Subject: Re: [linux-audio-dev] Gigasampler Clone
From: Benno Senoner (sbenno_AT_gardena.net)
Date: Sat Jul 01 2000 - 03:18:24 EEST
On Sat, 01 Jul 2000, Juhana Sadeharju wrote:
> >From: Benno Senoner <sbenno_AT_gardena.net>
> >
> >I will use the usual technique: PREFETCHING
> >
> >And getting a "cease-and-desist" letter for having implemented
> >a simple opensource prefetching technique (which using mmap() can even do
> >implicitly, because the kernel does it ) whould be kinda cool.
>
> The patented system is not about prefetching. The beginnings also have to
> be kept in the memory. There is two parts: (i) keep beginnings in the
> memory and (ii) stream audio from disk (with possible prefetching).
>
> For (i) I found prior art, and (ii) is old stuff. Together, these
> can make, for example, a multitrack application (mentioned in the prior
> art), or sampler (no prior art so far).
>
> It is not a trivial patent.
>
> Juhana
but look it at this way:
mmap() the entire sample in memory
mlock() the first part
and then as the sound get triggered just memcpy() !! the
data from the mmapped buffer to the soundcard buffer
and mlock() ahead a couple of KBytes in order to let the
disk shuffle in the pages into mem.
this loop is so simple that even a common sorting algorithm
looks much more complex.
But I don't understand , you as a Finnish guy , why do you worry
so much about patents ?
If we get sued by implementing the above mmap() algorithm,
then they MUST sue unix/linux developer for having invented mmap().
:-)
Benno.
This archive was generated by hypermail 2b28 : Sat Jul 01 2000 - 02:42:46 EEST