On Monday 09 October 2006 9:43 am, Christopher "Monty" Montgomery wrote: > On 10/9/06, David Brownell <[EMAIL PROTECTED]> wrote: > > On Monday 09 October 2006 9:02 am, Christopher "Monty" Montgomery wrote: > > > On 10/9/06, David Brownell <[EMAIL PROTECTED]> wrote: > > > > > > > Exactly why I'm reminding you that your driver -- queueing only one > > > > URB at a time -- is the source of a **LOT** of your problemss. > > > > > > I've not heard any other plausible suggestions other than repeated > > > assertions that 'Linux can't do that'. > > > > I fail to understand how "do it THIS way not THAT way" is anything > > other than a plausible suggestion.... > > So far, it has been "do it THIS way not THAT way" inoring the fact > that "THIS way" doesn't work. Cannot work.
Demonstrably can and does work. I'm not sure where the communications problem is coming from, but it sure feel like it's not me ... > You have not dusputed the > fact that 4ms of buffering requirement precludes 2ms of latency. Make up your mind, first ... is it 5 msec latency? Or 2 msec? And I sure thought that "2ms in one URB" and "2ms in the next" is fairly clear. If the pipeline always has at least 2ms buffered, then you'll be able to tolerate 2ms latency for the USB IRQ; QED. > "I need 2ms latency". > "Do it the way we're doing it." > "Uhh.. that is barely working for 50ms latency right now. And even if > we got optimal performance, it could never theoretically get down to > 2ms." > "Too bad. This is The Way." You have not ** YET ** tried to do it the way that's known to work, so it's no surprise you've been seeing the failures which lead to the requirement to use double buffering. >From where I sit, I see your response as "I refuse to even consider using that known-to-work solution, it's My Way Or the Highway." ------------------------------------------------------------------------- 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
