[Oberon] ProjectOberon on FPGA

[Jan de Kruyf]

Skulski, now its nearly zoom off time. Pff what a week.

> A loosely coupled band of enthusiasts is unlikely to agree on
> this, or on anything else for that matter

I am aware of the parallel boards,but I have not much use for them,
lets say I forgot a little bit. So yes I agree. I was merely having some
wild idea
and I do not really know where your thinking is or is going.
I merely respond to what I read here.

On the quote: Yes and  no, it depends who is in charge.
the Linux kernel group is especially blessed I suppose,
notwithstanding the moans from time to time.

A company is good for making money, but it is easy to loose contact with
"the band" that does have valuable things to say at times.

> RISC5 is not powerful.

Ok next time I will call it Nanowulf. :)

> "bucket brigade"

and then? Process it outside realtime?

> but first I need to retire

Bad for your brain. After 2  years you look in the mirror and you will not
recognize yourself.
Thats my experience, and I did not even retire, just took a contract below
my capacity.

I will get the book from the university library when I need to.
I did fpga design in the grey past. I remember positioning all blocks by
hand to get the timing right.
The compiler was no use. (I tried to put too much into it for the compiler)
and I remember I multiplexed 3 quadrature input channels through 1 digital
noise filter to save flipflops.

Enjoy your sunday

j.



On Sat, Aug 29, 2015 at 4:23 PM, <skulski at pas.rochester.edu> wrote:

> J:
>
> > Seriously though, just to push the boundary a bit more:
> > you think you can set up 2 parallel oberon cores for us in an FPGA?
>
> Concerning two cores, why only two? RISC5 is an ideal platform for
> multicore designs similar to Parallella.org from or Propeller from
> Parallax.com. Check out the specs. IMHO this would be an ideal direction.
> However, going there (or anywhere else) would require forming a commercial
> company. A loosely coupled band of enthusiasts is unlikely to agree on
> this, or on anything else for that matter.
>
> > We can build a mini Beowulf, software wise it should be doable as long as
> > we dont get too fancy,
>
> Beowolf was designed for large scale high performance computations. RISC5
> is not powerful. It is small and flexible. This kind of soft core would be
> great for on-chip designs like I mentioned above.
>
> > Unless of course someone would be so kind to release the Xoberon system
> > sources, but I have little hope.
> > Just as way of processing 2 streams of data at once.
>
> Our next design is processing thirty two streams at a rate that are
> outside reach of any CPU, either hard or soft. Even Parallella would not
> be able to keep up. The approach in such a case is to set up the pipeline
> loosely named a "bucket brigade". One day I will write a textbook on such
> things, but first I need to retire.
>
> > Depending of course how much Bram is available.
> > ( you note, I did do some more reading about what you are up to)
>
> The only book that I am aware of is the one by Sadrozinski and Wu. There
> are tons of journal articles, but these are hard to comprehend for a non
> specialist. In a month or so our grad student will defend a thesis where
> he put together a complete system. It should be available on the web.
>
> W.
>
>
> --
> Oberon at lists.inf.ethz.ch mailing list for ETH Oberon and related systems
> https://lists.inf.ethz.ch/mailman/listinfo/oberon
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://lists.inf.ethz.ch/pipermail/oberon/attachments/20150829/5e5a942d/attachment.html