This book is aimed at one person in particular—that is, myself in the year 2014, when I was trying to develop a GPU-based simulation for my doctoral studies in math. I was poring over multiple books and manuals on GPU programming, trying to make the slightest sense of the field; most texts seemed happy to throw an endless parade of hardware schematics and buzzwords at the reader on every page, while the actual programming took a back seat.
This book is primarily aimed at those who want to actually do GPU programming, but without getting bogged down with gritty technical details and hardware schematics. We will program the GPU in proper C/C++ (CUDA C) in this text, but we will write it inline within Python code by way of the PyCUDA module. PyCUDA allows us to only write the necessary low-level GPU code that we need, while it automatically handles all of the redundancies of compiling, linking, and launching code onto a GPU for us.
