RE: [oc] Handel-C replacement, name is done now for the coding.

["Paul McFeeters" <paul.mcfeeters@ntlworld.com>]

Martin,

<snip bit about CPU+Linux Kernel on chip>

> Sounds very ambitious - I'd start by flashing some lights :-)  I have to
say the Celoxica's
> video demos are *very* impressive, generating VGA real-time from C
impresses the pants off
> most softies, so a demo board with a RAMDAC on would be nice...

But I am ambitious, yes I have flashed lights (not in my C program yet just
VHDL) but I like
to dream of tomorrow and make it happen. Did it for 14 years as a programmer
and only failed
twice (pesky fractal image compression using higher maths than I ever
studied!). Admittedly it
was a few years ago but 2 years spent working on new compilers should help
me out with parsing C.
A couple of resisters attached to an FPGA and I will have my own 12bit
(hopefully soon 24bit)
VGA interface according to the designs I've found. I was hoping to use the
Block RAM SRAM units
but at only 512bytes each my initial screens would be rather blocky  shall
we say? Monochrome I
could do say 70 x 50 pixels off one but will use them for higher resolutions
for now. That
reminds me Tony has a pre-made VGA connector for my boards so will look to
order one or two of
those for later.

> They generate directly to EDIF normally, although they do have a VHDL
output flow now.  YOu
> might get some info from some of the papers written about Handel-C when it
was still an academic
> thing.

I did try to find some of them but Oxford (where the academic papers were
written have removed them
probably at the bequest of Mr. Page). I did place call for XNF and EDIF
information but I suspect
going to Verilog would be so much easier than straight into EDIF/XNF for now
(if not pls email me).
I'm studying Verilog at the moment as VHDL (or maybe just Webpack) is giving
me so much grief its
unreal. Even a simple 8 stage FSM I did, clocked at 1Hz doesn't work right
with Webpack not to
mention some of the freaky errors it likes to produce at perfectly legal
things like using
functions or procedures, I don't need extra headaches during development
phases.

> See my other reply on sequential processing...

Probably already have, sequential processing is rather akin to the first
single cylinder steam engine.
Yes it did open up a world of possibilities but theres only so much power
you can get from one cylinder
before you think maybe more cylinders would be better for this? Imagine if
car engines were only allowed
to have one cylinder! I've driven French cars and believe me it wouldn't be
good! lol

> You never know...  Of course, if you open the source, Celoxica can nick
your ideas, even without nicking
> the source-code can't they?

The idea of opening the source will be for everybody to be able to use it
(or nick it as you call it) but
then how would Celoxica and others expect to charge so much for something
thats 'public property'? Do you
think that people would get away with charging $25,000 for a copy of the
Linux Kernel? I will be a little while before I open the source even in
ideal conditions alas. Apart from writing it I will have to check
that any technique I use doesn't infringe on copyrights from my previous
programming days, then I have to
clean it up. Don't want everyday laughing at me do I? But my main point is
$25,000 a year for any program
is just taking the P*** out of us. If thats what they think their program is
worth then they can charge
that for it, if I can do mine and make it open source then I will. I think
more people should be writing
programs in hardware rather than programs that run on programs (CPU cores)
in hardware (no offence Damjan
lol). Will also boost the parallelism of code which I am a big fan of.

> >Handel-C is copyright of Celoxica, Buffy-C hmmm might be copyright of me
;-)

> I've never objected to Handel-C actually :-)  You maybe want to find
something completely different...
> or not as you like!

Handel-C is copyright of Mr. Page or Celoxica, whichever obviously we can't
use it, another company does
SystemC and I think theres also CycleC but I have an idea for a name (won't
tempt name-claim-jumpers by
mentioning it but its pretty good).

> Cheers,
> Martin

All input gratefully received as always.

Paul McFeeters


--
To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml