new Risc PC motherboard and Risc OS 4.0
Dmitry Petrov
Hello,
Below are some details:
The new motherboard will have
- 4 SIMM sockets supporting FPM and EDO DRAM
- 4 sockets for VRAM (up to 128 bit data path!)
- flash upgradable 256KB BIOS (self-diagnostics, low level routines,
hardware tables) with hardware protection against overwriting
- a special slot for Risc OS ROMs (instead of separate sockets there
will be an easily replacable card)
- optional on-board 15ns L2 cache
- 2 PCI slots
- CHRP compatible design (?)
- socket for optional Ethernet and/or SCSI-2 chip
- a separate socket for optional FPA accelerator
There is a plans to release Risc OS 3.8 before releasing Risc OS 4.00
which is a major rewrite. All hardware specific routines will be moved
to the first 256KB of Risc OS ROMs. The rest of OS will not contain any
direct hardware access.
During startup the new motherboard will map its own 256KB BIOS over Risc
OS. As you can see this approach maintains compatibility with existing
Risc PC/A7000 motherboards while allows to have new hardware without
rewriting oprating system. Right now Risc OS 3.7 has support for several
different types of motherboards with slightly different controllers.
Sorry for bad English.
Dmitry Petrov
D P Macdonald
Dmitry Petrov (dpe...@acorn.new-cent.transit.ru) wrote:
>Loads of stuff about a new Risc PC.
This is all pure speculation, right?
--
David MacDonald Trevelyan College, Durham, England
European Law Students Association, Durham
Home Page (unfinished): http://www.dur.ac.uk/~d402qa/
"There's a tale in that", Tammyn the bard.
Patrick Herborn
In message <326839...@acorn.new-cent.transit.ru> Dmitry Petrov wrote:
> Hello,
>
> Below are some details:
>
> The new motherboard will have
> - 4 SIMM sockets supporting FPM and EDO DRAM
That's an improvement. How is that achieved? I thought IOMD only had support
for two SIMMs....
> - 4 sockets for VRAM (up to 128 bit data path!)
128 bit data path. Cool. Does that go through a MUX into a VIDC20, or does it
go to a new VIDC / multiple VIDCs ?
> - flash upgradable 256KB BIOS (self-diagnostics, low level routines,
> hardware tables) with hardware protection against overwriting
That's a novelty....
> - a special slot for Risc OS ROMs (instead of separate sockets there
> will be an easily replacable card)
Like a PCMCIA card type thing?
> - optional on-board 15ns L2 cache
How much of it? This is 66 Mhz stuff... cool or what?
> - 2 PCI slots
What sort of PCI controller is being used? Is there a PCI <> DEBI bridge?
> - CHRP compatible design (?)
Fair enough.
> - socket for optional Ethernet and/or SCSI-2 chip
On the motherboard? Now there's an idea!
> - a separate socket for optional FPA accelerator
But what will we fill it with? FPA11 is broke, and I haven't heard that Arm
Ltd are working on an new FPA (well, they may be working on an FPA core, but
a separate device? that would be interesting, if a bit pointless since SA-110
doesn't have an external copro bus)
> There is a plans to release Risc OS 3.8 before releasing Risc OS 4.00
> which is a major rewrite.
So what's new in RO 3.8? And more importantly, what's new in RO 4.0?
> All hardware specific routines will be moved to the first 256KB of Risc OS
> ROMs. The rest of OS will not contain any direct hardware access.
Ah, I see. Is this to allow the flash-upgrade ?
> During startup the new motherboard will map its own 256KB BIOS over Risc
> OS.
Fair enough.
> As you can see this approach maintains compatibility with existing
> Risc PC/A7000 motherboards while allows to have new hardware without
> rewriting oprating system.
Where's the DEBI bus?
> Right now Risc OS 3.7 has support for several different types of motherboards
> with slightly different controllers.
Really? Cool!
> Sorry for bad English.
I thought it was quite good, actually.
> Dmitry Petrov
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. Starkle Starkle Little Twink What the hell you are I think?
Mark Gillman
In article <326839...@acorn.new-cent.transit.ru>, Dmitry Petrov
<dpe...@acorn.new-cent.transit.ru> wrote:
> Below are some details:
>
> The new motherboard will have
[etc]
Says who?
--
MG. Statements made from my personal account should not be attributed to the
company I work for: I pay the bills around here.
Stefan Brueck
> Below are some details:
Where on hell have you ripped them off ?
> The new motherboard will have
> - 4 sockets for VRAM (up to 128 bit data path!)
:-( Thus implies they still haven' bothered to let us choose 'real' graphics,
but again an 'on-the-motherboard-restriction' against modularity. Sad to hear
that. Endless VRAM doesn't help around the fact that others got
3D-acceleration by now, surely not being removed in the future.
> - flash upgradable 256KB BIOS (self-diagnostics, low level routines,
> hardware tables) with hardware protection against overwriting - a special
> slot for Risc OS ROMs (instead of separate sockets there will be an easily
> replacable card)
What do I need the second one for ? I could boot straigth from the harddisc,
whatever OS I prefer, right ?
> - 2 PCI slots
Less than expected.
> - socket for optional Ethernet and/or SCSI-2 chip
I'd call that a half-modular-design, if ever.
> - a separate socket for optional FPA accelerator
Still not for standard ?
> Sorry for bad English.
Can't do any better myself. :-)
> Dmitry Petrov
The guy who brought us !Filler, right ?
cheerio
Stefan
--
Stefan Brueck sbr...@mail.bs.provi.de Voice: +49-5362-66006
Kampstr. 3 Fax : +49-5362-66007
38442 Wolfsburg Data : +49-5362-66008
Stephen B Streater
In article <na.db8252471f....@argonet.co.uk>, Mark Gillman
<URL:mailto:mgil...@argonet.co.uk> wrote:
>
> In article <326839...@acorn.new-cent.transit.ru>, Dmitry Petrov
>
>
>
> Says who?
It's putting ART's new policy of customer responsiveness into practice.
--
Stephen B Streater
tord.s.eriksson
>
> Hello,
>
> Below are some details:
>
> The new motherboard will have
> hardware tables) with hardware protection against overwriting
> - a special slot for Risc OS ROMs (instead of separate sockets there
> will be an easily replacable card)
> - CHRP compatible design (?)
>
> to the first 256KB of Risc OS ROMs. The rest of OS will not contain any
> direct hardware access.
>
> Risc PC/A7000 motherboards while allows to have new hardware without
> different types of motherboards with slightly different controllers.
>
>
> Dmitry Petrov
As good as mine, and that ain't too bad :-)
--
Tord S Eriksson,
another friggin' foreigner
Mark Gillman
In article <ant192226bc8%2...@surprise.demon.co.uk>, Stephen B Streater
<ste...@surprise.demon.co.uk> wrote:
> > > Below are some details:
> > >
> > > The new motherboard will have
> >
> > [etc]
> > Says who?
> It's putting ART's new policy of customer responsiveness into practice.
So it's for real?
And if it is, does this chap have permission to broadcast the news?
I'm suprised to hear it this way, that's all.
S. W. Tengelsen
>
> In article <na.db8252471f....@argonet.co.uk>, Mark Gillman
> <URL:mailto:mgil...@argonet.co.uk> wrote:
> >
> >
> > > Below are some details:
> > >
> > > The new motherboard will have
> >
> > [etc]
> >
> >
> > Says who?
>
> It's putting ART's new policy of customer responsiveness into practice.
Huh? Sounds like an informed guesstimate to me. :) I even heard there
will be sneak preview of the new MB at AW96?
--
Best regards,
Sveinung W. Tengelsen
a.k.a. pixeleyes
_______________________________________________________________________
| Dolphin Design is ARM-powered | +47 22 67 22 21 - pixe...@sn.no |
| Scariest website on the web: http://www.igc.apc.org/millennium/inds/ |
Daniel Maloney
Pat Herborn wrote:
> In message <326839...@acorn.new-cent.transit.ru> Dmitry Petrov wrote:
^^^^^
So, is this official, or just to promote rumours?
> > The new motherboard will have
> > - 4 SIMM sockets supporting FPM and EDO DRAM
> That's an improvement. How is that achieved? I thought IOMD only had
> support for two SIMMs....
...and it doesn't support EDO RAM, so they can't have used (the current)
IOMD.
> > - 4 sockets for VRAM (up to 128 bit data path!)
> 128 bit data path. Cool. Does that go through a MUX into a VIDC20, or does
> it go to a new VIDC / multiple VIDCs ?
More to the point, will our old VRAM work with it? How much VRAM is that,
hopefully 8Mb?
