How to control multicore in C
category: general [glöplog]
Quote:
(and perhaps rightfully so, chaos' advice is probably the best for most developers)
meant the first post on this page. second one goes in a good direction :) i've seen people using fibers with some good results.
Quote:
Actually Intel's Thread Building Blocks stuff (that the guy talked about on Evoke) sounds pretty neat if you want to distribute workload in a nice way. It's only meant for actual computing work tho, so don't expect it to solve all of your sync and latency problems :)
kb: yes, Intel TBB exactly use a work-stealing scheduling system like outlined above under the hood. They also put a lot of higher-level stuff (pipelines, split&gather, etc.) on top.
Actually, there are a lot of similar solutions out there who mainly differ only slightly in the programming interface - including our own solution (main focus on embedded systems, but also implemented a Cell/ps3 version and others).
Because of this we made some effort to promote an API standard for this kind of libraries..
[quote=jar_]OpenMP is pretty much useless for more complex stuff other than simple loop constructs.[/quote]
Sure... it's not very powerful, but I wouldn't say useless. I just wanted to mention it since I couldn't find any other reference to it in this thread.
I've done some distributed utilities between multicore computers, and prefer a combination of Pthreads and MPI myself.
Sure... it's not very powerful, but I wouldn't say useless. I just wanted to mention it since I couldn't find any other reference to it in this thread.
I've done some distributed utilities between multicore computers, and prefer a combination of Pthreads and MPI myself.
hmm... seems like I should have read the BB Code standard before I posted. :p
zmurf: OpenMP has been mentioned multiple times already.
kus ma bite: Yepp.. I see that know... fuck! I have to learn how to read. :p
multicore: 60% - DX10 GPU: 46% - both high numbers, gamers don't know the crisis :D