> On 13 Sep 2018, at 21:30, Akim Demaille <a...@lrde.epita.fr> wrote: > >> Le 13 sept. 2018 à 20:56, Hans Åberg <haber...@telia.com> a écrit : >> >>> And the model is really different, >>> interruptions are controlled with coroutines, while threads are >>> preemptive. These two models are not equivalent at all imho. >> >> I found coroutines not so intuitive, so threads would be better if one can >> accept some overhead. Also, threads can be run a pool to lessen the >> overhead. This way, I made a tree-engine as opposed to a sequence of >> stackengines. > > No possible comparison. You can have 100.000 of coroutines > which any problems. And again, the sheer fact that you decide > when you yield to another coroutine changes everything; e.g., > most of the time you don’t have to lock.
The libcppsim library uses real coroutines and not threads according to the paper [1]. Download [2-3]. 1. https://pdfs.semanticscholar.org/a446/b51798dc776b0b1c88b562907a90f6a66de3.pdf 2. https://www.moreno.marzolla.name/software/libcppsim-0.2.5.tar.gz 3. https://www.moreno.marzolla.name/software/
