Modern GPU Architecture and Programming Mini Bundle

If you’ve ever wondered why your GPU code hits a wall long before the hardware’s limits, this book tells you why—and how to break through it.

Most programmers stop where the compiler starts. They trust nvcc to make the right decisions, to manage registers, to schedule instructions, and to use memory efficiently. But the compiler doesn’t know your problem. It guesses. And in GPU computing, guessing costs performance.

Mastering PTX and SASS – Volume I pulls back the curtain on NVIDIA’s virtual machine—the PTX instruction set that every CUDA kernel becomes before it touches silicon. You’ll learn how threads, warps, and memory really behave at the hardware level, how each instruction interacts with caches and pipelines, and how to read, write, and reason about PTX like an architect, not just a coder.

This isn’t a surface-level “how-to.” It’s a deep, methodical tour through the machinery of modern GPUs—built for professionals who want measurable, repeatable speedups, not guesswork. You’ll discover how the compiler transforms your high-level logic into executable reality, and where you can step in to take control.

By the time you finish, you won’t be relying on compiler magic. You’ll understand it, improve it, and surpass it.

Mastering PTX and SASS – Volume I gives you the foundation; Volume II takes you to the bleeding edge of optimization. Together, they turn GPU performance from a mystery into a science.

You’ve mastered the architecture—now it’s time to own the performance.

Every GPU developer hits the same wall: the profiler says you’re close to peak, but you know there’s still headroom. What’s missing isn’t another compiler flag—it’s visibility into the hardware’s final truth. That truth lives in SASS, the real machine code running on NVIDIA GPUs.

Mastering PTX and SASS – Volume II takes you past theory into the territory where nanoseconds matter. Here you’ll learn how to read, analyze, and tune instruction streams with surgical precision. You’ll uncover how schedulers pair ops, how register pressure throttles throughput, and how to turn your kernels into clock-cycle-balanced engines of pure efficiency.

This book is for engineers who refuse to settle for “good enough.” It turns profiling, disassembly, and optimization into a repeatable process—one grounded in data, not superstition. From tensor cores to warp shuffles, from atomic operations to multi-GPU scaling, you’ll learn how real experts bend hardware to their will.

Volume I built the foundation; Volume II shows you how to weaponize it.
If you’re ready to squeeze every drop of performance from your GPU—and understand exactly how you did it—this is the manual you’ve been waiting for.

Modern GPU Architecture Second Edition — Volume One
Graphics Pipeline Design and Hardware Implementation

Modern GPUs are the most complex and efficient parallel processors ever created—and this book shows you exactly how they work at the hardware level.

Unlike typical graphics or programming guides, this volume takes you inside the GPU itself:
how instructions flow through pipelines, how memory hierarchies sustain bandwidth, how shader cores and fixed-function units cooperate to render billions of pixels per second.

You’ll explore every major stage of the graphics pipeline in depth—geometry, rasterization, shading, texturing, and render output—all supported by clear mathematical models and synthesizable Verilog examples. This is not “theory for theory’s sake”; it’s engineering detail you can apply directly in design, simulation, or hardware verification.

By reading this book, you’ll gain:

• Architectural intuition — understand how throughput, latency, and bandwidth interact in real GPUs.
• Practical RTL-level insight — see how each stage can be implemented with clean, synthesizable Verilog.
• A foundation for advanced design — build the knowledge required for AI acceleration, compute architectures, or FPGA-based GPU prototyping.
• Confidence to analyze real silicon — reason about performance, bottlenecks, and tradeoffs like a hardware architect.

Every chapter bridges concept and implementation, making it invaluable for anyone designing graphics hardware, studying computer architecture, or seeking mastery of parallel computation systems.

Dense, detailed, and unapologetically technical, this book is written for those who want to understand modern GPUs—not just use them.

⚠️ This isn’t entertainment. It’s engineering.
If that excites you, welcome aboard.
If it intimidates you, this book isn’t for you.

From the Editor at Burst Books — Gareth Thomas

A Smarter Kind of Learning Has Arrived — Thinking on Its Own.

Forget tired textbooks from years past. These AI-crafted STEM editions advance at the speed of discovery. Each page is built by intelligence trained on thousands of trusted sources, delivering crystal-clear explanations, flawless equations, and functional examples — all refreshed through the latest breakthroughs.

Best of all, these editions cost a fraction of traditional texts yet surpass expectations. You’re gaining more than a book — you’re enhancing the mind’s performance.

Explore BurstBooksPublishing on GitHub to find technical samples, infographics, and additional study material — a complete hub that supports deeper, hands-on learning.

In this age of AI, leave the past behind and learn directly from tomorrow.

Modern GPU Architecture Second Edition — Volume One
Graphics Pipeline Design and Hardware Implementation

Modern GPUs are the most complex and efficient parallel processors ever created—and this book shows you exactly how they work at the hardware level.

Unlike typical graphics or programming guides, this volume takes you inside the GPU itself:
how instructions flow through pipelines, how memory hierarchies sustain bandwidth, how shader cores and fixed-function units cooperate to render billions of pixels per second.

You’ll explore every major stage of the graphics pipeline in depth—geometry, rasterization, shading, texturing, and render output—all supported by clear mathematical models and synthesizable Verilog examples. This is not “theory for theory’s sake”; it’s engineering detail you can apply directly in design, simulation, or hardware verification.

By reading this book, you’ll gain:

• Architectural intuition — understand how throughput, latency, and bandwidth interact in real GPUs.
• Practical RTL-level insight — see how each stage can be implemented with clean, synthesizable Verilog.
• A foundation for advanced design — build the knowledge required for AI acceleration, compute architectures, or FPGA-based GPU prototyping.
• Confidence to analyze real silicon — reason about performance, bottlenecks, and tradeoffs like a hardware architect.

Every chapter bridges concept and implementation, making it invaluable for anyone designing graphics hardware, studying computer architecture, or seeking mastery of parallel computation systems.

Dense, detailed, and unapologetically technical, this book is written for those who want to understand modern GPUs—not just use them.

⚠️ This isn’t entertainment. It’s engineering.
If that excites you, welcome aboard.
If it intimidates you, this book isn’t for you.

From the Editor at Burst Books — Gareth Thomas

A Smarter Kind of Learning Has Arrived — Thinking on Its Own.

Forget tired textbooks from years past. These AI-crafted STEM editions advance at the speed of discovery. Each page is built by intelligence trained on thousands of trusted sources, delivering crystal-clear explanations, flawless equations, and functional examples — all refreshed through the latest breakthroughs.

Best of all, these editions cost a fraction of traditional texts yet surpass expectations. You’re gaining more than a book — you’re enhancing the mind’s performance.

Explore BurstBooksPublishing on GitHub to find technical samples, infographics, and additional study material — a complete hub that supports deeper, hands-on learning.

In this age of AI, leave the past behind and learn directly from tomorrow.