[Open-graphics] Re: Setting video mode

[James Richard Tyrer]

Timothy Miller wrote:
> On 8/1/06, mykrowatt at comcast.net <mykrowatt at comcast.net> wrote:
> 
>>    The way si87 did it was to provide a DIP switch for boot mode 
>> selection.  4 bits on the DIP switch select stored modes for various 
>> common fixed frequency monitors, with mode 0 being VGA and I think 
>> mode 4 being the A1097C.  These modes are loaded in flash at the 
>> factory, and custom modes can be loaded on special order.  Other 
>> switch bits set sync options, such as sync-on-green, composite sync, 
>> etc.  The DIP switch is accessible on the back panel, without opening 
>> the case.
> 
> We can provide jumpers or a DIP switch (jumpers may be cheaper, but I
> don't know).  The problem here is identifying a good list of fixed
> modes that will satisfy everyone.
> 

This isn't going to work -- we can't satisfy everyone.  By definition 
odd ball monitors are not going to be standard.  That is why X11 has 
modelines.

We could have it possible for it to start in VESA modes that are larger 
than 640x480, but that isn't going to help with something odd ball.

We could have it possible to select some (or all :-() of the standard 
modelines built into the current XFree86 and X.org releases, but again, 
there are going to be real odd ball monitors that this isn't going to help.

I would suggest that a serial NVRAM with the X11 modeline or VESA mode 
for boot up in it is the best idea.  It might also be a good idea for 
this to store the VESA modes that were supported.  Since there is a 
finite list, we would only need a bit for each one (best to have a few 
to spare since I presume that they may add new ones).  And, it appears 
the type of sync the monitor needs (H&V, Composite, on Green).

You could load this NVRAM yourself if you borrowed a VGA monitor or 
could use your TV.  Otherwise, we could offer the programing service.

This leaves us with two jumpers (4 choices):

	VGA

	TV 480i

	TV 489p

	NVRAM stored mode

and a jumper to enable writing the NVRAM.

Perhaps we wouldn't need to populate the serial NVRAM on the standard 
product.

-- 
JRT