Deep Dive Documentation Series

AMD ROCm Platform

Comprehensive technical exploration of AMD GPU computing, from GCN ISA to CDNA architecture and HIP programming

🔥 ROCm 7.x | CDNA Next Architecture
Chapters

Select Your Module

01
🔧

GCN ISA & ROCm Binaries

Deep dive into the ROCm compilation pipeline. Understand HIP programming, LLVM IR, GCN/RDNA native code, hsaco binaries, and the code object bundles that enable portable GPU computing.

HIP LLVM IR GCN ISA hsaco hipcc
Enter Chapter →
02
🏗️

GPU Architecture Deep Dive

Evolution from GCN to CDNA Next. Explore compute units, matrix cores, memory hierarchies, and the revolutionary architectural advances powering AMD's Instinct accelerators.

GCN 5 CDNA 1 CDNA 2 CDNA 3 CDNA 4 CDNA Next
Enter Chapter →
03
⚡

ROCm API & Kernel Execution

Master the ROCm software stack. Compare HIP Runtime vs HSA APIs, understand ROCm libraries, and learn how kernels are launched and executed across AMD compute units.

HIP Runtime HSA rocBLAS MIOpen KFD
Enter Chapter →