This book is designed for complete beginners and people who have just started to learn parallel computing. It does not require any specific knowledge besides the basics of computer architecture, and experience with C/C++ programming is assumed. For deep learning enthusiasts, in Chapter 10, Deep Learning Acceleration with CUDA, Python-based sample code is also provided, hence some Python knowledge is expected for that chapter specifically. 

The code for this book is primarily developed and tested in a Linux environment. Hence, familiarity with the Linux environment is helpful. Any of the latest Linux flavors, such as CentOS or Ubuntu, are okay. The code can be compiled either using a makefile or the command line. The book primarily uses a free software stack, so there is no need to buy any software licenses.  Two key pieces of software that will be used throughout are the CUDA Toolkit and PGI Community Edition.

Since the book primarily covers the latest GPU features making use of CUDA 10.x, in order to fully exploit all the training material, the latest GPU architecture (Pascal onward) will be beneficial. While not all chapters require the latest GPU, having the latest GPU will help you to reproduce the results achieved in the book. Each chapter has a section on the preferred or must-have GPU architecture in the Technical requirements section.