> > - flash upgradable 256KB BIOS (self-diagnostics, low level routines,
> > hardware tables) with hardware protection against overwriting
> That's a novelty....
Hmmmm.
> > - a special slot for Risc OS ROMs (instead of separate sockets there
> > will be an easily replacable card)
> Like a PCMCIA card type thing?
Probably more likely an A5000 memory board style thing?
> > - optional on-board 15ns L2 cache
> How much of it? This is 66 Mhz stuff... cool or what?
How much will it cost? Will smaller ammounts be available for poor
people?
> > - 2 PCI slots
> What sort of PCI controller is being used? Is there a PCI <> DEBI bridge?
Will there *be* any DEBI slots? How will this work in single slice
machines?
> > - CHRP compatible design (?)
> Fair enough.
It that '(?)' possibly, or '(?)' probably?
> > - socket for optional Ethernet and/or SCSI-2 chip
> On the motherboard? Now there's an idea!
I'll have to get one and flog off my Cumana card then!
> > - a separate socket for optional FPA accelerator
> But what will we fill it with? FPA11 is broke, and I haven't heard that
> Arm Ltd are working on an new FPA (well, they may be working on an FPA
> core, but a separate device? that would be interesting, if a bit pointless
> since SA-110 doesn't have an external copro bus)
Yeah, this is wierd! Maybe a sort-of FPA/FPE traps FP instructions, like
the FPE and pushes them over, like people keep suggesting with the 5x86
card? It certainly seems to suggest that StrongARM won't be getting an
on-board FPU in the near future, or this *would* be rather pointless.
> > There is a plans to release Risc OS 3.8 before releasing Risc OS 4.00
> > which is a major rewrite.
> So what's new in RO 3.8? And more importantly, what's new in RO 4.0?
> > All hardware specific routines will be moved to the first 256KB of Risc
> > OS ROMs. The rest of OS will not contain any direct hardware access.
> Ah, I see. Is this to allow the flash-upgrade ?
> > During startup the new motherboard will map its own 256KB BIOS over Risc
> > OS.
> Fair enough.
But why is this BIOS flash upgradable? Surely if they do it right, it
will never need to be upgraded, just the ROMS, so why aren't they flash
upgradable? Very wierd.
> > As you can see this approach maintains compatibility with existing
> > Risc PC/A7000 motherboards while allows to have new hardware without
> > rewriting oprating system.
> Where's the DEBI bus?
Good question. Also, what happens about the open bus? Will there be more
processor slots? Will there be support for 64-bit processors? Will the PC
card and Hydra work in it? Will there be a need for the Hydra? This post
just didn't answer enough of my questions. Sorry.
> > Right now Risc OS 3.7 has support for several different types of
> > motherboards with slightly different controllers.
> Really? Cool!
Hmmmm?
Cheers,
Dan.
--
____
| \ __ _ _ __ # Where's my StrongARM? In my RiscPC of course!! :-)
| D / _` | ' \ # All this speed and no use for it!
|____/\__,_|_||_| # Mail to: dmal...@argonet.co.uk
Dmitry Petrov
> Hi!
>
> > Below are some details:
> >
> > The new motherboard will have
> > - 4 SIMM sockets supporting FPM and EDO DRAM
> > routines,
> > hardware tables) with hardware protection against overwriting
> > there
> > will be an easily replacable card)
> > - CHRP compatible design (?)
>
Hi!
Well, it is just a logical conclusion based on various (partially
confirmed) rumours and confirmed fact that ART is going to release
new improved motherboard. The above specification is just a
necessary *minimum*. Otherwise it will not be acceptable. Sorry if I
dissappointed you.
I thought the my original message will force people to think about
the future and it will create some useful thread helping ART to go in
a proper direction.
Best regards,
Dmitry Petrov
Patrick Herborn
In message <na.dfd814471f....@argonet.co.uk> Daniel Maloney wrote:
> Pat Herborn wrote:
>
> > In message <326839...@acorn.new-cent.transit.ru> Dmitry Petrov wrote:
> ^^^^^
> So, is this official, or just to promote rumours?
I didn't know that Acorn had a subsisary in Russia....
> > That's an improvement. How is that achieved? I thought IOMD only had
> > support for two SIMMs....
>
> ...and it doesn't support EDO RAM, so they can't have used (the current)
> IOMD.
Good point :-) So it must be a different IOMD.
> > 128 bit data path. Cool. Does that go through a MUX into a VIDC20, or does
> > it go to a new VIDC / multiple VIDCs ?
>
> More to the point, will our old VRAM work with it?
One would hope so....
> How much VRAM is that, hopefully 8Mb?
Hmm.. imagine 1600x1200 in true colour.... :-) Should just about fit...
> > > - flash upgradable 256KB BIOS (self-diagnostics, low level routines,
> > > hardware tables) with hardware protection against overwriting
>
> > That's a novelty....
>
> Hmmmm.
I think it's a good idea, actually. This reduces ART's overheads (they don't
need to produce different ROMs for each machine, and they don't need to put
support in RISC OS for <every device imaginable>; they just put that in the
BIOS) hence (hopefully) reducing cost.
> > Like a PCMCIA card type thing?
>
> Probably more likely an A5000 memory board style thing?
Hmmm :-( How about a JEIDA (sp?) card? You can get 8MB DRAm ones, so it must
be possible to make 4MB ROM ones... Just insert your OS like a disc....
> > How much of it? This is 66 Mhz stuff... cool or what?
>
> How much will it cost?
Probably no more than for PCs (ie UKP <not very much>). Current(ish) price
for a 512kB pipeline cache module is UKP 17+VAT so it's not a significant
cost, compared to the total cost of the machine.
> Will smaller ammounts be available for poor people?
Are you worried that they were going to use 8MB of the stuff (like DEC) ? Or
are you worried about whether the cache controller will support different
amounts?
> > What sort of PCI controller is being used? Is there a PCI <> DEBI bridge?
>
> Will there *be* any DEBI slots?
One would hope so....
> How will this work in single slice machines?
Presumably by using a backplane... but that rather defeats the object :-(
> > > - CHRP compatible design (?)
>
> > Fair enough.
>
> It that '(?)' possibly, or '(?)' probably?
Wouldn't you like to know... :-)
> > > - socket for optional Ethernet and/or SCSI-2 chip
>
> > On the motherboard? Now there's an idea!
>
> I'll have to get one and flog off my Cumana card then!
Well, why not? If it's Fast-Wide SCSI ][ then you'de expect a (very) large
performance boost over your Cumana. It is choked by the DEBI bus :-(
> > > - a separate socket for optional FPA accelerator
>
> > But what will we fill it with? FPA11 is broke, and I haven't heard that
> > Arm Ltd are working on an new FPA (well, they may be working on an FPA
> > core, but a separate device? that would be interesting, if a bit pointless
> > since SA-110 doesn't have an external copro bus)
>
> Yeah, this is wierd! Maybe a sort-of FPA/FPE traps FP instructions, like
> the FPE and pushes them over, like people keep suggesting with the 5x86
> card? It certainly seems to suggest that StrongARM won't be getting an
> on-board FPU in the near future, or this *would* be rather pointless.
Hmmm... or /maybe/ DEC have decided to produce an SA-100 CPU. But if that's
the case, how the hell do you get the signals for the FPA off the SA
processor card? There aren't enough pins left :-( Or will you just use a MUX
on the card, and a memory / not_fpa line ?
> But why is this BIOS flash upgradable?
Very simple. The cost of making ROMs is rather high, unless you make LOTS of
them. Acorn doesn't produce LOTS of machine (unfortunately) so they can put a
flash device on the motherboard and blow it with board specific routines at a
much reduced cost.
> Surely if they do it right, it will never need to be upgraded,
That's true. I would think that they've used this route so they can reduce
overheads. They could use socketed EPROMs, but then you'de have to fit them
to the board. This way you can just mount the devices on the production line,
and blow the board specific routines in situ....
> just the ROMS, so why aren't they flash upgradable?
Flash memory may be cheaper than EEPROM, but you don't get 10-a-penny. It's
better to stick to the smallest amount that will allow you to get machine specific routines into it. That way you can produce the RISC OS ROMs in huge
numbers (they will work on all hardware, since they contain NO hardware
specific code). In those sort of number it becomes more economical....
> Very wierd.
Not really...
> > Where's the DEBI bus?
>
> Good question.
Well, I for one won't be crying if we lose the DEBI bus....
> Also, what happens about the open bus?
Presumably it stays on non-CHRP machines.
> Will there be more processor slots?
Now /there's/ an idea! :-)
> Will there be support for 64-bit processors?
