The CINECA Summer School on heterogeneous computing offers an intensive, week-long training experience focused on high-performance computing (HPC) and its integration with modern AI workflows. The program introduces the principles and practical use of diverse computing architectures, with particular attention to the combined use of CPUs and GPUs to achieve improved computational performance and efficiency. In addition, fundamental and applied aspects of deep learning are addressed, including model design, training workflows, and strategies for scaling from single-GPU environments to distributed systems.

Participants engage with core topics such as parallel programming models, hardware-aware optimization strategies, scalable software design, and efficient deep learning training techniques. The course combines theoretical lectures with hands-on laboratory sessions and interactive discussions, enabling direct application of the concepts introduced. The overall objective is to provide a solid foundation for leveraging heterogeneous computing and deep learning across a range of applications, from scientific research to industrial innovation.

Skills Covered

C/C++, Fortran, Python

Target Audience

This program is intended for individuals seeking to develop or strengthen their expertise in HPC, including:

• Master’s and PhD students in computational science, engineering, or related fields who aim to expand their technical competencies
• Researchers and engineers interested in applying HPC methodologies to scientific and engineering challenges
• Industry professionals who require up-to-date knowledge of advanced computing technologies

Applicants with extensive experience in parallel programming may find the course content too introductory.

Prerequisites

A basic working knowledge of C/C++, Fortran, and Python is expected.