Persistent Kernel implementation, trying to reproduce results from the article: https://www.concurrent-rt.com/wp-content/uploads/2016/09/Improving-Real-Time-Performance-With-CUDA-Persistent-Threads.pdf
nvcc compiler and MPS daemon running (so sudo rights) If the execution is corrupted, one must stop the MPS server and restart again
-> To test, launch mult first, and in another terminal, run add <-