GPU Parallel Computing Second Edition: From Basics to Breakthroughs in GPU Programming

By Gareth Thomas

Second Edition updated for 2026. This book is a complete rewrite with entirely new code samples while covering the same core topics as the original edition.

All code and infographics from the book are available at the GitHub repository BurstBooksPublishing.

GPU Parallel Computing: From Basics to Breakthroughs - A Technical Guide to GPU Programming

If you want to understand how modern GPUs work and how to use them effectively for high-performance workloads, this book provides the technical foundation required.

This book assumes no prior exposure to GPU internals; however, a working knowledge of electronics and general computer architecture is recommended.

It is written for students, engineers, researchers, and data scientists who are new to GPU architecture and parallel programming and want a rigorous introduction before progressing into optimization and large-scale GPU systems.

If you are already an experienced CUDA performance engineer or low-level GPU architect seeking a specialized microarchitectural reference manual, this book is not positioned for that purpose.

What You Will Learn
GPU Architecture Fundamentals

Streaming multiprocessors and SIMT execution

Warp scheduling and instruction flow

GPU memory hierarchy and bandwidth considerations

GPU Programming Models

CUDA programming principles

OpenCL fundamentals

Kernel structure and execution behavior

Performance Optimization

Memory access patterns and coalescing

Warp divergence and latency hiding

Occupancy principles and kernel configuration

Real-World Applications

Scientific simulations

Machine learning workloads

Graphics and visualization pipelines

Advanced Topics

Multi-GPU communication

Tensor cores and mixed precision

Profiling, debugging, and performance analysis

The early chapters establish architectural clarity and programming fundamentals.
Later chapters address optimization strategies, scalability, and applied GPU workloads.

Who This Book Is For

Students entering GPU computing

Engineers transitioning into parallel architecture

Researchers and data scientists adopting GPU acceleration

This is a technical book. It builds understanding from architectural principles upward and focuses on performance-oriented reasoning rather than superficial overview.

Why This Book

Many GPU resources either assume too much prior knowledge or remain overly abstract.

This book emphasizes structured technical understanding:

How GPUs execute threads

Why performance bottlenecks occur

How architectural constraints shape results

How programming decisions map to hardware behavior

Clear explanations.
Practical code examples.
Architectural context.

• Format: Paperback
• Language: English
• ISBN: B0GR9TWTVJ
• ISBN13: 9798250799591
• Release Date: March 2026
• Publisher: Independently Published
• Length: 408 Pages
• Weight: 2.07 lbs.
• Dimensions: 0.8" x 8.5" x 11.0"