with two demo sessions: an introductory tutorial on PyTorch with Distributed Data Parallel (DDP), and a demo of the YOLOv8 model for scalable computer vision.

Learning Objectives:

• Understand the concepts of parallelism, distribution, scale

• Understand how appear or apply in HPC

• Understand the benefits of exploiting parallelism in HPC

• Recognize how parallelism underpins key domains such as AI, IoT, scientific computing.

• Become familiar with Pytorch DDP

• Become familiar with YOLOv8 computer vision model 

Learning Outcome:

• Be able to distinguish between parallelism, distribution, scale

• Compare expected benefits (e.g., speedup, scalability) from different HPC frameworks.

• Run and interpret a basic PyTorch DDP training workflow.

• Apply YOLOv8 in a parallel setting for basic computer vision tasks.

• Demonstrate informed judgment as end-users when selecting HPC tools and environments.

Use of interactive methods:

The course contains lectures dedicated to practical demonstrations of popular AI-related tools (the Pytorch DDP module, the YOLO v8 computer vision model), which will also provide code snippets to the audience. Furthermore, the course includes a dedicated Q&A session in order to foster interaction between trainers and trainees. 

Prerequisites: A prior knowledge of the basics of the Python language is beneficial for some parts of the course.

Note: Please enter your institutional/corporate email when registering.

 

Read the Privacy Statement here.pdf

10:00 → 10:05 Welcome and Agenda review

10:05 → 10:20 Introduction to EuroCC & the training events

Speaker: Nikos Bakas (GRNET)

10:20 → 10:50 Parallelism – Essential Concepts for HPC

Essential elements and fundamental principles of parallelism. Elementary techniques and programming frameworks. What are the main applications and what is the role of GPUs. What they mean for machine learning algorithms.

Speaker: Lena Kanellou (FORTH)

10:50 → 11:35 Building/Engineering high performance streaming solutions: experiences from our research for IoT

Experiences designing and engineering high-performance streaming solutions tailored for the unique challenges of IoT data. Architectural choices, scalability strategies, and lessons learned from prototyping and deploying systems that handle massive, real-time data flows. Open challenges and future directions for building robust, efficient streaming infrastructures that can keep pace with the rapid growth of IoT.

Speaker: Anastasios Gounaris (AUTH)

11:35 → 12:00 Scalable Simulations: Parallel Strategies for Plasma Science from the Plasma-PEPSC CoE

Speaker: Kallia Chronaki (FORTH)

12:00 → 12:15 Break

12:15 → 12:45 Tutorial: Introduction to Pytorch and the Distributed Data Parallel module (DDP)

Basics and main functions of PyTorch, one of the most popular tools for developing machine and deep learning models. Installation and basic use of PyTorch. Introduction into the Distributed Data Parallel library.

Speaker: Polidoros Dafnomilis (FORTH)

12:45 → 13:30 Demo - Tutorial: YoloV8

Core concepts and practical steps for using YOLOv8 in object detection tasks. Key components of the architecture and how to prepare custom datasets. How to train, fine-tune, and run inference on images and video. What it means for real-world computer vision applications and deployment.

Speaker: Klodjan Hidri (FORTH)

13:30 → 14:00 Open Discussion and Q&A