From: Peter Enderborg Subject: Re: Rawmidi bug or missed feature? Date: Sat, 02 Mar 2002 20:15:44 +0100 Message-ID: <3C8124E0.4FDDECD1@ufh.se> References: <200203021858.NAA13670@renoir.op.net> Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="------------E9092D81C8422CBA2BDEFA1B" Errors-To: alsa-devel-admin@lists.sourceforge.net List-Help: List-Post: List-Subscribe: , List-Unsubscribe: , List-Archive: To: Paul Davis Cc: "alsa-devel@lists.sourceforge.net" List-Id: alsa-devel@alsa-project.org --------------E9092D81C8422CBA2BDEFA1B Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Paul Davis wrote: > >How do you solve the problem with sharing hardware then? > > I don't. I intend to wait for (and contribute too, if I can) what I > consider the correct solution: > > a) sequencer genuinely split into: > > 1) a router/multiplexer > 2) a scheduler > > b) sequencer moves into user space > > As I've said before, I don't consider the current kernel-side > implementation of the sequencer, nor its merging of two completely > different functions, to be in anyone's interest. I think that Frank > was (like myself and everyone else) somewhat "blinded" by the fact > that OSS put the sequencer in the kernel, and we did not know that it > was possible to get similar performance in user space. How can we get the same performance i userspace? For me it is the processor/OS schedule that gives the limit for that, and in kernel we get the hardware as the limit. > > > For the time being, I have hardware solutions that work for me when I > need to do complex MIDI routing (i.e. dedicated h/w MIDI patchbays), > and I never run multiple apps using the same port. > > --p -- foo! --------------E9092D81C8422CBA2BDEFA1B Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Paul Davis wrote:
>How do you solve the problem with sharing hardware then?

I don't. I intend to wait for (and contribute too, if I can) what I
consider the correct solution:

  a) sequencer genuinely split into:

          1) a router/multiplexer
          2) a scheduler

  b) sequencer moves into user space

As I've said before, I don't consider the current kernel-side
implementation of the sequencer, nor its merging of two completely
different functions, to be in anyone's interest. I think that Frank
was (like myself and everyone else) somewhat "blinded" by the fact
that OSS put the sequencer in the kernel, and we did not know that it
was possible to get similar performance in user space.

How can we get the same performance i userspace? For me it is the
processor/OS schedule that gives the limit for that, and in kernel we get
the hardware as the limit.
 

For the time being, I have hardware solutions that work for me when I
need to do complex MIDI routing (i.e. dedicated h/w MIDI patchbays),
and I never run multiple apps using the same port.

--p

-- 
foo!
  --------------E9092D81C8422CBA2BDEFA1B-- _______________________________________________ Alsa-devel mailing list Alsa-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/alsa-devel