Theory of Multithreading

Basic

• Threads share cache and translation lookaside buffer

• There must be only one main thread path

• Concurrency - the idea of processing several tasks at the same time

• Process-based multitasking - concurrent processes

• thread-based multitasking - concurrent threads

Concurrency vs Parallelism

Concurrency - when things run at the same time (doesnt have to be related tasks)

Parallelism - when multiple parts of a program run at the same time (might be related tasks)

Pros of Concurrency

• Speed
  • better on multi core
  • process wont get blocked while waiting for say and IO

• Availability and Distribution
  • most real world things are happening simultaneously

• Controllability
  • functions can be started, stopped or interrupted by other concurrent functions

Thread Life-Cycle

How many threads?

threads ≤ cores on machine

Synchronization

Synchronisation is needed since without it, threads may clash when reading and writing data, resulting in false results.

Deadlock - when X thread is waiting for Y thread, but Y thread is waiting on X thread

Mutex

A thread locks a block of code,

no other thread can then run that code until it is unlocked