Re: Thread Pre-Emption Question . . .

Tech-Archive recommends: Speed Up your PC by fixing your registry

No matter what code he writes, he cannot guarantee a max latency of 40 ms.
He could write a driver and do all processing in an ISR and still not
guarantee that, or any, fixed latency. The ISR might not even start to
execute until more than 40 ms after the interrupt was raised.



"Pops" <dude@xxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:eI%23X8lIJIHA.5860@xxxxxxxxxxxxxxxxxxxxxxx
m wrote:

As to the OP's problem: My understanding was that he has an impossible
task, to provide a guarantee of execution
within 40 ms on Windows,

I disagree, especially if its interrupt driven.

This is not uncommon design requirement and there are many everyday
applications since the annals of computing history that require a smaller
residence time than 40ms, e.g, sockets and telecommunications.

It is when you run into the WIN32 QUANTUM limits, is when you begin to
need to consider RTOS designs. In other words, if he needed a residence
time of less than ~15ms, then he would be fighting an uphill battle with
WIN32.

But then again, he didn't say who is the audience and what further
specialization he needs, e.g, is this for "USERS", can it be a dedicated
server application? In such environments, he can use higher priverledges
or even use device drivers? Can he use embedded windows, CE?

The key though, and again, this design is par for the course, is that he
will need to queue the PUMP. Processing an interrupt, and/or event signal
moving the data to a queue SHOULD not take 40ms.

I don't wish to be disrespectful but if I need to further explain this,
then I see that I'm talking to a few individuals with little practical
process control experience and/or with large projects and product lines.
Maybe you do, but there has nothing that has been stated that would trump
my own input here.

Thanks

--
HLS


.

Relevant Pages

  • Re: USBD_STATUS not in header
    ... "Philip Lukidis" wrote: ... Add more latency to ... histogram of the ISR and DPC time for each of the drivers in my system. ... atapi.sys driver can take up to 6ms during its ISR. ...
    (microsoft.public.development.device.drivers)
  • Re: USBD_STATUS not in header
    ... >>> What does this code means in the context of an isochronous USB OUT and ... Add more latency ... > histogram of the ISR and DPC time for each of the drivers in my system. ... > atapi.sys driver can take up to 6ms during its ISR. ...
    (microsoft.public.development.device.drivers)
  • Re: Blind People Protest - Hybrid Cars Are Too Quiet
    ... The only way to GUARANTEE you won't harm anyone else is to not operate ... delivered (by a different driver, of course) and the emergency brake ... almost hit a pedestrian, as Nate did. ...
    (rec.bicycles.misc)
  • Re: Attempted summary of "RT patch acceptance" thread
    ... > linux-RTOS has a guaranteed worst case latency of 50usec. ... every single app that don't require a worst-case guarantee of ... send the line "unsubscribe linux-kernel" in ...
    (Linux-Kernel)
  • Re: New approach to transfer exact data By FT2232H
    ... Windows CANNOT guarantee a timed capture of high speed data, ... NO USB driver can buffer 10's or 100's MB ...
    (comp.arch.embedded)