The aim of the training is to provide a detailed introduction to the GPU partition of the Komondor supercomputer in order that our users to take full advantage of the supercomputer's capabilities.
Following a detailed overview of the hardware architecture of GPU partitions (including CPU and GPU architectures, node architecture, and multi-node communication), we will examine the hardware and software characteristics that fundamentally influence, or may influence, the computational performance of the programs being executed. The knowledge gained will help further accelerate applications running on Komondor and enable more efficient scaling of the size and number of computational tasks to be solved.
Lecturer_ Dr. Zoltán Juhász, HPC expert