| Message ID | 20130919112043.5ad16147@armhf (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Thu, Sep 19, 2013 at 11:20:43AM +0200, Jean-Francois Moine wrote: > This patch extends the min and max number of bytes per period. > It mainly permits to reduce the sound delay in MIDI real-time playing. Applied, thanks. For the minimum limit is there any hardware imposed limit that could be used rather than tweaking the numbers?
On Thu, Sep 26, 2013 at 11:24:08AM +0100, Mark Brown wrote: > On Thu, Sep 19, 2013 at 11:20:43AM +0200, Jean-Francois Moine wrote: > > This patch extends the min and max number of bytes per period. > > It mainly permits to reduce the sound delay in MIDI real-time playing. > > Applied, thanks. For the minimum limit is there any hardware imposed > limit that could be used rather than tweaking the numbers? I looked at that, and apart from interrupt rate, I don't see any. The values I have for these in my tree are: #define KIRKWOOD_SND_MIN_PERIODS 2 #define KIRKWOOD_SND_MAX_PERIODS 16 #define KIRKWOOD_SND_MIN_PERIOD_BYTES 256 #define KIRKWOOD_SND_MAX_PERIOD_BYTES 0x100000 #define KIRKWOOD_SND_MAX_BUFFER_BYTES 0x100000 which are partly based on patches applied to Rabeeh's kernel, and then further adjusted. Yes, the max period bytes won't be reachable since the minimum period will cause it to be half the max buffer bytes.
On Thu, Sep 26, 2013 at 11:28:40AM +0100, Russell King - ARM Linux wrote: > The values I have for these in my tree are: > #define KIRKWOOD_SND_MIN_PERIODS 2 > #define KIRKWOOD_SND_MAX_PERIODS 16 > #define KIRKWOOD_SND_MIN_PERIOD_BYTES 256 > #define KIRKWOOD_SND_MAX_PERIOD_BYTES 0x100000 > #define KIRKWOOD_SND_MAX_BUFFER_BYTES 0x100000 > which are partly based on patches applied to Rabeeh's kernel, and then > further adjusted. Yes, the max period bytes won't be reachable since > the minimum period will cause it to be half the max buffer bytes. Those don't look implausible, obviously it'd be pretty CPU intensive application that pushed down to the minima but perhaps someone is willing to pay the cost. As you say the maximum buffer might want tweaking (or perhaps the maximum period bytes).
diff --git a/sound/soc/kirkwood/kirkwood.h b/sound/soc/kirkwood/kirkwood.h index f8e1ccc..bf23afb 100644 --- a/sound/soc/kirkwood/kirkwood.h +++ b/sound/soc/kirkwood/kirkwood.h @@ -123,8 +123,8 @@ /* need to find where they come from */ #define KIRKWOOD_SND_MIN_PERIODS 8 #define KIRKWOOD_SND_MAX_PERIODS 16 -#define KIRKWOOD_SND_MIN_PERIOD_BYTES 0x4000 -#define KIRKWOOD_SND_MAX_PERIOD_BYTES 0x4000 +#define KIRKWOOD_SND_MIN_PERIOD_BYTES 0x800 +#define KIRKWOOD_SND_MAX_PERIOD_BYTES 0x8000 #define KIRKWOOD_SND_MAX_BUFFER_BYTES (KIRKWOOD_SND_MAX_PERIOD_BYTES \ * KIRKWOOD_SND_MAX_PERIODS)
This patch extends the min and max number of bytes per period. It mainly permits to reduce the sound delay in MIDI real-time playing. Signed-off-by: Jean-Francois Moine <moinejf@free.fr> --- sound/soc/kirkwood/kirkwood.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)