cppvalleyC++ · Systems · Low Latency · HFT
HFT Core Systems

HFT CORE SYSTEMS

The C++ memory model and data-race freedom

Lesson 57 of 96Phase 06: Concurrency and lock-free engineering
Video lesson coming soon

The written objectives and hands-on lab are available now.

cppvalley on YouTube ↗

Learning objectives

What you should understand

  1. Define happens-before and conflicting actions
  2. Recognize undefined behavior from data races
  3. Separate compiler and CPU reordering models

Lab

Hands-on exercise

Analyze and repair a deliberately racy handoff.

Deliverable

Evidence of completion

A happens-before graph plus ThreadSanitizer evidence.

Prerequisite
EP 56

Phase 06

Concurrency and lock-free engineering

Full curriculum
57The C++ memory model and data-race freedom58Sequential consistency and total order59Acquire/release publication60Relaxed atomics, counters, and misconceptions61SPSC ring buffer from invariants first62Bounded MPMC queues and sequence numbers63Michael–Scott queue and linked-node trade-offs64Spinlocks, atomic_wait, futexes, and backoff65ABA: mechanism, not slogan66Hazard pointers and safe reclamation67Epoch reclamation and user-space RCU68Phase gate: queue library and concurrency defense