Problem with that is getting data to/from the CPU card. You'de need bigger
slots. That shouldn't be a problem, though. Ie, the default bus master is the
wide card, IF it's there, else it's narrow card....
> Will the PC card and Hydra work in it?
One would hope so.
> Will there be a need for the Hydra?
Just how far do you go with CPU sockets on the main board? ART's answer to
that question will decide the answer to yours, I would presume.
> This post just didn't answer enough of my questions. Sorry.
If it /was/ for real, it gives us something to talk about... but WHY aren't
ART talking to the Clan about this? They want to produce a high performance
board, and who's likely to buy it? The Clan members perhaps? So why not
encourage a dialogue about the board's design?
> > > Right now Risc OS 3.7 has support for several different types of
> > > motherboards with slightly different controllers.
>
> > Really? Cool!
>
> Hmmmm?
I was referring to the very very very very slim chance it may even support
the older hardware.... that would make a few people out there happy!
> Cheers,
>
> Dan.
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. A phaser on stun is like a day without orange juice.
Patrick Herborn
In message <3269EA...@acorn.new-cent.transit.ru> Dmitry Petrov wrote:
> Well, it is just a logical conclusion based on various (partially
> confirmed) rumours and confirmed fact that ART is going to release
> new improved motherboard.
So your nodename has nothing to do with Acorn....
> The above specification is just a necessary *minimum*.
PCI isn't necessary. It would be nice though. Don't forget that ART are
working on (allegedly) two boards. One that's CHRP ompliant, the other on
isn't.
> Otherwise it will not be acceptable. Sorry if I dissappointed you.
Well...
> I thought the my original message will force people to think about
> the future and it will create some useful thread helping ART to go in
> a proper direction.
Nice thought. :-)
> Best regards,
> Dmitry Petrov
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. So many lawyers, so few bullets.
Thomas Boroske
In message <na.dfd814471f....@argonet.co.uk> Daniel Maloney wrote:
> Pat Herborn wrote:
>
> > In message <326839...@acorn.new-cent.transit.ru> Dmitry Petrov wrote:
> > > - optional on-board 15ns L2 cache
>
> > How much of it? This is 66 Mhz stuff... cool or what?
>
> How much will it cost? Will smaller ammounts be available for poor
> people?
If they use standard "PC" 2nd level cache than there won't be much of
a problem with the price. The VRAM prices are much higher.
I still don't know what to make of the original message, Dmitry still
didn't say where he got all this from.
--
Thomas Boroske
.. --T-A+G-L-I+N-E--+M-E-A+S-U-R+I-N-G+--G-A+U-G-E--
Stuart Halliday
Dmitry Petrov wrote the following...
> > This seems to be very interesting. Where did you find this info?
> Hi!
>
> Well, it is just a logical conclusion based on various (partially
> confirmed) rumours and confirmed fact that ART is going to release
> new improved motherboard. The above specification is just a
> necessary *minimum*. Otherwise it will not be acceptable. Sorry if I
> dissappointed you.
>
> I thought the my original message will force people to think about
> the future and it will create some useful thread helping ART to go in
> a proper direction.
So the bottom line is that you made it all up and had the cheek to post it
as fact?
From rumours I've heard from various people ART are going to be producing a
3rd gen RiscPC motherboard, at present their Labs have various different
models being tried out none of which could make it out of the door. So we
should not take to much notice of any 'facts'.
But for my two pence worth the main priority is the data bus speed to cope
with the 200MHz+ StrongARMs that will be coming out next year and an extra
SIMM slot would be nice. Everything else is just a luxury IMHO.
P.S.
IMHO I think we'll see a RISCOS CHRP PCI card before a Acorn CHRP
motherboard in about 12 months time.
--
Stuart Halliday - Web Manager of the
___ ___ _ __ ___ _ _
/ _ \ __ ___ _ _ _ _ / __| _| |__ ___ _ \ \ / (_) | |__ _ __ _ ___
| _ / _/ _ \ '_| ' \ | (_| || | '_ \/ -_) '_\ V /| | | / _` / _` / -_)
|_| |_\__\___/_| |_||_| \___\_, |_.__/\___|_| \_/ |_|_|_\__,_\__, \___|
|__/ |___/
ZFC HM
http://www.cybervillage.co.uk/acorn/
Mike Wilson
In article <ant19170...@s-74588.swipnet.se>, tord.s.eriksson
<mailto:tord.s....@swipnet.se> wrote:
>
> In article <326839...@acorn.new-cent.transit.ru>, Dmitry Petrov
> >
> > The new motherboard will have
[BIG SNIP]
> > Dmitry Petrov
How the hell does he know?
--
Mike Wilson, Tel: 0113-253-3722 email: mi...@barc.demon.co.uk
Proud to be a member of the Wakefield Acorn Computer Group
But views expressed are mine alone
Robert Templeman
Dmitry Petrov <dpe...@acorn.new-cent.transit.ru> wrote:
>Hello,
>
>Below are some details:
>
>The new motherboard will have
> - 4 sockets for VRAM (up to 128 bit data path!)
>hardware tables) with hardware protection against overwriting
>will be an easily replacable card)
> - CHRP compatible design (?)
>
So the question is where are Acorn buying these PC motherboards from?
see you!
RobT
--
_/_/_/ _/_/_/ _/_/_/ _/_/_/ _/_/_/ _/ _/ _/ _/
_/ _/ _/ _/ _/ _/ _/_/ _/ _/ _/
_/ _/ _/_/ _/_/_/ _/ _/ _/ _/ _/ _/_/
_/ _/ _/ _/ _/ _/ _/ _/_/ _/
_/_/_/ _/_/_/ _/_/_/ _/ _/_/_/ _/ _/ _/
http://mphhpc.ph.man.ac.uk/~mbcaprt/DESTINY.html for the Destiny home page
http://mphhpc.ph.man.ac.uk/~mbcaprt for my home page, 9p on a modem!
Stuart Bell
In article <19961020....@quantumsoft.co.uk>, stu...@quantumsoft.co.uk
(Stuart Halliday) wrote:
>
> But for my two pence worth the main priority is the data bus speed to cope
> with the 200MHz+ StrongARMs that will be coming out next year and an extra
> SIMM slot would be nice. Everything else is just a luxury IMHO.
>
--
*** Stuart Bell (sab...@argonet.co.uk) Running Risc OS and ***
*** RiscBSD on an Acorn Risc PC in a Wintel-free environment ***
*** with a StrongARM RISC processor running at 202.4MHz :-)***
Stuart Halliday
Stuart Bell wrote the following...
> In article <19961020....@quantumsoft.co.uk>, stu...@quantumsoft.co.
uk
> (Stuart Halliday) wrote:
> >
> > But for my two pence worth the main priority is the data bus speed to
cope
> > with the 200MHz+ StrongARMs that will be coming out next year and an
extra
> > SIMM slot would be nice. Everything else is just a luxury IMHO.
> >
> . . . and apart from more VRAM, surely, Stuart?
No, in my opinion extra VRAM is a luxury. I can live with 2MB of VRAM.
3MB of VRAM would be nice to get 1024x768 in 32bit or 1280x1024 in 15bit
for a big monitor, but I can't afford a 20"+ monitor to allow me to use
these large resolutions seriously.
But then the average customer doesn't need 2MB let alone 3MB+ do them?
I'm not saying that ART shouldn't provide the expansion capabilities just
that *very* few people would actually use it.
Perhaps in ART's eyes they would say why should everyone else pay to
support something that is hardly going to be used. I believe that was the
original argument about why the RiscPC only had 2 SIMM slots and 2MB of
VRAM in the first place. Has the Market altered to make ART change their
minds?
Well, the price of SIMMs has come down to make it affordable to buy more
than 2 SIMMS. Can a similar concept be said of VRAM?
Stuart Bell
In article <19961021....@quantumsoft.co.uk>, stu...@quantumsoft.co.uk
(Stuart Halliday) wrote:
>
> Stuart Bell wrote the following...
>
> > In article <19961020....@quantumsoft.co.uk>,
> stu...@quantumsoft.co.
> uk
> > (Stuart Halliday) wrote:
> > >
> > > But for my two pence worth the main priority is the data bus speed to
> cope
> > > with the 200MHz+ StrongARMs that will be coming out next year and an
> extra
> > > SIMM slot would be nice. Everything else is just a luxury IMHO.
> > >
> > . . . and apart from more VRAM, surely, Stuart?
>
> No, in my opinion extra VRAM is a luxury. I can live with 2MB of VRAM.
