Messages in this thread | | | From | Bill Huey <> | Subject | Re: real-time threaded IO with low latency (audio) | Date | Sun, 25 Jul 1999 16:00:06 -0700 (PDT) |
| |
> So, what you need is instant response to "start recording"? If so, a > suffuciently large circular buffer in mlock()ed memory should be enough, > I think. Have a lower priority disk I/O task do the writing to disk. > > I think I'm missing the problem here, really... Bandwidth?
Nope, you got it correct. :-P
I didn't think about using the solution you've outlined above, which is quite excellent if the low priority I/O task doesn't interfere with any soft realtime tasks.
But part of the line of thought was to get around the normal overhead of the FS layer and data structure manipulations.
Using a dedicated partition was the first idea that came to mind and then unification of that feature with the swap was the other.
The data structures that manage the swap seem to be more predictable and simpler to deal with than the entire FS layer. This including the management of the swap system.
What I'm thinking about is kind of functional overloading the of the swap system so that folks [would *not*] have to use a dedicated spooling parition.
I could be on crack about this issue. I'm a kernel newbie.
> Ok, no extreme bandwidth and/or real time requirements then? This
Yeah, not for this application. The problems for me are related to dropouts, scheduling of realtime threading decoding without funny latencies and side effect associated with using various kernel subsytems (network, disk, etc...).
[Stuff about disk seeking and a multimedia file system]
Yeah, I know. SGI's got some paper on video serving that might be relevant for constructing a design proposal or something for realtime streaming to disk.
> //David
bill
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |