This mini course is held by Dániel Berényi and based on the similarly named course – now entering its 10th year – at Eötvös Loránd University (ELTE). The goal is to let scientists and developers familiarize themselves deeper with the capabilities and efficient usage of modern computing devices.

The presentations will cover:

• Modern CPU architectures
• Architectures and services of modern Graphical Processors (GPUs)
• Similarities and important differences of the two architectures
• Operating principles of memories and caches and how to make efficient use of them
• General overview of GPU programming interfaces (APIs)
• More detailed introduction to CUDA and OpenCL programming
• Various examples illustrating how different technological aspects influence the performance of applications

The course is theoretical in nature; it will not have a hands-on session, but sample codes and the presentation materials will be available to the participants. There are only minimal prerequisites: the presentations intend to provide insights independent of specific programming languages; however, the examples and sample codes will be in C++, so some basic knowledge of C, C++, or some other C-style language is advantageous.

• Lunch: on your own
• Inquiries should be addressed to Gábor Drótos.
• Participation is free, everyone is welcome to attend.