[sdiy] The penultimate soft DIY platform?

[Jerry Gray-Eskue]

<<> There has been all these discussions about FPGA this, TI/AD/MOT DSP 
> that, dsPIC here, over clocked AVR there, Beagleboard here and so on, 
> i mean most of these approaches are dead ends, right?>>

Umm not really, this stuff can be moved to other target systems without a
lot of hassle, especially the stuff written in the C language. I have
written various code packages in C with porting in mind and have had the
code up and running on an entirely different target system in as little as a
few hours. 

One particular piece of code without ANY change to the core code is running
on 2 different 68302 custom embedded system in one of these cases without
any operating system the other a custom Unix like kernel, Windows, AIX, DEC,
SCO, SUN, LINUX, and was recently ported to yet another custom embedded
system in less than a day by a second party working only from a porting
guide I supplied.

- Jerry  

-----Original Message-----
From: synth-diy-bounces at dropmix.xs4all.nl
[mailto:synth-diy-bounces at dropmix.xs4all.nl] On Behalf Of Noah Vawter
Sent: Thursday, March 10, 2011 9:43 PM
To: karl dalen
Cc: synth-diy at dropmix.xs4all.nl
Subject: Re: [sdiy] The penultimate soft DIY platform?


	If I may be so bold, I support your idea, but it seems that synth
dev  
follows instrument dev in that people tend to build them out of  
whatever they have handy.  And to confound that matter, culture steers  
us toward re-creating specific designs rather than a wide variety.  
e.g.  in the U.S. we keep making guitars, in Russia they keep making  
balalaikas even though they're both just common string instruments.

	Nevertheless, there are some standards which simplify this whole  
process and I support them wholeheartedly.  Publishing source code is  
probably the most successful, although simply switching between int  
and float implementations and other necessary optimizations tends to  
fragment even this practice.

	I do wish there were a way to avoid the dead ends you mention.  It  
seems the life of any one of these devices gets extended the more  
people share information about them.

	The other thing that evades any single solution is that people  
whimsically change context...  The moment we get a cool synth, someone  
wants to put it in their car, or walkman, or underwater...  Or  
suddenly quality needs to be improved....  I remember being flamed at  
2400 baud in 1989 for (in jest) saying that 8-bits wasn't enough.   
Well now, 16-bits isn't enough and 96 kHz isn't enough.  Quadrophonic  
isn't enough.

	One notable effort a friend of mine did was the AudioPint...  a mini

PC running Linux embedded in a Pelican case with mic pre-amps.  You  
can see some pics here:  http://audiopint.org/gallery   But it  
honestly wasn't perfect for my purposes...

	I do like your suggestion to include development tools inside the  
system.  That single act would catapult development forward.  It would  
be nice if there were an OS that was "freezable" - that didn't need to  
be updated.  It would be almost as good as buying a laptop and never  
using it for anything else.


On Mar 10, 2011, at 9:48 PM, karl dalen wrote:

> Well, this has been up here some years ago, i tought i make retake
> at it once again, so why havent anyone made a open surce Music/Sound
> DSP platform?
>
> There has been all these discussions about FPGA this, TI/AD/MOT DSP
> that, dsPIC here, over clocked AVR there, Beagleboard here and so
> on, i mean most of these approaches are dead ends, right?
>
> A dual Atom core mini ITX or so with everything on board with a
> dedicated audio, im not saying OS but a simple audio frame work
> firmware that are dedicated to audio only could easily outperform
> most dedicated solder-yourself-attempts-BGA-2000-balls + extra-hassle
> one-wendor-FPGA-lock-yourself-into-tool-stuff and such.
>
> I mean something that allows the developer sync to USB subframes
> with 4 lines of code, 16/32 sample buffering audio and some kind of
> API that simulates a big donkey OS so one could perhaps, maybe run
> the regular standard development tools...No just a frame firmware
> who just sits and wait for code from your regular PC who runs the
> regular dev tool set whatever Cobol ,Fortran version you may use.
>
> That board could easily be any Chinese sub 100 euro, and when it
> becomes obsolete you lift the code over to whatever new there are.
>
> Yes there would be software/hardware implications but could be
> solved due to the lack of insane stack of OS based extras runtime
> whatever code, one just dont need drivers for everything that's on
> the board.
>
> One problem (HW) i can see are the closed system aproach regular
> PC's have but many of these ITX boards have some openings with
> integrated CF card reader HW trough a bridge, it may not be
> blazing fast but for interfaceing your own HW without aditional
> 8bit PIC's on a UART kind of thing it could be well enough.
>
> Quite frankly i think Korg is doing this with their Kronos
> at least they uses a dual Atom core platform. ;)
>
> Regards
> KD
>
>
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy

_______________________________________________
Synth-diy mailing list
Synth-diy at dropmix.xs4all.nl
http://dropmix.xs4all.nl/mailman/listinfo/synth-diy