>
> 3MB of VRAM would be nice to get 1024x768 in 32bit or 1280x1024 in 15bit
> for a big monitor, but I can't afford a 20"+ monitor to allow me to use
> these large resolutions seriously.
I agree that you probably need a 21" screen for 1600 x 1200, but I use 1320
x 1024 (a slightly stretched 1280 x 1024) on my Iiyama 17" pro, and find it
very usable. However, I'm limited to 256 colours, even with 2Mb VRAM. To
get 16 bit colour I have to drop to 1112 x 834.
I don't think that asking a new motherboard, three years on from the first
RPC, to at least support 4Mb VRAM is unreasonable, especially if it is to be
in any way 'future proof'.
cheers,
Stefan Brueck
In article <3269EA...@acorn.new-cent.transit.ru>, Dmitry Petrov
<URL:mailto:dpe...@acorn.new-cent.transit.ru> wrote:
> > > Below are some details:
> > This seems to be very interesting. Where did you find this info?
> The above specification is just a necessary *minimum*.
More or less correct.
> Otherwise it will not be acceptable.
Even if all things you mentioned are there, it'll be less than I expect.
> Sorry if I dissappointed you.
Bet it's beyond your control to disappoint me (others) , since it's not
you playing dice which parts of a development-machine get stripped for
cost reasons.
> I thought the my original message will force people to think about the
> future and it will create some useful thread helping ART to go in a
> proper direction.
I would like ..... [it'll never become true] ....
> Best regards,
> Dmitry Petrov
2u2
Stefan Brueck
In article <na.44c24c471f....@argonet.co.uk>, Mark Gillman
<URL:mailto:mgil...@argonet.co.uk> wrote:
> > > > The new motherboard will have
> > > [etc]
> > > Says who?
> > It's putting ART's new policy of customer responsiveness into practice.
> So it's for real?
> And if it is, does this chap have permission to broadcast the news?
As long as he hasn't signed anything who'd be able to stop him ?
And even if he did sign, who could physically stop him ?
> I'm suprised to hear it this way, that's all.
I'm happy to hear at least something ... :-)
Hope to hear more news at Acorn World, eg. news, not the old hat named
SA.
cheerio
Patrick Herborn
In message <19961020....@quantumsoft.co.uk> Stuart Halliday wrote:
> So the bottom line is that you made it all up and had the cheek to post it
> as fact?
Ahhhr! The poor guy... he's been at a M$ training camp! :-)
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. Fax me no questions, I'll Fax you no lies!
Mr. R. Tierney
>
> So the bottom line is that you made it all up and had the cheek to post it
> as fact?
Ouch Stuart! That's a bit nasrsty considering Dmitry was trying to be
constructive, all he was trying to do is get people mentioning what they
would like to see on the new motherboard, and I for one think it is a much
needed discussion point.
> From rumours I've heard from various people ART are going to be producing
> a 3rd gen RiscPC motherboard, at present their Labs have various different
> models being tried out none of which could make it out of the door. So we
> should not take to much notice of any 'facts'.
That doesn't surprise me in the slightest. ART have all-but-confirmed that
they will release /two/ boards in the comming year (or so). One will be a
direct upgrade of the current Risc PC to allow more data-bus headroom for
such things like StrongARM and Hydra. The other will be some form of CHRP
board.
As you say...
> IMHO I think we'll see a RISCOS CHRP PCI card before a Acorn CHRP
> motherboard
I agree. What is likely to happen (and has been spoken quietly within ART)
is that the Guy's in Cambridge are going to get Risc OS and StrongARM
working in a modular form on a PCI plug-in card.
ART will get people like Digital and Xemplar to fund these developments.
Then, once the system works on the CHRP architecture, ART, suddenly, will
find that about 80% of the work (to make a /truly native/ version) will have
been done (and paid for) already. They just need then to do the necessary
development to make the system a native machine option.
>...in about 12 months time.
Take /that/ long? I hope not :-( Intel will be releasing their new machines
by then, and M$ will have (probably) managed to get Win'97 out by then too,
so we'll be trying to play catch-up again... :-(
For the former option, the Risc PC based board, Archive have already
suggested that the whole board is likely to be redesigned. They said that
the new motherboard for the Risc PC would probably cost around 475UKP
because that is about the minimum price they can build upgraded motherboards
for in the numbers which would be required (no matter what spec's it has).
Custom ART boards will always have this high premium because the market is
not just large enough. Even current Mk II boards cost well over 300UKP IIRC!
That will be one of the major advantages of the CHRP boards; they will be
built for PC users, PowerPC users, Sun (low-end) users and many, many more.
ART will just buy these mass-produced boards in (cheaply) and fit their
processors in instead. It may even allow Acorn users to buy motherboards
(like we can these days for PC's) to their chosen specifications and just
plug the Acorn items in to let it run, that would be quite nice IMHO.
I for one *will not* consider paying anything like 400+ (when I can get a
two-processor PPro one for under 100) /UNLESS/ it has at least most of
Dmitry's list as a Minimum! IMHO, it should be given most of the features
which Dmitry suggested to justify the high cost, if nothing else.
What I want is an excellent machine which does the work I want it to do; at
a reasonable cost. It's a lot to ask for, I know, but I don't want to find
that I can get PPro's (which does the job, but less efficiently) at half the
cost.
Even /I/ (Mr Acorn Fanatic) would find it hard to find the financial sence
to purchase a computer system because it costs more, has fewer programs
which are not capable of the 'I-can-do-anything-because-I-cost-mega-bucks',
and are used by 1% of the market in this country. It would mean the only
workable argument would be that it were British.
I just pray that ART can come up with a decent system quite quickly, to stay
ahead.
This is /my/ wish list (BTW)...
1) 8Mb VRAM should be made possible to keep up with current market trends in
the PC world. That means a new design for VIDC probably, and I have heard
(rumors only) that ARM are already working on that...
2) Four SIMM slots. EDO would be nice, or better still SDRAM IMHO.
3) 2nd Level cache should be an option, mainly because quite a few machines
are now running OS's which would benefit from it - RISC OS probably wouldn't
but BSD certainly will.
4) Very Fast bus architecture. We need, as a /bare/ minimum, to allow an SA
to stretch it's legs fully. Ideally, an SA powered Hydra should not be
forced to get a /real/ headache - and seeing as that option is not /that/
expensive for those who wish to use it (BTW, I would).
5) Enhanced-IDE support. Current RPC's /have/ got E-IDE Interfaces on the
main mother-board, but they are switched to use the bog-standard levels,
why? I don't know. Perhapse Pat could shed some light?
6) FP in hardware. This isn't really anything to do with the new
motherboard, but I think it's worth repeating, again, and again... ;-)
> But for my two pence worth the main priority is the data bus speed to cope
> with the 200MHz+ StrongARMs that will be coming out next year and an extra
> SIMM slot would be nice. Everything else is just a luxury IMHO.
You call a singular lone extra SIMM slot a necessity? At 400+ quid???
You can still get 64Mb (2 x 32Mb) for around 400UKP! Two more would be far,
far better for those who can't afford the larger sized SIMM's, as you seem
to be suggesting. Then people could still have 16Mb for under 100UKP. They
then don't need to spend much to upgrade from 4's to 8's.
High-powered users would then be able to (fairly) easily get 128Mb! I'm sure
many ppl would find 2 extra slots far more welcome.
Hmmm, that's a lot of stirring for only 2p! :-)
--
Cheers,
-The Tarmac Terrorist
Dmitry Petrov
Patrick Herborn wrote:
>
> In message <19961020....@quantumsoft.co.uk> Stuart Halliday wrote:
>
> > So the bottom line is that you made it all up and had the cheek to post it
> > as fact?
>
Dmitry
Gerry Iles
Mr. R. Tierney wrote:
>
> I for one *will not* consider paying anything like 400+ (when I can get a
> two-processor PPro one for under 100) /UNLESS/ it has at least most of
Are you sure about that? (Or is this price limited to you, personally?)
Gerry Iles
--
EMail: ge...@xara.com / All opinions in this message
/ are mine so go get your own.
Stuart Bell
In article <na.17c7784720...@argonet.co.uk>, Stuart Bell
<sab...@argonet.co.uk> wrote:
> I don't think that asking a new motherboard, three years on from the first
> RPC, to at least support 4Mb VRAM is unreasonable, especially if it is to
> be in any way 'future proof'.
>
And, I should have added, with VIDC(s) fast enough to get a decent refresh
rate with 32 bit colour at high resolutions. 63Hz for 16 bit colour at 1112
x 834 is simply too slow. 70Hz at 32 bit 1280 x 1024 should be made at least
possible if the new motherboard is to have the video circuit built-in.
