Animated interactive study companion
General-Purpose Graphics Processor Architectures
GPU architecture is about what happens each cycle — warps diverging and reconverging, schedulers picking, memory accesses coalescing. So this companion animates it: every dynamic process is a diagram you can play, pause, and step cycle by cycle.
Based on General-Purpose Graphics Processor Architectures by Tor M. Aamodt, Wilson Wai Lun Fung, and Timothy G. Rogers (Synthesis Lectures on Computer Architecture #44, Morgan & Claypool, 2018). Personal study companion — not affiliated with the authors or publisher.
Your progress
0 / 22 sections
1
Foundations
Why GPUs exist, and how you program them.Chapter 1 · pp. 1–8
Introduction
2
The Two Halves of the Machine
The SIMT compute core, and the memory system that feeds it.Chapter 3 · pp. 21–66
The SIMT Core: Instruction and Register Data Flow
- One-Loop Approximation & SIMT Execution Masking
- SIMT Deadlock, Stackless SIMT & Warp Scheduling
- Two- & Three-Loop Approximation, Operand Collector
- Research: Warp Compaction
- Research: Intra-Warp Divergent Path Management
- Research: MIMD Capability & Complexity-Effective Divergence
- Research: Scalarization and Affine Execution
- Research: Register File Architecture
3
The Research Frontier
Scheduling, parallelism, transactional memory, heterogeneity.Chapter 5 · pp. 85–101