View this page "Project: HR Hack'n Crack (DSP56858)"
Mitch Davis
development environment.
Click on http://groups.google.com/group/connected-community-hackerspace/web/project-hr-hackn-crack-dsp56858
- or copy & paste it into your browser's address bar if that doesn't
work.
Mitch Davis
non-free-as-in-libré toolchain.
Richard Andrews
Looks like too much trouble.
There is mention in mail archives of an ancient fork of gcc-1.37 for
the motorola 56k instruction set called gc-56k - but I've failed to
locate any copies. I did manage to find a GPLd assembler for 56k
called a56 but it looks unmaintained since c 1995.
Mitch Davis
> There is a free-as-in-beer "evaluation version" with code size restrictions.
> Looks like too much trouble.
Looks like free pain.
If you're after a conventional DSP board, Olimex have a TMS6722 DSP
eval board in development:
http://www.olimex.com/dev/ (look for TMS320-PC6722)
Olimex prices are generally very good.
If you are after a programmable DSP that is designed to be an effects
processor, you may like to have a look at this device:
http://www.spinsemi.com/
http://www.spinsemi.com/products.html
http://www.spinsemi.com/Products/datasheets/spn1001/FV-1.pdf
You can't beat the price for a board:
http://www.oct-distribution.com/
Mitch.
Clifford Heath
developers (his designs have been made into ASICs, etc), and he had
the following to say about the 56K:
----- QUOTE -----
If the hardware works and has lots of useful IO, _possibly_ worth
their trouble.
We used a 56K DSP at Martin Comms - ~20 years ago!
The compiler was a DOS based thing and relied on a phar lap memory
manager/extender or some such crap. Eerrrrk!
Found this: http://www.bdti.com/faq/3.htm
Most modern DSPs have a built in debug/port, so I'd be inclined to go
for something like an EZKIT-Lite from AD.
Their windows tools include a proper IDE, etc.
----- END QUOTE -----
In my opinion, built-in debug support is mandatory *especially* for
hobbyists.
Your time might not be worth much, but you'll spend an awful lot of it
without a debugger.
Clifford Heath.
> --
> You received this message because you are subscribed to the Google
> Groups "Connected Community HackerSpace" group.
> To post to this group, send email to connected-commu...@googlegroups.com
> .
> To unsubscribe from this group, send email to connected-community-h...@googlegroups.com
> .
> For more options, visit this group at http://groups.google.com/group/connected-community-hackerspace?hl=en
> .
>
Richard Andrews
The chip does have a built-in JTAG port with on-chip debugging.
I've found a patch on GNU binutils to add 56k but it doesn't compile.
The barrier to entry is looking pretty high. I'm pretty sure I'll toss
this in the seemed-like-a-good-idea-at-the-time bucket.
Clifford Heath
development
tools, incase he needs to update the firmware in some old hardware he
built. It's a "commercial" compiler derived from gcc, which means it's
freely
available if you want - just ask (through me - except I'm away for 3
weeks).
He actually said (of the compiler):
I still have this around somewhere should you wish. In case I ever
feel like
recompiling the foxhunt DSP code. It had some bugs (one notably not
saving
sufficient context for ISRs). David no longer uses the 56k DSP board.
It's a gcc port done by Motorola, and then sold by Motorola in a dubious
un-GPL fashion !
There is also a debugger (someone else wrote) that uses the serial port;
superceeded by the better hardware debug port port that came with
later 56ks
(560002 on ?).
Clifford Heath.
Richard Andrews
Assuming I can hack it into a buildable state for a modern toolchain
bootstrap, it would be enough to convince me to have another look at
the boards.
Richard Andrews
> http://www.olimex.com/dev/ (look for TMS320-PC6722)
>
> Olimex prices are generally very good.
>
Both good hardware but the toolchain is the problem again. Both only
have closed toolchains.
Mitch Davis
Curious. The spinsemi.com home page says "Integrated IDE is FREE" and
"Open source library, ready to go". Can you point me to the "bad
news" page?
Mitch.
michaelc
post a link? All I could find was a Windows XP gui tool for compiling
assembler, with no source code for the assembler in sight. I'm not
sure if you can even buy Windows XP any more, but I know it used to
cost a least a couple of hundred dollars: that's a long way from free.
The SpinAsm user manual (http://www.spinsemi.com/Products/datasheets/
spn1001-dev/SPINAsmUserManual.pdf) seems to have instruction coding
included, so it might not be so hard to write your own assembler.
Michael
On Feb 27, 12:18 am, Mitch Davis <m...@afork.com> wrote:
Richard Andrews
Richard Andrews
builds when I'm not falling asleep).
https://sourceforge.net/projects/dsp56800/
This project advertises the binutils packages but delving deeper (all
files) there is also what appears to be a patch on gcc-3.3.2 which the
README say can be built using said binutils. There is also what
advertises itself as a JTAG transport and flash programmer built on
linux parallel port. This might be hard to get working. Parallel is
mostly abandonned these days. My home PC does have one.
Even without JTAG if I can get a mildly sensible 568xx ELF out of gcc
I should be able to use objdump to build a program image and stream it
into the serial port at boot.
Richard Andrews
patched for dsp56800. Back to assembler I guess.
Mitch Davis
> No such luck. gcc-3.3.2 doesn't build for me - even without being
> patched for dsp56800.
How did it fail? Can you dump a build transcript somewhere?
> Back to assembler I guess.
That's probably your best best for a blinkenled hello world program anyway.
Mitch.
Richard Andrews
> On Sun, Mar 7, 2010 at 9:22 PM, Richard Andrews <bflat...@gmail.com> wrote:
>> No such luck. gcc-3.3.2 doesn't build for me - even without being
>> patched for dsp56800.
>
> How did it fail? Can you dump a build transcript somewhere?
First it fails to compile because of cast code which is flagged by
gcc-4 as an error.
Reworking the offending cast leads to a segfault later in the
bootstrap process. I already deleted the build directory in disgust.
If anyone is really curious I can redo it.
>> Back to assembler I guess.
>
> That's probably your best best for a blinkenled hello world program anyway.
Yep - should be sufficient for that. A lot of flailing in the dark though.