Single-Producer Single-Consumer (SPSC) lock-free ring buffer

Single-Producer Single-Consumer (SPSC)

• SPSC Introduction
• SPSC Lock-Free Ring Buffer for High-Frequency Trading
• Huge Pages & TLB in SPSC
• Detailed Design Notes and Trade-Offs
• Memory Ordering in SPSC Lock-Free Ring Buffers
• Atomics vs. Mutexes in SPSC

High Frequency Trading and SPSC

• Application in HFT
• Application in HFT: In-Depth Analysis
• Importance of queue size in HFT
• Vertical scaling in HFT
• Horizontal scaling in HFT
• Message Passing Interface (MPI) in HFT
• Five Leading commercial HFT platforms

High Frequency Trading Pipeline and SPSC

• Stage 1: NIC → Market Data Handler (Kernel Bypass)
• Stage 2: Market Data Parser → Order Book Engine
• Stage 3: Order Book Engine → Alpha / Signal Engine
• Stage 4: Signal Engine → Order Management System (OMS)
• Stage 5: OMS → Exchange Gateway (Outbound)
• Stage 6: Exchange Gateway → OMS (Inbound Execution Reports)
• Stage 7: OMS → Risk Manager
• Stage 8: Any Thread → Logger

Specific Data Structures in HFT

• Order book architecture in HFT systems
• Matching Engine Architecture for HFT
• Intrusive Linked List for an HFT Order Queue
• Flat Circular Array Order Queue for HFT
• Flat Hash Map + Sorted Intrusive Tree for an HFT Order Queue

Operating Systems in HFT

• Kernel Bypass: DPDK & Solarflare OpenOnload
• CPU Pinning
• Cache Pathologies
• Rocky Linux
• RHEL (Red Hat Enterprise Linux)
• AlmaLinux
• Fedora Linux

Hardware in HFT

• x86-64 CPUs for High-Frequency Trading
• Setup of x86-64 Linux HFT System
• x86-64 vs. IBM POWER on Linux
• Non-Uniform Memory Access (NUMA) Architecture
• Cache Coherency - Intel CPUs
• Cache Coherency - AMD CPUs
• Cache Coherence Across NUMA Nodes

Compiler Optimization in HFT

• Performance Analysis Tools for C++ HFT Systems
• Compiler Optimization Techniques for C++ in HFT Systems

Logging Libraries in HFT

• Quill Logger
• NanoLog Logger
• SPDLog Logger
• Measuring Execution Time

Market Data Protocols in HFT

• High-Frequency Trading Protocols:
  • FIX 5.0 (Financial Information eXchange)
  • OUCH (NASDAQ Order Entry Protocol)
  • ITCH 5.0 (NASDAQ Market Data Feed)
  • PITCH (CBOE/BATS Market Data Protocol)
  • SBE (Simple Binary Encoding)

Monitoring and Observability in HFT

• Grafana: pros and cons
• Prometheus: pros and cons

Databases in HFT

• KDB+
• QuestDB
• InfluxDB
• TimescaleDB
• ClickHouse
• Aeron Messaging Transport Library
• LMAX Disruptor Messaging Transport Library

Message Queue Systems in HFT

• Apache Kafka
• Kafka's Role
• Solace PubSub+
• IBM MQ
• Kafka vs. Solace

Multi Producer Multi Consumer (MPMC)

• Thorough Overview
• Detailed Design Breakdown

C++ Specific Problems

• The ABA Problem in Lock-Free Data Structures
• In C++: volatile vs. std::atomic
• Static and Dynamic Inheritance in C++
• Static Inheritance in C++: A Deep Technical Analysis

My LinkedIn Posts Related to HFT

• Trading at light speed by David Gross, Meeting C++ 2022
• Lock-Free Single Producer Single Consumer FIFO, CppCon 2023
• Lock-Free Programming by Herb Sutter, CppCon 2014
• Microseconds matter. Ring Buffer (SPSC queue) by Erik Rigtorp
• Ultrafast Trading Systems by David Gross
• Unlocking Modern CPU Power by Fedor G Pikus
• C++ atomics, from basic to advanced by Fedor G Pikus
• Every nanosecond counts. Lock-Free Ring Buffer by Joe Wood

Tests for SPSC implemenation

• Description of Basic Tests
• Description of Concurrent Tests

Compare ChatGPT (OpenAI) with Claude (Anthropic)

• Atomics vs. Mutexes
  • Atomics vs. Mutexes in SPSC - Claude
  • Atomics vs. Mutexes in SPSC - ChatGPT
• QuestDB
  • QuestDB - Claude
  • QuestDB - ChatGPT
• SPSC queue size
  • Importance of queue size in HFT - Claude
  • Importance of queue size in HFT - ChatGPT

---

📖 References

Video Presentations

• 🎥 "Lock-Free Programming (or, Juggling Razor Blades), Part I", 16 Oct 2014
• 🎥 "Lock-Free Programming (or, Juggling Razor Blades), Part II", 16 Oct 2014
• 🎥 "Low-Latency Lock-Free Ring-Buffer in C - Lock Free Programming (Part #2)", 13 Mar 2024
• 🎥 "Single Producer Single Consumer Lock-free FIFO From the Ground Up", CppCon 2023, 22 Feb 2024
• 🎥 "SPSC Queues: From Naive to Lock-Free", 24 Jan 2026
• 🎥 "Trading at light speed: designing low latency systems in C++", Meeting C++ 2022, 2 Jan 2023
• 🎥 "What is Low Latency C++? (Part 1)", CppNow 2023, 18 Aug 2023
• 🎥 "What is Low Latency C++? (Part 2)", CppNow 2023, 18 Aug 2023
• 🎥 "When Nanoseconds Matter: Ultrafast Trading Systems in C++", CppCon 2024, 28 Feb 2025

Articles and Software

• Erik Rigtorp, "Optimizing a Ring Buffer for Throughput", 14 April 2026
• Erik Rigtorp, SPSCQueue GitHub repo
• Jackson Allan, "An Extensive Benchmark of C and C++ Hash Tables", 14 Apr 2026
• Peter Mbanugo, "Building a Lock-Free Single Producer, Single Consumer Queue (FIFO)", 15 Apr 2026
• C++ Memory Model (cppreference)
• Boost C++ Libraries, SPSC Lock Free Queue
• Dmitry Vyukov Implementation
• C implementation of Dmitry Vyukov's Bounded MPMC queue
• Ode to a Vyukov Queue
• Memory Bounds for Concurrent Bounded Queues
• JF Bastien, Deprecating volatile, Published Proposal, 1 Oct 2018

Books

• Paul E. McKenney, "Is Parallel Programming Hard, And, If So, What Can You Do About It?", 18 December 2025, Release v2025.12.18a
• Paul E. McKenney, "Is Parallel Programming Hard, And, If So, What Can You Do About It?", 11 June 2023, Release v2023.06.11a
• Paul E. McKenney, "Is Parallel Programming Hard, And, If So, What Can You Do About It?", Original repository for the book
• Paul E. McKenney, "Is Parallel Programming Hard, And, If So, What Can You Do About It?", GitHub mirror repository for the book
• Paul E. McKenney Home Web Page