Hi Pavel,
I don't think so, but I can be wrong....
These are the times i have defined:
VOID
DpxMttFrameArrivalEvtTimerFunction(
IN WDFTIMER Timer
)
{
PDEVICE_EXTENSION devContext =
GetDeviceContext(WdfTimerGetParentObject(Timer));
// this timer handler looks for new frames to process and process them
// the frames come from usb at a 30 hz frame. the timer is at the double of
the //frequency
if (devContext->Context.HasNewFrame == 1)
{
// Process frame
devContext->Context.HasNewFrame = 0;
}
// 66 hz timer
WdfTimerStart(Timer,
WDF_REL_TIMEOUT_IN_MS(15)
);
}
VOID
DpxMttReadReportEvtTimerFunction(
IN WDFTIMER Timer
)
{
PDEVICE_EXTENSION devContext =
GetDeviceContext(WdfTimerGetParentObject(Timer));
// this timer looks for new touch information to inject in windows subsystem
at a 66 hz rate
DpxMttCompleteReadReport(WdfTimerGetParentObject(T imer));
WdfTimerStart(Timer,
WDF_REL_TIMEOUT_IN_MS(15)
);
}
Any ideas?
Thanks,
Nuno
"Pavel A." wrote:
> A timer resolution issue again? Do you use delays/time waits somewhere?
> --pa
>
>
>
> "sinosoidal" <> wrote in message
> news:77709963-9F90-4AEA-933F-...
> > Hi,
> >
> > I'm having a lack of responsiveness problem in the multitouch driver i'm
> > developing.
> >
> > in order to have decent feedback i need to move my finger thru the surface
> > very slowly.
> >
> > in my test plataform (linux) where I develop the code before putting it in
> > the device driver (user mode) i have suberb responsiveness,
> >
> > but when i put it into the kernel, it really goest down.
> >
> > i think it is somwhow related with the injection of the touch information
> > because, even in windows, my test plataform needs to get the frames from
> > the
> > device from the driver, and i can get good performances results this way
> >
> > the problem is on the touch information that is sent by the driver to the
> > windows touch subsystem.
> >
> > can anyone more experienced in this kind of drivers point me out typical
> > implementation problems?
> >
> > I currently have a serial mode touch report of 10 contacts in my
> > descriptor
> >
> > and i'm reporting frames at 30 hertz.
> >
> > i also thought that this problem could be from the 30 hz but i have tested
> > for 65 and the problem continues, so i'm dropping the idea that the frame
> > rate could be a problem.
> >
> > i have also used kernrate to see how much the driver was spending (6% in a
> > dual core 2.5 Ghz machine)
> >
> > is there any way of defining the priority a driver has in kernel?
> >
> > any ideas?
> >
> > with my best regards,
> >
> > Nuno
>
|
Similar Threads
Linear Mode
