On Wed, 17 Oct 2001, Rodolphe Ortalo wrote: > Yes? Well... > > Do you have more specific pointers? (Either web, research papers, etc.)
I haven't got to the point of figuring out what the specific term for our problem is in realtime-eese, but a google search for "rate monotonic linux realtime" is a good direction to push off in. To throw in some more terms: Our range fitting algorithm is different in that if we converted it into real-time terms (i.e. pixels/bytes => timeslices, textures => tasks) we are interested in "non-harmonic" rate monotonic scheduling with a hard guarantee and reclaiming timeslices when tasks are canceled. (Of course, our "time" axis doesn't actually progress, so any features of an algorithm that carry state during time progression don't really help us. :-) BTW, Eric has had a go at implementing a basic first cut. -- Brian
