concurrency

A small repo for my concurrency stuff

there are two header files:

• custom_locks.hpp : two mutexes are defined, a spin_lock and a ticket_mutex, based on these two but with an added try_lock() method, required for the task_system.

• task_system.hpp : implements a task system based on this one but using a semaphore instead of a condition_variable, a spin_lock, and with new methods to stop and clear running tasks, and the added ability to add task parameters.

Inside main.cpp there's a simple example I used to benchmark my task_system against the one displayed in the youtube video. The idea is to dispatch a ton of extremely simple tasks, in this case decrement a std::latch, so that the difference in timing has to come from the tasking system used. Using hyperfine, the original task_system runs in 3.753 s ± 0.184 s, while mine in 2.092 s ± 0.552 s, a performance improvement of roughly 40-45%.