Back To Course Home

Modern C++ Concurrency: Get the most out of any machine

0% completed

Introduction

A Guide to Follow This Course

A Quick Overview

C++11 and C++14: The Foundation Multithreading in C++Case Studies C++17: Parallel Algorithms of the Standard Template Library C++20: The Concurrent Future Challenging Issues in C++

Memory Model: The Contract

The Contract The Foundation & Challenges

Memory Model: Atomics

Strong Memory Model Weak Memory Model The Atomic Flag Spinlock vs. Mutex std::atomic<bool>User Defined Atomics All Atomic Operations Free Atomic Functions Atomic Operations on std::shared_ptr

Memory Model: Synchronization and Ordering Constraints

Introduction Types of Synchronization & Ordering Constraints Sequential Consistency Acquire Release Semantic Is the Acquire-Release Semantic Transitive?Acquire Release: The Typical Misunderstanding std::mem_order_consume Data dependencies with std::memory_order_consume Relaxed Semantic

Memory Model: Fences

Fences as Memory Barriers The Three Fences Acquire and Release Fences Synchronization with Atomic Variables Synchronization with Fences

Multithreading: Threads

Introduction to Threads Creation of Threads Managing Thread Lifetime Thread Lifetime Management: Warnings and Tips Passing Arguments to Threads Arguments of Threads: Undefined behavior Arguments of Threads - Race Conditions and Locks Methods of Threads Methods of Threads in Practice

Multithreading: Shared Data

Introduction to Shared Data Introduction to Mutexes Mutex Types and Locking Methods Issues of Mutexes: Deadlocks Issues of Mutexes: Avoiding Exceptions Types of Locks: std::lock_guard Types of Locks: std::unique_lock Types of Locks: std::shared_lock Thread-Safe Initialization Thread-Safe Initialization: Constant Expressions Thread-Safe Initialization: call_once and once_flag Thread-Safe Initialization - Static Variables with Block Scope

Multithreading: Local Data

Thread Local Data

Multithreading: Condition Variables

Condition Variables The Caveats of Condition Variables

Multithreading: Tasks

Introduction to Tasks Threads vs Tasks Introduction to std::async async: Start Policy async: Fire and Forget async: Concurrent Calculation Introduction to std::packaged_task Introduction to Promises and Futures Promise and Future : Return an Exception Promise and Future: Return a Notification Introduction to std::shared_future

Case Study: Calculate Sum of a Vector

Introduction Single Threaded Summation: Ranged Based for Loops Single Threaded Summation: Addition with std::accumulate Single Threaded Summation: Protection with Locks Single Threaded Summation: Protection with Atomics Multithreaded Summation: Using std::lock_guard Multithreaded Summation: Using Atomic Variable Multithreaded Summation: Using fetch_add Method Multithreaded Summation: Using fetch_add Method with Relaxed Semantic Thread Local Summation: Using Local Variable Thread Local Summation: Using an Atomic Variable with Sequential Consistency Thread Local Summation: Using an Atomic Variable with Relaxed Semantic Thread Local Summation: Using Thread Local Data Thread Local Summation: Using Tasks Calculate Sum of a Vector: Conclusion

Case Study: Thread-Safe Initialization of a Singleton

Introduction Double-Checked Locking Pattern Performance Measurement Classical Meyers Singleton Introduction to Thread-Safe Meyers Singleton Thread-Safe Singleton: std::lock_guard Thread-Safe Singleton: std::call_once with std::once_flag Thread-Safe Singleton: Atomics Conclusion

Case Study: Ongoing Optimization with CppMem

Introduction to CppMem CppMem: An Overview CppMem: Non-Atomic Variables CppMem: Locks CppMem: Atomics with Sequential Consistency CppMem: Atomics with an Acquire-Release Semantic CppMem: Atomics with Non-Atomics CppMem: Atomics with a Relaxed Semantic Conclusion

Parallel Algorithms of the Standard Template Library

Execution Policies Parallel & Vectorized Execution Algorithms The New Algorithms New Algorithms - A Functional Perspective

The Future: C++20

Atomic Smart Pointers Thread-Safe Linked List Using Atomic Pointers Introduction to Extended Futures Attaching Extended Futures Creating New Futures Latches and Barriers Transactional Memory - An Overview The Two Flavors of Transactional Memory Introduction to Coroutines Coroutines: More Details Task Blocks

Coding Examples

Blocking Issues

Breaking of Program Invariants

Lifetime Issues of Variables

Race Conditions

Best Practices

General Memory Model Multithreading: Threads Multithreading: Shared Data Multithreading: Condition Variables

The Time Library

The Interplay of Time Point, Time Duration, and Clock Time Point From Time Point to Calendar Time Time Duration Time Duration Calculations Clocks Sleep and Wait

Glossary

Glossary Running Source Code on your own machine

About the Author

Data Races

This lesson gives an overview of data race problems which might occur during the implementation of concurrency in C++.

A data race is a situation in which at least two threads access a shared variable at the same time. Within that, at least one thread tries to modify the variable.

If your program has a data race, it will have undefined behavior. This means all outcomes are possible and, therefore, reasoning about the program makes no sense anymore. Let me show you a program with a data race.

Get hands-on with 1400+ tech skills courses.