Learn CUDA Programming: A beginner’s guide to GPU programming and parallel computing with CUDA 10.x and C/C++

GPU computing can offer a different way to approach computationally demanding work, but making use of it means understanding how a GPU executes code—not simply moving a program from one processor to another. Learn CUDA Programming builds that understanding through CUDA 10.x examples in C and C++, progressing from first kernels and thread organization to memory behavior, optimization, and larger-scale GPU applications.

Start with CUDA’s execution model

The opening chapters introduce GPU programming and the CUDA model, including how threads and blocks are organized and how kernels are launched. Early examples give readers a practical setting for exploring those concepts, while also introducing the performance considerations that make parallel programming distinct from conventional CPU-focused development.

Understand memory, threads, and performance

From global and shared memory to unified and pinned memory, the book examines how data movement and access patterns affect GPU applications. It also explores occupancy, warp behavior, parallel reduction, synchronization, and profiling—ideas that help explain why a kernel performs as it does and where optimization efforts may matter.

Move from individual kernels to larger applications

Further chapters address streams, asynchronous execution, pipelining, dynamic parallelism, cooperative groups, and interactions with tools such as OpenMP and MPI. The broader scope extends to scalable multi-GPU programming, parallel patterns, GPU-accelerated libraries, OpenACC, and deep-learning acceleration.

A code-led route into GPU computing

Written for readers beginning with CUDA, this guide is also relevant to C and C++ programmers, students, and developers who want a structured introduction to GPU architecture and parallel application development. Its version-specific focus is explicit: the examples center on CUDA 10.x, giving readers a clear reference point as they study the platform’s programming model and optimization techniques.

Explore parallel programming with purpose

With its progression from foundational concepts to profiling and broader application strategies, Learn CUDA Programming connects the mechanics of writing CUDA code with the practical questions of memory use, execution, and performance.