Yun-HI554×694 165 KB

SungHwan Yun

Grokking Parallel Programming: With Examples in CUDA by SungHwan Yun takes a much friendlier route. It starts with a question that matters: how do you stop thinking sequentially and learn to recognize work that can happen at the same time?

The book uses visual explanations, approachable analogies, and small CUDA programs to build that way of thinking from the ground up. You’ll learn how to:

• Write and launch your first CUDA kernels
• Organize work across threads, blocks, and grids
• Move data safely between the CPU and GPU
• Recognize reusable patterns such as Map and Reduction
• Avoid race conditions and out-of-bounds memory access
• Understand when parallelization will help—and when it won’t

One detail I particularly like is that you don’t need to own an NVIDIA GPU to begin. The exercises run in a free browser-based lab, so anyone comfortable with basic C functions and loops can start experimenting with actual CUDA code.

Five chapters are currently available through Manning’s Early Access Program. Because it’s a MEAP, readers can follow the book as it develops, receive new chapters as they’re released, and send feedback that can help shape the finished edition.

If you’ve been curious about what happens underneath frameworks such as PyTorch and TensorFlow—or you simply want to develop a better instinct for designing parallel algorithms—this is a practical place to start.

• Full details: Grokking Parallel Programming - SungHwan Yun

Don’t forget you can get 45% off with your Devtalk discount! Just use the coupon code “devtalk.com” at checkout