Andre Timmermans
Enhancing the VIDC20 with support of more VRAM, Ok nice but
I think it should be first enhanced to give a better support
of low resolution screens. Why not allow the VIDC20 to replicate
pixels or lines of pixels send to the VDU, so that a 320 x 256 screen
would be send to the VDU as a 640 x 512 screen?
Also, IIRC the A4 LCD screen only supports one screen resolution.
Could the same replication of pixels or lines (completed perhaps
with sending "dummy" pixels or lines of pixels) be used to offer
support of different modes? Even if this meams a smaller
screen centred in the physical LCD pixel matrix, it is better
than nothing!
Andre Timmermans
t...@swn.sni.be
M R Atkinson
< everything snipped >
IMHO the order of priority for a new motherboard is
Memory bandwidth
Video capability
Expansion bus
Interfaces
Flash ROM for RISCOS
Sound
Memory bandwidth
----------------
This should be capable of supporting current StrongARM cards, a
cached StrongARM card at higher frequencies and a hydra made from
(cached) StrongARM cards. Wether it is 32 or 64 bit, uses EDO or
SDRAM, etc. is a design issue. However the bandwidth needed to
support the above configurations is approx. 500 Mbyte/s peak,
250 Mbyte/s average, 100Mbyte/s random access to single words.
This implies a 40ns (random) access time, with 8ns serial access
time.
32 bytes use only half the SIMM slots.
Video capability
----------------
Processor to video RAM bandwidth should be similar to that to
main memory. Displays of at least 1280 x 1024 in 32 bits should
be possible (implies 8Mbyte video RAM), higher resolutions desirable.
Upgradable to use 3D graphics chip (GLINT, or similar).
Video in and genlock as standard.
Expansion bus
-------------
2 slot backplane for current bus as legacy (i.e. no enhancements).
2 slot backplane for PCI bus.
Interfaces
----------
IDE - as standard, using high speed DMA.
SCSI III - as standard, using high speed DMA.
Serial - as standard, 2 high speed serial (better than 230kbits/s)
Parallel - as standard, bidirectional
PMC-CIA - as standard, NC personality card
MIDI - as standard?
Network - upgrade (because there are so many different types of
network)
using high speed DMA.
Flash ROM for RISCOS
--------------------
8 (16?) Mbyte of flash ROM for RISCOS. Hardware protected from
overwriting. Upgradeable without opening case (probably puting
key in lock and upgrading from CD ROM is the best option).
Sound
-----
CD quality stereo sound output as standard.
CD quality stereo sound input as standard.
Security
--------
Unique Machine ID. Something to make motherboards traceable.
=========
To round out the motherboard into a system; storage, power supply
case, keyboard, mouse, speakers and monitor are needed.
Storage
-------
Large hard disc
CD-ROM
DVD
Some backup system capable of saving all hard disc in one go
Power Supply
------------
Large enough for expansion
Case
----
Although the present one is OK, it could do with improvements, you can't
get at the motherboard without taking the whole thing apart, its quite
hard to fit disc drives, etc.
keyboard & mouse
----------------
Good quality ones essential.
Speakers
--------
Good quality stereo speakers.
Monitor
-------
Let the user choose.
--
Mike Atkinson (Senior Research Engineer)
E-Mail: M.R.At...@nortel.co.uk Nortel Technology
Telphone: +44 1279 402660 (6-742-2660) London Road, Harlow
http://bhars219.bnr.co.uk/~mra/ (behind firewall) Essex, CM17 9NA, UK
Andrew Clover
Mark Gillman <mgil...@argonet.co.uk> wrote:
>> It's putting ART's new policy of customer responsiveness into practice.
> So it's for real? And if it is, does this chap have permission to
> broadcast the news?
Educated rumours don't seem to be as discouraged under the current regime!
> I'm suprised to hear it this way, that's all.
Well, if they announced it officially they'd be held to it. :-)
BCNU, AjC
Patrick Herborn
In message <19961021....@quantumsoft.co.uk> Stuart Halliday wrote:
> 3MB of VRAM would be nice to get 1024x768 in 32bit or 1280x1024 in 15bit
> for a big monitor, but I can't afford a 20"+ monitor to allow me to use
> these large resolutions seriously.
1280x1024 is precfectly usable on a 17" monitor. Anything higher and VIDC
falls over, since it can't handle the PIXEL rate... I have some high-res
modes that are limitied not by memory bandwidth but by the VCO. I get to
about 145 Mpix / sec and that's about it. Pity, since my MT-9017E will do
1600x1200 @ 76 Hz....
> But then the average customer doesn't need 2MB let alone 3MB+ do them?
I wounldn't mind...
> I'm not saying that ART shouldn't provide the expansion capabilities just
> that *very* few people would actually use it.
Given the opportunity, many people would use it, just because it's there....
> Perhaps in ART's eyes they would say why should everyone else pay to
> support something that is hardly going to be used.
Perhaps.
> I believe that was the original argument about why the RiscPC only had
> 2 SIMM slots
Hmmm. IOMD only supports two SIMMs.. that might also have had something to do
with it...
> and 2MB of VRAM in the first place.
They /planned/ on 4MB but found they had trouble with memory bandwidth :-(
> Has the Market altered to make ART change their minds?
Let's hope so... :-)
> Well, the price of SIMMs has come down to make it affordable to buy more
> than 2 SIMMS.
Now if only it wasn't going uo again :-(
> Can a similar concept be said of VRAM?
Nope. VRAm is a different kettle of silicon...
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. No new mail - start whine-pout sequence (Y/N)?
Ian Molton
M R Atkinson <M.R.At...@nortel.co.uk> wrote:
> IMHO the order of priority for a new motherboard is
> Memory bandwidth Video capability Expansion bus Interfaces Flash ROM for
> RISCOS Sound
> Memory bandwidth ----------------
<Truly Humongous Snip>
And the person to volunteer to build this (rather nice) beastie is....
Um.
Rik Griffin
In message <326CFDD0...@nortel.co.uk> M R Atkinson wrote:
> Processor to video RAM bandwidth should be similar to that to
> main memory. Displays of at least 1280 x 1024 in 32 bits should
> be possible (implies 8Mbyte video RAM), higher resolutions desirable.
> Upgradable to use 3D graphics chip (GLINT, or similar).
>
> Video in and genlock as standard.
Right, and are you willing to pay another grand for a RiscPC? Features
that less than 5% (at a guess) of users are going to use belong on a
podule. Or PCI card or whatever.
> IDE - as standard, using high speed DMA.
> SCSI III - as standard, using high speed DMA.
Both as standard? Again, who is going to use this?
> Serial - as standard, 2 high speed serial (better than 230kbits/s)
Serial ports usually get used for mice and modems. I suppose ISDN / cable
modems would need the faster ports, and 2 ports would be very useful.
> Parallel - as standard, bidirectional
Of course.
> PMC-CIA - as standard, NC personality card
> MIDI - as standard?
More things that I think the average user would have no interest in.
> 8 (16?) Mbyte of flash ROM for RISCOS. Hardware protected from
> overwriting. Upgradeable without opening case (probably puting
> key in lock and upgrading from CD ROM is the best option).
How expensive is flash rom? If it's not cheap, I think a beter option
would be normal ROMs for the main version of RISCOS, with a half meg
(maybe) to load patches, replacement modules etc into.
> CD quality stereo sound output as standard.
We've got that already.
> CD quality stereo sound input as standard.
PCs do it cheaply, so I don't see why Acorn can't.
> Monitor
> -------
> Let the user choose.
Agreed!
--
Top 100 reasons why Captain Kirk is better than Captain Picard ...
48. If something doesn't speak English - it's toast.
Daniel Maloney
M R Atkinson wrote:
[massive, long snip]
I agree with most of this, but I think a) that it has quite a few of those
things already, and b) that your video expectations are a little optimistic,
but we have to look to the future right?
So in summary: Seconded.
Cheers,
Dan.
--
____
| \ __ _ _ __ # Where's my 202MHz StrongARM? In my RiscPC of course!! :-)
Iain Anderson
Patrick Herborn (p...@nanobyte.demon.co.uk) wrote:
: 1280x1024 is precfectly usable on a 17" monitor. Anything higher and VIDC
: falls over, since it can't handle the PIXEL rate... I have some high-res
: modes that are limitied not by memory bandwidth but by the VCO. I get to
: about 145 Mpix / sec and that's about it. Pity, since my MT-9017E will do
: 1600x1200 @ 76 Hz....
