On 29 June 2012 23:15, Tony Duell <ard at p850ug1.demon.co.uk> wrote:
Tony
*might* approve if the published documentation included detailed
instructions on how to mine your own copper ore, smelt it, build a
silicon refinery, fabricate your own CPU, spin glass fibre and
synthesize resin and then manufacture your own circuit board.
But I do emphasize the "might" here...
Actually, I'd be very likely to approve of it if :
It came complete, that is to say I could just plug it in and go without
having to pay to download and print manuals, download the OS, etc
Then it would cost significantly more than =C2=A330.
True. But a prodcut I can't use is surely a total waste of money.
IMHO documentation is not an option, it's essential. If you can't work
out how to use something, it's *(by defintion) useless.
The
docuemtnation let me figure out how to use the thing.
It's a device to teach kids programming. The bulk of the information
is on the web. You'd need a Web-capable computer and a broadband
connection.
That is one of the most worrying things I've read in a long time (not
your fault at all, I am not blaming you).
To use the Rpi you need various bits from a PC (USB keybaord, some kind
of PSU, normally a mouse, monitor, etc). And you need access to the web,
the conventioanl way to get that is to have a PC.
So, the Rpi is goign to be sold mostly to people who already have a
computer (at lleast in Eirope and the States). In whci hcase, why not
simply install prgrammign tools on said computer and program that. Heck,
you could dual-boot linus and Windows about 15 years ago (at least).
Given that, I might actualyl buy one.
It is really not a device for you, given your predilection for
vintage, technically-documented hardware. It's a closed-spec
Actually, I _did_ have an appication in mind where I might haev used one.
The reason I was asking about the GPIO port was that I wanted to see how
easy it would be to inteface it to an HPIB system. Given that the RPi
already has soem kind fo filesystem nd cna presuambly communicate with
USB stoeage deviecs and, of course, SD cards, I was wondering if it might
be the unit to use to make a HP disk drive emulator, soemthign to conenct
to my HP9000s, etc in place of the CS/80, SS/80 and Amigo drives.
undocumented system-on-a-chip with some supporting
hardware, which
Origianlyl it was climed ot be 'open'. I have now relaised that this
'open' is not what I would call it.
assumes that the owner has a basic set of 21st century
equipment to
hand: USB hub, USB keyboard, USB mouse, USB power supply, SD media and
a host PC that can read/write SD media, an HDMI-capable
high-definition flatscreen TV and so on. None of which you own and
from what you have told me before none of which you want to own.
For an embeeded use, presumably you don't; need all that once it's
working...
A couple of serious questions :
1) AS has been confiremd the data sheet on the main 'chip' doesn't
include docuemtnation on certain parts of it (video-related?). Are thos=
e
parts simply not used by the standard linux for
the Rpi, or are the
drives supplied as binary-only, or what?
Binary drivers for Linux only, no other OS. GPU is proprietary,
undocumented and protected Broadcom IP; they will not disclose
That's the reall downer for me.
details. Even the Linux driver only exposes certain,
limited APIs and
functions that the R=CF=80 Foundation have paid for, such as H.264 video
decode. There are lots of other functions in the GPU that Linux cannot
access and never will unless someone pays quite a lot of license fees
to Broadcom.
The device /has/ no discrete CPU; the processor is a small, low-spec,
old-fashioned ARM core of a type long superseded and no longer
supported by most current Linux distributions, which has been masked
onto some spare gates on one corner of the GPU die.
Right... It is presumably powerful enough for the application I've just
suggested (given that hte origianl HP drive units used a 68B09, albeit
with supporting hardware).
The machine boots because the GPU reads a FAT filesystem on the SD
card, looks for files of certain names, loads them into RAM and then
boots the ARM core and starts it executing.
Is that fully docuemtned? In otehr wordsm if you don't want video
functions at all, can you work with the bare metal?
It's not a CPU with an attached GPU as most modern computers; it's a
GPU with an afterthought of a CPU bolted onto the side.
THat makes little differnece practically, I do realise it's essnetially
one big 'chip' in the middle (I had heard it was actually deveral
silicon dice in the package, but that makes no practical difference either).
2) I understnad the's some kind of GPIO/user
port. How many lines, are
they individually selectable for direction? Can this be easilly used fr=
om
C (I assuem there's a C vompiler included
with the OS).
Don't know.
Wikipedia says:
Low-level peripherals: 8 =C3=97 GPIO, UART, I=C2=B2C bus, SPI bus with tw=
o chip
selects, +3.3 V, +5 V, ground
... If that means anything to you; it doesn't to me mostly.
YEs, it does.
The 'UART' is presumably an ansynchronous serial port, the sort of thing
ypou could feed into a level shifter IC and turn into an RS232 port.
I2C (Inter-IC Communication) is a Philips serial bux using 2 signals
(clock and data). It was origianly designed to send control infformation
to the ICs in a TV set or whehrever (so the microcontorlelr could send
the contrast value to the video decoder chip, send the teletext page
numbner ot that decoder, etc). It's varily low-speed, but fast ewnough
for that. More useuflly there are some general-purpose ICs iwht I2C
interfaces -- 8 bit paralle I/O, ADCs, DACs, etc
SPI is an older (and IMHO inferiour) serial conmmunicatio nstandard for
things like ADCs, DACs, etc. One differnece is that I2C includes
addressing (you need nothing other than the clock and data lines), while
SPI has clock, data in, data out and select lines, the last being a
separate signal for each IC (hende '2 chip selects' implies only 2 SPI
peripherals.
It's accessible from Linux and therefore from any
Linux programming
I wonder how much of it is easily accessible from Linux. Are there
supplied libraries to talk to I2C devies, etc?
Anyway, the problems for me now are hte lack of docuemntation, the
closed-source OS (not somethign I want on any machien I have to debug)
and the fact it appears you can't actually get one without a long wait.
-tony