Re: A question about period

[Takashi Iwai <tiwai@suse.de>]

On Fri, 03 Aug 2018 19:24:05 +0200,
Yu-hsuan Hsu wrote:
> 
> Sorry for my unclear question. The device usually provide the range of
> period_size to us, such as 16 to 1024 (We can get it from
> snd_pcm_hw_params_get_period_size_min and max). It means we can choose
> period_size 16 for this hardware device. However, the device only consumes
> 40 frames at each time.
> It only issue IRQ after it consumes 40 frames. My
> question is, whether it can support period_size 16 as it said? Or it need
> to change the range of period_size it provides.

In that case, allowing period_size = 16 is obviously wrong.
The practical minimum of period size is 40.  This must be set up via
either snd_pcm_hardware or some hw_constraints.


Takashi

> 
> Thank you for your patience and assistance.
> 
> yuhsuan
> 
> 
> On Fri, Aug 3, 2018 at 6:14 PM Takashi Iwai <tiwai@suse.de> wrote:
> 
> > On Fri, 03 Aug 2018 12:04:34 +0200,
> > Yu-hsuan Hsu wrote:
> > >
> > > For example, we can set period_size 32 to one device. But the device
> > > consumes 40 frames each time. Can we said it support period_size 32? I
> > > found many devices have this situation.
> >
> > The question remains: how can the hardware issue an IRQ of period size
> > 32 while it can consume 40 frames at each time?
> >
> > As you asked in the first post, the period size is the size where the
> > hardware notifies the update.  Your description above doesn't match
> > with the definition...
> >
> >
> > Takashi
> >
> > >
> > > On Fri, Aug 3, 2018 at 4:30 PM Takashi Iwai <tiwai@suse.de> wrote:
> > >
> > > > On Fri, 03 Aug 2018 10:09:30 +0200,
> > > > Yu-hsuan Hsu wrote:
> > > > >
> > > > > Thank for your answer!
> > > > >
> > > > > The other question is if we set period_size to N, but the frames it
> > > > > consumes once is 5N.
> > > >
> > > > Then the assumption is wrong.  How the hardware can issue an IRQ at N
> > > > consumption while it processes for 5N in once?
> > > >
> > > >
> > > > Takashi
> > > >
> > > > > Can it said it supports period_size N? How do we check
> > > > > whether the period_size can be set correctly?
> > > > >
> > > > > On Fri, Aug 3, 2018 at 3:31 PM Takashi Iwai <tiwai@suse.de> wrote:
> > > > >
> > > > > > On Fri, 03 Aug 2018 08:28:37 +0200,
> > > > > > Yu-hsuan Hsu wrote:
> > > > > > >
> > > > > > > Hi all,
> > > > > > >
> > > > > > > I have a question about the period we set in hw_params. I found
> > > > different
> > > > > > > boards may have different explanations about it. I have two
> > guesses
> > > > about
> > > > > > > its meaning.
> > > > > > >
> > > > > > > 1. The period_size is the size of each hardware's consumption.
> > If we
> > > > set
> > > > > > > period size to N, the pcm will consume N frames each time.
> > > > > > >
> > > > > > > 2. The period_size is the size to control when hardware call
> > > > interrupt.
> > > > > > If
> > > > > > > we set period size to N, the pcm consume frames in its step.
> > When the
> > > > > > > number of frames it consumes more than N, it will call interrupt.
> > > > > >
> > > > > > 2 is the correct answer.
> > > > > >
> > > > > >
> > > > > > HTH,
> > > > > >
> > > > > > Takashi
> > > > > >
> > > > > > > We can use snd_pcm_avail function to check the real available
> > frames
> > > > in
> > > > > > > the device. If guess 1 is correct, the size of consumption
> > should be
> > > > > > fixed.
> > > > > > > Else, setting period_size is nothing to do with hardware's
> > > > > > > consumption. I've checked some boards and found that each board
> > has
> > > > > > > different behavior (Most of them meet guess 2). I'm confuse which
> > > > one is
> > > > > > > correct. Thanks!
> > > > > > > _______________________________________________
> > > > > > > Alsa-devel mailing list
> > > > > > > Alsa-devel@alsa-project.org
> > > > > > > http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
> > > > > > >
> > > > > >
> > > > > [2  <text/html; UTF-8 (quoted-printable)>]
> > > > >
> > > >
> > > [2  <text/html; UTF-8 (quoted-printable)>]
> > >
> >
> [2  <text/html; UTF-8 (quoted-printable)>]
>