1600x1200@76 implies a pixel rate of about 200MHz. Even the MT-9017E has not
got that sort of bandwidth. It also hasn't got a fine enough AG to display all
of those pixels. Although I suppose it would be worth a go - it might be like
hardware anti-aliasing (?!) (time to warm up those ramdacs I think).
____
Iain
Iain Anderson
M R Atkinson (M.R.At...@nortel.co.uk) wrote:
: Video capability
: ----------------
: Processor to video RAM bandwidth should be similar to that to
: main memory. Displays of at least 1280 x 1024 in 32 bits should
: be possible (implies 8Mbyte video RAM)
I make it 5M for that.
: Interfaces
: ----------
: IDE - as standard, using high speed DMA.
EIDE. Supporting whatever those new fast DMA modes are.
: SCSI III - as standard, using high speed DMA.
Fat chance :(
: Flash ROM for RISCOS
: --------------------
: 8 (16?) Mbyte of flash ROM for RISCOS. Hardware protected from
: overwriting. Upgradeable without opening case (probably puting
: key in lock and upgrading from CD ROM is the best option).
I think I'd rather have to open the case actually. But it should be easy to
enable whatever needs enabling after taking the lid off.
: Storage
: -------
: Large hard disc
: CD-ROM
: DVD
Not so sure about DVD. (Are they even available for anything yet?)
: Some backup system capable of saving all hard disc in one go
Tricky. All the good backup systems (ie DAT on SCSI) seem to be still way
expensive.
Other than that, I totally agree.
____
Iain
Thomas Boroske
In message <54l7ve$2...@lyra.csx.cam.ac.uk> Owen M. Astley wrote:
> In article <54l99t$i...@prism.prismtech.co.uk>,
> Iain Anderson <ia...@prismtech.co.uk> wrote:
> How about using an expandable graphics system? Base computers have
> VIDC20 1/2 Mb VRAM. Expansion cards allow better features.
What about using some sort of very wide, very fast but cheap DRAM type ?
Of course that would mean that processor -> video ram bandwidth is slower,
but
1. In the desktop, the remaining bandwidth will be enough anyway, and
memory bandwidth isn't everything here (the picture has to be calculated
first).
2. For games, the screenmodes used will almost certainly not use the full
video bandwidth, so most of the bandwidth will be there for those
highly optimised plotting routines.
3. In the current RiscPC, the processor can't write to VRAM with the
maximum speed anyway (first, VRAM is clocked higher, but for write
access the DRAM speed is used, second, the data bus is only half as
wide as the video bus).
I think this will also be the case with the new motherboard. Even if
they make the data bus 64bit, they'll have to use 128bit for video
(how else can you use those 4MB or even 8MB of VRAM everybody wants ?).
In this case, it would be probably better to put a sort of buffer
just before the video memory, wich will be written 64bit wide by
the processor and will then write 128bit wide to the video memory.
I'm normally all for the power approach. The reason I suggest this
is that 4MB or more of VRAM will be far too expensive to justify in
any way, I think (8MB of SDRAM would be cheaper. Leaves the problem
to get it in small enough DIMMS to get the data bus wide enough).
Another solution probably would be to go the PC route entirely and use
a PCI graphics card. Or use some sort of standard-VRAM, this Acorn-only
solution is just too expensive.
Thomas
P.S.: Please correct me if my above assumptions about available bandwidth
are wrong.
--
Thomas Boroske
.. I can SPELL, I just can't TYPE worth a hoot !
Ian Molton
ia...@prismtech.co.uk (Iain Anderson) wrote:
> 1600x1200@76 implies a pixel rate of about 200MHz. Even the MT-9017E has
> not got that sort of bandwidth. It also hasn't got a fine enough AG to
> display all of those pixels. Although I suppose it would be worth a go -
> it might be like hardware anti-aliasing (?!) (time to warm up those
> ramdacs I think).
If you can get this mode @76 Hz, Gimme a modedef - Mine does 1600x1200@56
Oh yes. The AG is plenty fine enough - Clear as a bell.
-Ian
John Martin
M R Atkinson <M.R.At...@nortel.co.uk> wrote:
> Speakers --------
> Good quality stereo speakers.
I'd much sooner route the sound via a decent hifi system.
I do not _need_ another pair of speakers to clutter up my study.
> Monitor -------
> Let the user choose.
So, you'd let the user choose the monitor, but not the method of sound
output?
John
--
Owen M. Astley
>: main memory. Displays of at least 1280 x 1024 in 32 bits should
>: be possible (implies 8Mbyte video RAM)
>I make it 5M for that.
Even so, I think that 1280x1024 in 32 bits would cost *way* too much.
I'm afraid that I don't know much about the electronics of it, but comparing
it with PC graphics cards, 4M cards are still quite expensive, and remember
that Acorn computers tend to cost more anyway.
How about using an expandable graphics system? Base computers have
VIDC20 1/2 Mb VRAM. Expansion cards allow better features.
>: Interfaces
>: SCSI III - as standard, using high speed DMA.
>Fat chance :(
IIRC Acorn were considering putting SCSI (prob SCSI I) into the original
RPC, but didn't because of the cost.
I'm not sure that this is necessary with EIDE anyway (especially if you have
a 4 device motherboard)
>: Storage
>: Some backup system capable of saving all hard disc in one go
>Tricky. All the good backup systems (ie DAT on SCSI) seem to be still way
>expensive.
Agreed.
Basically there is always going to be a problem with cost.
The computers need to at least pretending to have a similar price to a PC.
The advantage of RISC OS is that it doesn't need huge resources, but this
price advantage isn't enough. Anything else must be considered optional,
surely?
Of course some of these things can be bought afterwards...SCSI, more IDE
hard discs, ram (I have 24Mb :-) ).
Owen
Thomas Boroske
In message <19961022....@cheesey.demon.co.uk> Rik Griffin wrote:
> In message <326CFDD0...@nortel.co.uk> M R Atkinson wrote:
>
> Serial ports usually get used for mice and modems. I suppose ISDN / cable
> modems would need the faster ports, and 2 ports would be very useful.
Yes, very inportant. I guess more and more people will use their serial
port for a modem / isdn 'modem' and that leaves no port free for
a graphics tablet / joystick / whatever.
--
Thomas Boroske
.. The surest way to be late is to have plenty of time.
D P Macdonald
: : Video capability
: : ----------------
I would have thought that at least the spec of a Diamond Stealth VRAM or
Matrox Millenium would be needed - most new PCs come with one or the
other. Something like expansion to 8Mb VRAM, maybe 1Mb on board to cut
costs - as this is really a minimum standard in the PC world.
Interfaces : :
----------
: EIDE. Supporting whatever those new fast DMA modes are.
Mode 4? - someone recently mentioned that RPC has EIDE but that it's high
speed modes have been disabled, any truth in this?
: : SCSI III - as standard, using high speed DMA.
Is SCSI III available on a card yet (is that Fast and Wide?)
Data Bus
--------
Probably the most important ATM. With a 233MHz SA now available a 33 or
66 MHz bus is really a necessity.
--
David MacDonald Trevelyan College, Durham, England
European Law Students Association, Durham
Home Page (unfinished): http://www.dur.ac.uk/~d402qa/
"There's a tale in that", Tammyn the bard.
David Gamble
> [snip]
> I don't think that asking a new motherboard, three years on from
the first
> RPC, to at least support 4Mb VRAM is unreasonable, especially if it
is to be
> in any way 'future proof'.
>
>
Future proof?
What possible use do you have for a screen mode larger than 1280 x
1024?
(except straining your eyes *badly*, or connecting to cinema screens)
----------------------------------------------
____ _ __
/ __ \___ __ __(_)__/ /
/ /_/ / _ `/ |/ / / _ /
/_____/\_,_/|___/_/\_,_/
(David Gamble : gda...@enterprise.net)
---------------------------------------------
Thomas Hoenl
In message <ant20174...@barc.demon.co.uk> Mike Wilson wrote:
> > > Dmitry Petrov
>
> How the hell does he know?
>
Maybe KGB?
Bye
Thomas
--
_____ _ _
|__ __|| | | | Es ist bemerkenswert, dass nur vielleicht 10% aller Pro-
| | | |_| | grammierer Programme ohne Verwendung von Flussdiagrammen er-
| | | _ | folgreich schreiben koennen. Ungluecklicherweise glauben aber
|_|O |_| |_|O 90%, dass sie der Gruppe dieser 10% angehören. (Rodnay Zaks)
... When I get a better mouse trap built, mice will be an endangered species.
M R Atkinson
Daniel Maloney wrote:
> M R Atkinson wrote:
> I agree with most of this, but I think a) that it has quite a few of those
> things already, and b) that your video expectations are a little optimistic,
> but we have to look to the future right?
