[LAD] RME madi latency

[Adrian Knoth]

On 02/03/16 14:04, Fokke de Jong wrote:

> Hi all,

Hi!

> I have a RME madi fx card, but i found out that the minimal buffersize
> for those cards is 8192 samples, which is way to big for my use.

Hardware buffer size. This buffer is divided into sub-buffers (periods)
depending on what you configure.

If you choose 32 samples, then it's 8192/32 == 256 periods that fit into
one buffer. You get an interrupt every 32 samples, and ALSA reads the
just completed sub-buffer.

The HW buffer size really has zero relation to your RTT.

Spin up jackd and then use Fons' jack_delay. Maybe we have to add
buffer_size_min to the driver, but I'm pretty sure returning the card is
the entirely wrong approach.


HTH