Timothy Miller wrote:
On 6/25/06, James Richard Tyrer <[EMAIL PROTECTED]> wrote:
In message
<[EMAIL PROTECTED]>, "Timothy
Miller" writes:
> I'm not even sure this could fit in the XP6 we've chosen, although
> there's probably a bigger (but more expensive) one we could use.
> Usually, more integration is less expensive, but with FPGAs, it's the
> other way around. Two small ones are cheaper than one large one. We
> could split the design across two FPGAs for this.
If you are going to use two chips, would it be possible to use an
existing chip for part of the design? A simple NorthBridge for memory
control, for example. Or does someone still make memory controller
chips? Probably not. :-(
We looked into using a northbridge chip as a graphics chip. It seems
that we can't do it because northbridge chips are not designed to talk
to other northbridge chips. More precisely, they're designed not just
to be bus masters but a CENTRAL RESOURCE, and you can have only one of
those.
Having looked at the DOX for the Intel 945G series, I would say yes and
no. This chip has 4 connections:
1. Front Side CPU bus
2. Video outputs
3. Memory
4. I/O buss (PCIe like)
The limitation that I see is that you have to connect something to #1
that will initialize and control the chip. Naturally, this (CPU) bus is
designed for an Intel CPU but that doesn't mean that you couldn't
connect something else -- even a PCI, AGP, or PCIe interface -- using a
CPLD for glue. IAC, you have to load a bunch of data into registers to
initialize it. The simplest way to do this as far as the host PC is
concerned is to have a CPU on the card to read the data out of ROM and
load it. If that is all the CPU does, it wouldn't have to be much of a CPU.
There are issues of memory mapping. When the host PC accesses memory
through the NorthBridge on the card, it has to be mapped into an
appropriate (and empty) address space. But this is always an issue with
any graphics card.
Memory mapping would also be an issue if the host PC was going to IPL
the NorthBridge but this is a common issue when programing any
peripheral chip. If the standard memory space for the graphics card is
not sufficient, windowing could be used. IAC, this is probably a bad
idea since it would require drivers for other OSes. It would be best
for the thing to boot up by itself and act like a VGA card till told to
do something else.
The I/O bus could also be used to access the NorthBridge through a PCI,
AGP, or PCIe interface.
Either way (I/O access) there would be issues with the PCI addresses for
the video hardware since the host PC would need to access it when
running in the VGA mode. This appears to be permanently assigned to
D2:F0 and D2:F1. If this would conflict with the host PC's PCI bus
assignments, it would be necessary to fix this. If this can not be
addressed by memory mapping, you could use, or emulate, a PCI bridge.
Did you find additional issues that I have not considered?
--
JRT
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)