I think my video expectations are just right for a motherboard released
in 6 moths time with a lifespan of 2 years.
But then I would wouldn't I?
M R Atkinson
> > main memory. Displays of at least 1280 x 1024 in 32 bits should
> > Upgradable to use 3D graphics chip (GLINT, or similar).
> >
> > Video in and genlock as standard.
>
> Right, and are you willing to pay another grand for a RiscPC? Features
> that less than 5% (at a guess) of users are going to use belong on a
> podule. Or PCI card or whatever.
< more of my standard features questioned >
The reason I put so many features in as standard is that with the
current (and probably future) small sales of the RISC PC it makes
good sense. Take video in and genlock: adding these features to
the VIDC and motherboard might cost 500,000 UKP in design and add
another 10 pounds per motherboard total cost per RISC PC sold is
(assuming 20,000 sold) is 35 UKP, so add something for profit, etc,
and we get 50 UKP extra cost.
5% of 20,000 is 1000 say the design cost of an add-in board is 300,000
add selling costs, distributer margins and profit and a selling price
of 1000 UKP is not too unreasonable. Finding 1000 people to buy
something at 1000 UKP is not easy.
All my "standard" features proably add less than 200 UKP to the selling
price over a bare bones system.
=========
Having thought about it some more I would make a change to the spec.
1 slot (no backplane) for current bus as legacy.
removing the backplane saves a bit of money and makes the system more
reliable. With all the features as standard few if any people would
require 2 legacy cards. It is a slight overkill to have 2 PCI bus
slots for the same reason.
==========
I should have made it clear that I'm not expecting this system to
appear tomorrow! So I have extrapolated in one or two areas.
SCSI III is probably the most difficult, I'm not sure about the
state of play as regards to chips, if there aren't samples available
now, it is unlikely there will be production versions available in
(say) 6 months. In this case SCSI II would do.
Similarly, DVD may not be available, however if it is I regard it as
extremely important to fit as standard because it will be the next
data distribution and interchange standard (however this isn't a
motherboard issue).
M R Atkinson
Iain Anderson wrote:
> M R Atkinson (M.R.At...@nortel.co.uk) wrote:
> : Processor to video RAM bandwidth should be similar to that to
> : main memory. Displays of at least 1280 x 1024 in 32 bits should
> : be possible (implies 8Mbyte video RAM)
> I make it 5M for that.
But you can't get 5M easily, so round up to 8M. I expect that
a low end system will be sub-equiped with 2M. This incidently
enables even larger screen resolutions, 1600x1200 or even higher
in less bits, but cheap monitors for these are unlikely to be
available in the next couple of years.
> : IDE - as standard, using high speed DMA.
> EIDE. Supporting whatever those new fast DMA modes are.
Yes latest standard, I have rather lost track of what is happening
with IDE these days, so I wasn't more specific in case I got it
wrong.
> : SCSI III - as standard, using high speed DMA.
> Fat change :(
Unlikely I agree, but we can hope! I would settle for SCSI II,
which I know to be available cheaply in single chip form from
several manufacturers.
> : 8 (16?) Mbyte of flash ROM for RISCOS. Hardware protected from
> : overwriting. Upgradeable without opening case (probably puting
> : key in lock and upgrading from CD ROM is the best option).
> I think I'd rather have to open the case actually. But it should be
> easy to enable whatever needs enabling after taking the lid off.
Yes you could do that, but whatever it is the whole machine should
not need to be dismantled.
> : DVD
> Not so sure about DVD. (Are they even available for anything yet?)
They should be available in the next few months, and should drop in
price rapidly if the sales predictions hold up.
> : Some backup system capable of saving all hard disc in one go
>
> Tricky. All the good backup systems (ie DAT on SCSI) seem to be
> still way expensive.
Yes, its probably going to be expensive, but backup is becoming
even more important as disc sizes increase (and it is now essential
for anything other than a games machine!).
Stuart Bell
I can get 1600x1200 to 62Hz - not really fast enough to be called
flicker-free.
Dickon Hood
Owen M. Astley (oma...@thor.cam.ac.uk) wrote:
: In article <54l99t$i...@prism.prismtech.co.uk>,
: >: Storage
: >: Some backup system capable of saving all hard disc in one go
: >Tricky. All the good backup systems (ie DAT on SCSI) seem to be still way
: >expensive.
: Agreed.
Rubbish. I paid #250 (ish) for my DAT drive. Very useful, too.
--
Dickon Hood.
Due to Windoze NiceTry, my .sig is out of order.
Normal service will be resumed as soon as possible.
We apologise for the inconvenience.
Dickon Hood
Patrick Herborn (p...@nanobyte.demon.co.uk) wrote:
: In message <na.dfd814471f....@argonet.co.uk> Daniel Maloney wrote:
: > Pat Herborn wrote:
: > > 128 bit data path. Cool. Does that go through a MUX into a VIDC20, or does
: > > it go to a new VIDC / multiple VIDCs ?
: >
: > More to the point, will our old VRAM work with it?
: One would hope so....
Yes indeed. Perhaps a second VRAM slot could be used?
: > How much VRAM is that, hopefully 8Mb?
: Hmm.. imagine 1600x1200 in true colour.... :-) Should just about fit...
Yes, but it'd have to be at about 100fps (I can see flicker above 75
:-( - I have mine at 1280x1024x75fps and that's still too low). I have
a DOS mode which I use for my PC card at 640x480x120fps - makes using
Linux consoles very pleasant.
: > > > - flash upgradable 256KB BIOS (self-diagnostics, low level routines,
: > > > hardware tables) with hardware protection against overwriting
: > > That's a novelty....
: I think it's a good idea, actually. This reduces ART's overheads (they don't
: need to produce different ROMs for each machine, and they don't need to put
: support in RISC OS for <every device imaginable>; they just put that in the
: BIOS) hence (hopefully) reducing cost.
Hardly. The OS generally needs to know about these things at a higher
level than that.
: > > Like a PCMCIA card type thing?
: >
: > Probably more likely an A5000 memory board style thing?
: Hmmm :-( How about a JEIDA (sp?) card? You can get 8MB DRAm ones, so it must
: be possible to make 4MB ROM ones... Just insert your OS like a disc....
A flash upgradable OS would be nice, 'tis true. It would remove part of
the need for my Cumana SCSI-II podule (currently I have RO 3.5 and three
1GB hard drives - new FileCore is on the podule).
: > > How much of it? This is 66 Mhz stuff... cool or what?
: >
: > How much will it cost?
: Probably no more than for PCs (ie UKP <not very much>). Current(ish) price
: for a 512kB pipeline cache module is UKP 17+VAT so it's not a significant
: cost, compared to the total cost of the machine.
Making it compatible with standard PC L2 caches would be useful (if
there s a standard, of course. I suspect not, but then I'm like that).
: > Will smaller ammounts be available for poor people?
: Are you worried that they were going to use 8MB of the stuff (like DEC) ? Or
: are you worried about whether the cache controller will support different
: amounts?
: > > What sort of PCI controller is being used? Is there a PCI <> DEBI bridge?
: >
: > Will there *be* any DEBI slots?
: One would hope so....
Very definitely.
: > How will this work in single slice machines?
: Presumably by using a backplane... but that rather defeats the object :-(
Badly?
: > > > - CHRP compatible design (?)
: > > Fair enough.
: > It that '(?)' possibly, or '(?)' probably?
: Wouldn't you like to know... :-)
Yes, please. I don't like CHRP and it seems to have died somewhat.
: > > > - socket for optional Ethernet and/or SCSI-2 chip
: > > On the motherboard? Now there's an idea!
: > I'll have to get one and flog off my Cumana card then!
: Well, why not? If it's Fast-Wide SCSI ][ then you'de expect a (very) large
: performance boost over your Cumana. It is choked by the DEBI bus :-(
It's not a new idea, either. AFAIK the Medussa (RPC prototype) machines
had SCSI on the motherboard.
: > > > - a separate socket for optional FPA accelerator
: > > Arm Ltd are working on an new FPA (well, they may be working on an FPA
: > > core, but a separate device? that would be interesting, if a bit pointless
: > > since SA-110 doesn't have an external copro bus)
: > Yeah, this is wierd! Maybe a sort-of FPA/FPE traps FP instructions, like
: > the FPE and pushes them over, like people keep suggesting with the 5x86
: > card? It certainly seems to suggest that StrongARM won't be getting an
: > on-board FPU in the near future, or this *would* be rather pointless.
