On Sunday 25 June 2006 10:13, John Culp wrote:
> I've finally completed a quick-n-dirty EPP driven framebuffer.  The full
> design, Linux driver code, design documents, and some misc pictures are
> posted at www.johnculp.net.  This is a new hosting account, so DNS may
> or may not resolve.  Also try  http://0087a8d.netsolhost.com/.   I am
> placing everything in the public domain.
>

Hey cool. Hardware & all. very nice.

> This design takes up about 42% of a Xilinx spartan XC2S200E FPGA and is
> currently driven by a 50Mhz clock.  I don't think reaching 100Mhz is out
> of the question, but this was not a design goal.
>
> If I had the time to work more on this in the short term I would:
>
> Fix the display bugs. Row 0 of the framebuffer is displayed near the end
> of the screen?! The display then wraps over and looks almost normal.
> There is also some misc trash on the screen that I have not tracked down.
>
> Fix whatever bug exists in writing single pixels, the problem is seen in
> the line drawing picture on the diagonal lines.
>
> Improve the EPP performance.  This would be a half hardware/half
> software solution.
>
> Implement Bresenham's Algorithm in hardware.
>

I've been posting one to the list with improvements as I learn more about 
verilog. The thing it doesn't do is actually render pixels. I'll pull your 
code & see about integrating my bresline module into your design & see how 
far I get.

> Modify the MCU to provide pixel read/modify/write operations for Memory
> Users and to add unaligned accesses. (probably the toughest improvement)
>

Why bother with unaligned accesses? (I take it you mean the host CPU reading a 
byte/word at an odd address.

> Add block fill.
>
> Add a hardware cursor.
>
> Add a bit-blit.
>

OK... I'll pick up some of that, just to learn more about verilog (Even if 
noone else likes my code :)

> None of the above are very difficult modifications, I think another
> month of work would do it, but I do not have the time right now.  After
> the above changes, I think re-targeting it to a PCI/PCI-Express board
> would not be difficult and would result in a  pretty competent 2D solution.
>
> I think a full 3D (depth buffering and texture mapping) core could fit
> ?easily? in a webpack supported FPGA.  Performance (of everything)
> pretty much depends on available memory bandwidth.  This is probably
> another 3-4 months of work.
>
> In an ideal world other folks would pick this code up and retarget it to
> other dev kits (preferably with sdram on board) with other types of
> computer connections.  As I've said before I am a ME, not an EE, so to
> do this I've had to (partly) learn VHDL over the past few weeks.  This
> is by far the most complex digital design I've done, and beyond flashing
> a LED a few years ago, this is the first successful FPGA project I've
> implemented.
>

Hmm... I'd love to have a devkit. Not sure I could afford one though, I'm 
saving (OK, my company is saving :) for an OGD1 board when they roll out the 
door.

Hamish.

Attachment: pgp8ylrYBa6PQ.pgp
Description: PGP signature

_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)

Reply via email to