Book-length deep dive: virtual memory from first principles — page tables, TLBs, NUMA and performance
abhi9u · x · 2026-09-06
Abhinav Upadhyay of Coding Confessions published a months-in-the-making, nearly book-length deep dive on virtual memory, with a typeset 60-page PDF/EPUB edition.
- Covers why virtual memory exists: process isolation, memory protection, and the illusion of abundant memory.
- Explains page tables, TLBs, and page faults: random access over large datasets causes TLB misses, which is why data layout and structure-size optimizations matter beyond CPU caching.
- Details multi-core realities: per-core TLBs requiring TLB shootdowns, and NUMA topology where nearby RAM is far faster than remote RAM.
- Practical topics include demand paging, copy-on-write and fork, mmap avoiding page-cache copies, page reclaim/swap, huge pages, and their performance impact on data-intensive systems.
More from Infra
- Independent Sweep Puts DiffusionGemma Peak Throughput at 3k tok/s, 3x Paper's Claim — bodonoghue85 · 2026-09-06
- 推理时给预训练 LLM 加滑窗注意力，64K 上下文 KV 缓存仅 3.5MB — ahsaor8 · 2026-09-06
- A One-Stop Dashboard to Find the Cheapest GPUs for Running MiniMax H3 — CategoryFew5869 · 2026-09-06
- Villager sim game built locally on 16GB VRAM with Qwen3.8-27B Q3 quant at 75 tok/s — Fancy-Snow7 · 2026-09-06
- US golf courses use 30.5x more water than all data centers combined — MaxUnfried · 2026-09-06
- Qwen3.8 Flash hits 45 tok/s on M4 Max, matching Qwen3.8 27B speed on Apple Silicon — DerTomsn · 2026-09-06