Hello all! Now that the train seems to have caught a locomotive or two in it's front-way, I've been lurking the forums for long enough at once to talk again.
Lately, after beeing presented with the magnificient productivity - performance problem by himself, the comunity has sprouted a great deal of recommendations, references to articles and posts about parallel computing, multithreading and the likes. It made me want to fiddle with that sort of library too.
Of course, is going to make his own implementation (if he hasn't already), superior and tailor-fitted for the LT base engine, when he finally sets for a language.
But seeing that many around here know an actual fair share of multithreading and/or the game industry inner workings, I cannot restrain myself from asking:
What libaries do you know that allow you to build a portable (linux - windows) multithreaded task scheduler library in C++ robust/flexible enough to deal with the uglinesses of game engines?
Heck, be it even "just" a portable multithreaded for-loop implementation.
I've found around here (forget where, sorry D: ) the magnificient Cascade PPF Paper, but I haven't been able to find the library itself. Hence, this post.
As always, all the creative smog that infests this place numbs minds and provokes delusional thoughts (that's a compliment there); so I'll open a post on the off-topic game section with the details on the wanna-be game this would be used on if any of you has sincere interest in knowing about it. (Because if opening such a post means ending up throwing a full day at it, as it's first public notice demands, and that's not gonna happen without provocation).
Tue Feb 07, 2017 1:15 am
I used OpenMP (not to be confused with MPI) on my final project for a class last semester. It's not as robust and flexible as pthreads or MPI but is way, way easier to use it in the cases where it's usable.
Tue Feb 07, 2017 3:20 am
I'm having a look at OpenMP, after your recomendatins. I'll start simple. But this beeing a game engine, parallelizing just whole for loops might not be optimal. Still in the (parallel ... hehe) search for the library used in the article or similar.
May luck be by your side for the matters not under your responsability.