On 10/10/06, Alan Stern <[EMAIL PROTECTED]> wrote:
> You're talking about detecting URB completion before a completion IRQ
> arrives.  In other words, polling.  I would prefer not to see that added
> to the HCDs, but it is a possibility.

Ah, EHCI (or USB in general) specifies a completion interrupt happens
only at the end of a frame?  I was suggesting the interrupt be thrown
as soon as the URB completes, even if that's mid-frame.

If that's not possible, than I guess it would mean polling, yes.  I'm
not enthusiastic about that either.


> > Given isoch and the design of low-latency devices, I wouldn't be
> > surprised if devices designed for low latency aren't already playing
> > the data in frame 3 of your scheme, as the samples pour in.
>
> The packet transfer may occur anywhere during the frame; the device would
> not be able to count on the data being available in time to play when the
> frame begins.  I suppose it could start playing the data a short time
> before the end of the frame, though; if the packet is going to arrive at
> all then it has to arrive before the frame ends.

Oh, good point.

> > I will point out that low latency applications generally consider the
> > machine to be an embedded appliance.  They own the box.  This is
> > accepted.
>
> If they own the box then they also own the OS.

Yes, at least, they have relatively uncontested priority to whatever
resources they desire.

It is common in my sound setups for the boxes to be multiuser; the
various stage managers have a login and run the controller software
remotely over ssh/ncurses.  The software on the controller box is
running realtime and the ncurses interface is used for interaction
(during shows, little more than the Go button).

> Make up your mind.  Do you want "as good as other systems that have been
> around a while" or do you want "a reliable 5ms"?  The two are not
> obviously compatible.

I want a reliable 2ms, but it sounds like I'm going to have to take
what I can bloody well get.  You and David have suggested (not stated
explicitly) that I'm probably going to have to be content with 5ms.

Monty

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
linux-usb-devel@lists.sourceforge.net
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to