Subject: Re: [linux-audio-dev] [ANN] A novel approach to real-time free software
From: David Olofson (david_AT_gardena.net)
Date: Mon Jun 10 2002 - 03:05:56 EEST
On Saturday 08 June 2002 12.32, Ingo Oeser wrote:
> On Sat, Jun 08, 2002 at 01:11:10AM +0200, David Olofson wrote:
> > (Software patents should be invalidated and made illegal world
> > wide, IMHO. They serve only the interests of the biggest
> > companies.)
>
> Nod, partially. Just invalidate after 2-3 year on the market
> and after 1-2 year(s) if not marketed.
Yes, something like that. Despite being a strong believer in
Free/Open Software (to the extent that I'd rather stop making money
on coding than giving up my freedom), I think it might be nice if
closed source developers could use patents as a way of securing a
"window".
Eventually, though, I think hardware/solution oriented companies will
start complaining about closed source developers holding patents that
prevent new technology from being used at all in Free/Open Source
software.
If we can't avoid software patents entirely (I certainly don't want
them, but I can understand those who do), maybe software patents
should guarantee special rights to Free/Open Source projects by
definition? (As it is now, I'm afraid someone like Microsoft could
start suing Free/Open Source projects - whether there are clear
patent infringements or not - to scare the community to death. What
Free/Open Source project would go to court against Microsoft, even if
they would *most likely* win?)
[...]
> > From the application POV, both approaches give the same result:
> > Two "worlds" with different schedulers and services - and
> > unfortunately, different drivers. That is, if you want real time
> > I/O, you still have to port Linux drivers to RTAI/ADEOS [...].
>
> That is intended. You are encouraged to write hybrid RT/Non-RT
> programs by limiting you to a small core that is RT and a
> full-blown (Non-RT-) OS, that provides you with all additional
> services.
Yes, I'm perfectly aware of that - I'm the one who suggested running
audio under an RTK in the first place, and even started hacking a
"Driver Programming Interface" to make porting drivers from Linux to
RTL easier.
The problem is that getting people to use Linux/lowlatency + user
space applications is hard enough already, and as to driver
development/porting, getting ALSA into mainstream kernels seems to be
quite enough work, without also having to port ALSA to RTAI/ADEOS.
That said, there's always mmap() mode and hooking the audio card IRQ
with an RTAI ISR... :-)
> <shameless plug>
> Our own project[1] is going even further by splitting the RT part
> into a DSP program and a host program doing Non-RT-tasks. That
> way nobody can beat you on low-latency, if you get your hands on
> the right DSPs[2] ;-)
> </shameless plug>
RTLinux or RTAI is quite sufficient to work with latencies
corresponding to a fraction of the minimum PCI DMA burst size, so
latency is not a good reason to use a DSP. (Even a good
Linux/lowlatency setup seems to be sufficient to hit the hardware
limits of most PCI cards.)
Processing power/heat ratio is a *very* good reason, though. :-)
//David
.- M A I A -------------------------------------------------.
| Multimedia Application Integration Architecture |
| A Free/Open Source Plugin API for Professional Multimedia |
`---------------------------> http://www.linuxdj.com/maia/ -'
.- David Olofson -------------------------------------------.
| Audio Hacker - Open Source Advocate - Singer - Songwriter |
`-------------------------------------> http://olofson.net -'
This archive was generated by hypermail 2b28 : Tue Jun 11 2002 - 04:03:33 EEST