Massively Parallel Processing or programming is both taught and extensively used at GRIS. Our integrated course “Programming Massively Parallel Processors” and the lab “Programming Massively Parallel Systems” teach not only the concepts of general purpose GPU (GPGPU) programming but allow the students to apply this knowledge to real-world problems in today’s research.
Massively Parallel Processing explained
Massively Parallel Processing originally appeared in form of the PRAM architecture which is a parallel extension to the traditional von Neumann Model that features a theoretically unlimited number of processors connected to a single shared memory. While this model was actually implemented as a real hardware, it was not widely spread. This, however, changed with the appearance of programmable GPUs. With the advent of general purpose GPU (GPGPU) programming, massively parallel processing became a widely available reality that is being used in a multitude of research areas, including machine learning, real-time of offline rendering, etc.