Date : Wed, 28 Dec 2011 06:32:47 +0100
From : rick@... (Rick Murray)
Subject: About ARM PC
On 28/12/2011 00:35, Alan Williams wrote:
> and whatever was necessary to get BeebEm running.
Just, you know, out of interest... why not an actual Beeb? Is it for
networking support?
> "Documentation/Datasheets: There is not going to be any low level
> documentation for the BCM2835
Useful.
> other than info regarding the GPIO stuff.
At least there's that, then.
> This is one topic that typically becomes less than cival
Doesn't this suggest something?
> The short answer I have for you is that Broadcom does not
> operate at this level of transparency
WTF? I've read the DM320 datasheets cover to cover. The DSP is only
sort-of mentioned (you need a specific datasheet for that, like it
doesn't cover the ARM instruction set). It's a multi-level framebuffer
(two video feed frames (two?), a video display frame, a mask frame, and
IIRC a cursor frame). It can run varying levels of transparency and
paste that over top of a live video feed in the video feed frame.
There's auto white balance and such. A co-pro DSP. A co-pro hardware
resizer. A co-pro this and that. Basically the DM320 is a device that
works with a CCD imager (either camera or video, in the case of my PVR,
the tvp5150 chip converts analogue video into a bitstream, the DM320
thinks it is talking to a video CCD) and optionally a McBSP audio codec
chip to be a still or video camera/recorder. There's a lot of techie
junk in here, but I cannot imagine *anything* that would warrant NDA?
[I think it is a sort of pre-DaVinci:
http://en.wikipedia.org/wiki/Texas_Instruments_DaVinci ]
After all, god knows there are enough fab houses capable of bolting
video hardware to an ARM to make a similar chip. The Beagle uses a TI
OMAP, the RPi will use a Broadcom. Essentially they're the same thing,
just a different implementation.
> Since RISC OS is struggling with USB sound
? Is this chip unable to generate its own sound?
> it will be interesting to see if HDMI sound is any better.
I think it depends how the sound is generated. I can imagine that
anything that isn't a cyclic buffer arrangement (a la VIDC(2)) could
present... difficulties.
> I would hope RISC OS has enough 'friends' in Broadcom that ROOL
It seems like this is a likely or else we wouldn't have RISC OS at all
(you need hardcore info to be able to start up and initialise the chip),
but how many people *actually* have access to the data? If only a few,
then it will take longer to get the job done.
If they have gone the NDA route, will this affect the the RPi specific
parts of the code? Will it be in the source bundle, or available as a
binary blob (or, worse, prebuilt firmware image)?
> I would hope that given the RP is entirely aimed at being an
> educational product that Broadcom might make more of the resources
> available.
It just seems, to me, like they're shooting themselves in the foot. I
mean, come on, the core chip is old enough that Ubuntu is no longer
supporting it. Surely any PROPRIETARY TRADE SECRETS (or whatever the NDA
is supposed to cover) is so far behind the state of the art that it
isn't such a big deal to reconsider. But, if nothing has happened thus
far, I wonder if it ever will.
> In a thinly valid attempt to get back on topic how would owning a
> BEEB have been without 6502 and 6522 datasheets, without The advanced
> user guide or Rodnay Zaks 'Programming the 6502'.
Damn right.
Or look to The Dark Side - do you remember that MASSIVE tome that took
apart the PC BIOS (I think, AMI's one?) and described exactly what it
did, how, then went on to describe the x86, ISA bus, interrupt
controllers, why the COM1/COM3 and COM2/COM4 nonsense, etc etc? It was
bigger than a family bible. It had everything except a schematic. So you
could get in and *understand* the machine.
Now it's a piddly half-inch chip with an equally piddly lump of memory
stuck on top. It can blow away any of the old hardware without breaking
a sweat, but god help you if you ever need to try to understand what is
actually happening.
It's no wonder humanity is turning into imbecile user-consumers that
can't see beyond the next Twitter update. But, given that corporations
run the show and corporations have a lot of tat to flog to the mindless
sheep, they're quite happy with this arrangement.
> Rick's poor experience with TI doesn't seem to hold for the
> Beagleboard. There are a lot of data books downloadable for the OMAP
> cpu used on it.
My poor experience dates from just under two years ago. Post-Beagle
days. [ http://www.heyrick.co.uk/blog/index.php?diary 100123 ]
I believe that TI is a little more forthcoming with docs now, but I
wonder if the information would have been around if not for the
popularity of the Beagle? Maybe TI have seen the light and realised that
having geeks onboard is not an entirely bad thing, even if you might end
up with some unheard of operating systems like Symbian, or RISC OS, or
Haiku... :-P
Best wishes,
Rick.
--
Rick Murray, eeePC901 & ADSL WiFI'd into it, all ETLAs!
BBC B: DNFS, 2 x 5.25" floppies, EPROM prog, Acorn TTX
E01S FileStore, A3000/A5000/RiscPC/various PCs/blahblah...