[Open-graphics] Definition of Open Hardware

[Lourens Veen]

On Wednesday 04 October 2006 13:34, Hamish Marson wrote:
>
> Personally open-hardware to me means open specifications. And a
> license to use the hardware to it's best ability without prejudice.

So, that would more or less match the idea of open standards. Photoshop 
reads and writes standard PNGs, and you can draw anything you like with 
it (although there are supposed to be some banknote-editing protections 
in there). Photoshop itself is proprietary, but at least it talks to 
everyone else.

> As a couple of examples?
>
> At the other end of the spectrum is chips like the ATI x1400 (R5xx
> family IIRC). You can't do squat without a non-disclosure from ATI.
> No register level docs etc.

So this is more like Microsoft's undocumented interfaces and proprietary 
file formats that they use to gain competitive advantage over rival 
companies.

> Sparc (And powerPC now?) can also be considered open... Although I'm
> not sure what you get with open powerpc is that CPU or whole system
> with bits inside still hidden?

I'm not sure about PowerPC, but at least OpenSPARC gives you all the RTL 
under the GNU GPL. So that would be comparable to completely free/open 
source software.

> Maybe we do need several levels of openness... With the lowest being
> complete register level...

I'm still not sure whether I would call anything but this last example 
Open Hardware. If the interface is fully documented then it could be 
called Standards Compliant Hardware, or perhaps Generally Useable 
Hardware. I agree that this is a useful distinction to make. And I like 
the term "generally useable".

It probably also depends on your point of view. A software developer 
probably only cares about having the interface data available, so that 
a free driver can be written. A hardware developer isn't going to have 
much to work with if she only has the interface description. To a 
hardware hacker hardware with closed HDL but a documented interface 
isn't very open at all.

> But what about something like ipw2100? I know the hardware itself
> isn't open, but there is a relatively open interface to the
> proprietary firmware... Maybe there needs to be a level for hardware
> that although runs closed proprietary firmware, there's still a
> usable & generic documented interface to it...

Well, essentially there is not much difference between a hardware 
description that says "to do X, write number Y to register Z" and one 
that says "to do X, write numbers (a, b, c, ...) to register A, then 
write number Y to register Z". In that sense, the firmware is part of 
the hardware. If the HDL were available but the source to the firmware 
wasn't, then I still wouldn't call it open hardware however. But if it 
were properly documented then it could still be standards 
compliant/generally useable.

Incidentally, I believe RMS would disagree with this. IIRC he runs 
LinuxBIOS, claiming that the firmware is user-modifiable and therefore 
should be free. I think this is a rather subtle and tricky point. The 
current ATI/nVidia cards with proprietary Linux drivers also present a 
usable & generic documented interface: OpenGL. If you accept the driver 
as part of the hardware that is.

> Hmm... I'm thinking maybe good docs for the interface to the hardware
> are probably more important than the HDL itself... What if someone
> like ATI released an obsfuscated HDL description of what they claimed
> was an r5xx chip just to claim openness(If any open hardwaer spec was
> based on HDL alone)... You still wouldn't necessarily have any
> register level description... And their own hardware would still be
> slightly incompatible with it... But how long & how much effort to
> prove it...

Well, the GPL says that source code should be in human-readable form, 
the same could go for the HDL. As an analogy, the GIMP file format XCF 
was undocumented for a long time. Recently, some noise was made about 
this, and someone took it upon themselves to read the source and create 
a description of the file format. So, there is now an XCF 
specification. It's possible.

Lourens
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 191 bytes
Desc: not available
Url : http://kayan.duskglow.com/pipermail/open-graphics/attachments/20061004/d583d003/attachment.bin