Re: [oc] wishlist/todolist

[John Dalton <john.dalton@bigfoot.com>]

> What is a wiki?

Refer to:

http://www.wikipedia.com/wiki/What_is_a_wiki

A wiki is sort of like a sheet of paper coupled
to CVS.  Anyone is free to scribble on it.  If
a malicious attack is made, the record of the past
allows the changes to be undone.  The lack of control
takes some getting used to, but it can achieve great
results.

The best known wiki project to date is the wikipedia
(http://www.wikipedia.com/).  It is an attempt to
collaboratively write an entire encyclopedia.  It's
success so far is quite impressive.

For instructions on building your own wiki, see:
http://www.wikipedia.com/wiki/How_can_I_build_a_Wiki_of_my_own

> Do you have concrete ideas what resources to develop to link ideas and
> projects?

It's still just a gem of an idea.  As details occur to me, I will post them
for discussion.

> What I have in mind is a backend environment with EDA tools where opencores
> web serves as a frontend. For start I think we could start with a web based
> HDL editor and a HDL simulator. Basically it would look something like this:
> 1) login to opencores
> 2) start a project
> 3) write some HDL
> 4) run simulation and log off
> 5) come back later (let say 2 hours later) and analyze simulation results
> 
> Of course for running simulations current dual PIII won't be enough. We'll
> probably need a small cluster farm. ;-)

Some ideas:
- Implement the simulator as an FPGA, with appropriate sampling logic
  for the nodes being observed.  Ideally, it would offer all the functions
  of a software only simulator.  A user accessing the simulator over
  the Internet, should only be aware of an increase in simulation speed.
  This offers the possiblity of near real time simulation, faster than any
  processor could do it.  Of course there are difficulties including
  compilation/syntheis time, damage due to conflicting outputs, the
  non-trivial task of automatically generating appropriate sampling logic to
  name a few.  I suspect this is basically what you are already talking about.
- If we can come up with a truly distibuted development environment, it
  might be possible to implement a distributed simulator, along the lines
  of SETI at home.  Each opencores user donates excess CPU cycles.
- Again, with a trully distributed environment, it should be possible for
  a developer to do most of their work locally.  Any computationally intensive
  tasks could draw on the distributed resources of opencores (transparently?).
  Maybe it will be possible for opencore's data store to be distributed among
  all users (need a way to avoid losing info if someone drops out).

> Definitely. So any concrete ideas?

None yet, though I think a wiki system would help in this regard.

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