Definition

• Ability of a program to be decomposed into parts that can run out of order or in partial order without affecting the final outcome
• Composition of independently executing tasks
• Concurrency is about managing multiple tasks, not speed
• On a single CPU core, only one task executes at any given moment

Multiple tasks start, run, and complete in overlapping time periods

Time-slicing

On a single-core processor, the CPU switches back and forth between tasks so fast it seems like they are happening at the same time

graph LR
  T1[Task A
Part 1] --> T2[Task B
Part 1]
  T2 --> T3[Task A
Part 2]
  T3 --> T4[Task C
Part 1]
  T4 --> T5[Task B
Part 2]
  T5 --> | ... | T6[End]

  style T1 fill:#f9f
  style T2 fill:#ccf
  style T3 fill:#f9f
  style T4 fill:#ff9
  style T5 fill:#ccf

Concurrency is a high-level concept

Multithreading

Multithreading is a specific way to achieve concurrency using a single process

Race Condition

All threads within a process share the same memory space

This makes communication fast, but risky:
multiple threads might change the same variable "at once"

sequenceDiagram
    autonumber
    actor T1 as Thread 1
    participant Mem as Shared Memory
(counter = 10)
    actor T2 as Thread 2

    Note over T1, T2: Goal: Both threads increment 'counter'

    %% --- The Intersection ---
    activate T1
    Note right of T1: Step 1: READ
    Mem->>T1: Load counter (10) into Register A

    activate T2
    Note left of T2: Step 2: READ (T2 runs before T1 writes)
    Mem->>T2: Load counter (10) into Register B

    Note left of T2: Step 3: INCREMENT Reg B = 11

    Note left of T2: Step 4: WRITE (T2 overwrites T1's work)
    T2->>Mem: Store Register B (11)

    Note right of T1: Step 5: INCREMENT Reg A = 11

    %% --- The Collision ---
    Note right of T1: Step 6: WRITE
    T1->>Mem: Store Register A (11)

    deactivate T1
    deactivate T2

    Note over T1, T2: Result: RACE CONDITION (Data Loss: counter = 11, expected: 12)
    

Multiprocessing is about parallel computing to improve performance (true parallelism)

It requires using two or more CPU cores to execute tasks simultaneously

Each process has its own private memory space and processes communicate via message passing

If one process crashes, the remaining processes can continue

Mini-example: make -j8 compiles up to 8 source files in parallel

Example: Modern Web Browsers

Modern browsers like Firefox moved to a Multi-Process Architecture

• The browser window and each individual tab are separate processes
• If a tab's process crashes, other tabs and the main window stay alive
• This also improves security (each tab has its own private memory space)

However, even though each tab uses its own process, each of those processes is still heavily multithreaded