The rapid growth of complexity in modern applications have exceeded the General Purpose Central Processing Units (GPCPUs) ability to deliver. Modern High Performance Computing (HPC) applications demand more than what current processors can deliver, thus they have created a performance gap between the demand for computational power and performance. This forces us to search for new architectural solutions as we are reaching the end of Moore's Law predictions. One solution is to use hardware accelerators to offload compute-intensive tasks from the main processor. Some examples of hardware accelerators are Graphic Processing Units (GPUs) and Field Programmable Gate Arrays (FPGAs). GPUs in particular, have recently proven to be a much more efficient architecture compared to GPCPUs. In recent years Graphic Processing Units have seen widespread adoption in many scientific fields, from Machine Learning (ML) to Genomics. Their use makes it possible to achieve significant speedups and improvements in power efficiency over computationally intensive algorithms compared to General Purpose Central Processing Units. However, algorithms require specific knowledge of the GPU architecture and expertise to achieve significant results. The aim of this course is to provide knowledge and hands-on experience in developing applications for heterogeneous systems accelerated with GPUs. This course will provide the knowledge necessary to understand the architecture of GPUs used for acceleration of general-purpose applications, and how it can be made available in a heterogeneous computing system. At the end of the course the student will be able to exploit the GPU architecture to accelerate various applications using the CUDA programming language.

dal March 2023 a June 2023