: Hmmm... or /maybe/ DEC have decided to produce an SA-100 CPU. But if that's
: the case, how the hell do you get the signals for the FPA off the SA
: processor card? There aren't enough pins left :-( Or will you just use a MUX
: on the card, and a memory / not_fpa line ?
I wouldn't be surprised if DEC produced an SA with an FPU on chip fairly
soon. They are known to be interested in the hand-held embedded market
(ie. Newtons and the like), which would benefit somewhat with FP.
: > But why is this BIOS flash upgradable?
: Very simple. The cost of making ROMs is rather high, unless you make LOTS of
: them. Acorn doesn't produce LOTS of machine (unfortunately) so they can put a
: flash device on the motherboard and blow it with board specific routines at a
: much reduced cost.
But you're advocating a BIOS: Basic Input/Output System; the OS would be
loaded in from disc in the situation you're describing. A BIOS just
handles the hardware and does very little else.
: > Surely if they do it right, it will never need to be upgraded,
: That's true. I would think that they've used this route so they can reduce
: overheads. They could use socketed EPROMs, but then you'de have to fit them
: to the board. This way you can just mount the devices on the production line,
: and blow the board specific routines in situ....
: > just the ROMS, so why aren't they flash upgradable?
: Flash memory may be cheaper than EEPROM, but you don't get 10-a-penny. It's
: better to stick to the smallest amount that will allow you to get machine specific routines into it. That way you can produce the RISC OS ROMs in huge
: numbers (they will work on all hardware, since they contain NO hardware
: specific code). In those sort of number it becomes more economical....
But I don't think you could. Certain routine need to know what hardware
they're unning on, and they'd be in the wrong level of the OS if they
were in the BIOS.
: > Very wierd.
: Not really...
Yes, really.
: > > Where's the DEBI bus?
: > Good question.
: Well, I for one won't be crying if we lose the DEBI bus....
I will. technically, it's slower than the PCI bus, but in fact the
devices that you see on a PCI bus tend to be slower.
: > Also, what happens about the open bus?
: Presumably it stays on non-CHRP machines.
One would hope so.
: > Will there be more processor slots?
: Now /there's/ an idea! :-)
There is no reason why more slots couldn't be on the present motherboard
(AFAIK), that is the point about the Hydra card. What they've done is
to add a lot more functionality by adding the possibility of an L2 (or
L3 - depends on the processors) cache to it.
: > Will there be support for 64-bit processors?
: Problem with that is getting data to/from the CPU card. You'de need bigger
: slots. That shouldn't be a problem, though. Ie, the default bus master is the
: wide card, IF it's there, else it's narrow card....
Yuk, yuk, yuk.
: > Will the PC card and Hydra work in it?
: One would hope so.
Indeed. Although there might be no need for the hydra card.
: > Will there be a need for the Hydra?
: Just how far do you go with CPU sockets on the main board? ART's answer to
: that question will decide the answer to yours, I would presume.
: > This post just didn't answer enough of my questions. Sorry.
: If it /was/ for real, it gives us something to talk about... but WHY aren't
: ART talking to the Clan about this? They want to produce a high performance
: board, and who's likely to buy it? The Clan members perhaps? So why not
: encourage a dialogue about the board's design?
: > > > Right now Risc OS 3.7 has support for several different types of
: > > > motherboards with slightly different controllers.
: >
: > > Really? Cool!
: >
: > Hmmmm?
: I was referring to the very very very very slim chance it may even support
: the older hardware.... that would make a few people out there happy!
:-)
M R Atkinson
Owen M. Astley wrote:
> Even so, I think that 1280x1024 in 32 bits would cost *way* too much.
> I'm afraid that I don't know much about the electronics of it, but comparing
> it with PC graphics cards, 4M cards are still quite expensive, and remember
> that Acorn computers tend to cost more anyway.
>
> How about using an expandable graphics system? Base computers have
> VIDC20 1/2 Mb VRAM. Expansion cards allow better features.
I was talking about capabilities, it should be possible to sub-equip
with
2M or 4M of video RAM. Please note I did not say VRAM, it should be up
to
the designer to work out which is the cheapest alternative to meet the
specification. It may be possible to use SDRAM, cache DRAM, RAMBUS RAM
or
something else, it may even be possible to combine it with main memory.
<SNIP>
> Basically there is always going to be a problem with cost.
> The computers need to at least pretending to have a similar price to a PC.
> The advantage of RISC OS is that it doesn't need huge resources, but this
> price advantage isn't enough. Anything else must be considered optional,
> surely?
RISC PCs cannot compete on cost with low end PCs there are just not
enough made, so the alternative is to compete on features, ease of use
etc.
Anton Erasmus
M R Atkinson <M.R.At...@nortel.co.uk> wrote:
>< everything snipped >
>IMHO the order of priority for a new motherboard is
> Memory bandwidth
> Video capability
> Expansion bus
> Interfaces
> Flash ROM for RISCOS
> Sound
[Lots of detail snipped]
>Interfaces
>----------
>IDE - as standard, using high speed DMA.
>SCSI III - as standard, using high speed DMA.
>Serial - as standard, 2 high speed serial (better than 230kbits/s)
>Parallel - as standard, bidirectional
>PMC-CIA - as standard, NC personality card
>MIDI - as standard?
>Network - upgrade (because there are so many different types of
>network)
> using high speed DMA.
[More stuff snipped]
What about USB (Universal Serial Bus), This seems to be on the point
of becomming a standard in the PC market. It will be standard on the
new ATX PC motherboards being developed by Intel and others. This bus
will be fast enough for real time audio, and will also be suitable for
fast modems etc. The spec includes Plug&Play capability. Hence one
should be able to take a USB peripheral made for the PC market and use
it without problems on an Acorn with USB as standard. The USB can also
be used for a cheap and dirty network. The speed is something like 1
Mbits/s. AFAICR.
What about the new high speed serial busses for Disk drives ? Such as
Fire wire and SSA. AFAICR IBM or someone has developed an SSA chipset
which uses the ARM7 core. These two busses support speeds in the
region of 400Mbits/sec
All the above serial busses are of course hot plugable. One can remove
and ad peripherals while the system is running.
Cheers Anton
Iain Anderson
: > M R Atkinson (M.R.At...@nortel.co.uk) wrote:
: > : Displays of at least 1280 x 1024 in 32 bits should
: > : be possible (implies 8Mbyte video RAM)
: > I make it 5M for that.
: But you can't get 5M easily, so round up to 8M. I expect that
: a low end system will be sub-equiped with 2M.
I'm sure it could be done. Recently I saw a review of a PCI graphics adapter
that had 2.25MB on board. This allows 1024x768 at 24 bpp without rounding up
to 4MB.
____
Iain
Darren Salt
In article <na.03d6d04720...@argonet.co.uk>
Mr. R. Tierney <kra...@argonet.co.uk> wrote:
[snip]
> This is /my/ wish list (BTW)...
> 2) Four SIMM slots. EDO would be nice, or better still SDRAM IMHO.
How about setting jumpers to say whether you've got matched pairs of SIMMs?
Or even *four* matched SIMMs?
--
+---------------------------------------------------------------------------+
| Darren Salt - arc...@spuddy.mew.co.uk - ToonArmy - darre...@unn.ac.uk |
| Acorn A3010 - Spectrum +3 - BBC Master - Season ticket - Wintel free zone |
+-01268-515441-free-email-&-Usenet---Newcastle-5--Whingeing-Cockney-Mancs-0-+
When working towards a solution, it helps to know the answer.
Patrick Herborn
In message <54l99t$i...@prism.prismtech.co.uk> Iain Anderson wrote:
> : Displays of at least 1280 x 1024 in 32 bits should
> : be possible (implies 8Mbyte video RAM)
>
> I make it 5M for that.
And just how do you get 5M ????
> : Interfaces
> : ----------
>
> : IDE - as standard, using high speed DMA.
>
> EIDE. Supporting whatever those new fast DMA modes are.
The new modes are PIO, IIRC DMA isn't any faster, but it might be...
> : SCSI III - as standard, using high speed DMA.
>
> Fat chance :(
Pity :-(
> : 8 (16?) Mbyte of flash ROM for RISCOS. Hardware protected from
> : overwriting. Upgradeable without opening case (probably puting
> : key in lock and upgrading from CD ROM is the best option).
>
> I think I'd rather have to open the case actually. But it should be easy to
> enable whatever needs enabling after taking the lid off.
What about a lock like PCs have? That would save you having to open the case...
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. "Virtual" means never knowing where your next byte is coming from.
Stuart Bell
In article <54o7su$t...@prism.prismtech.co.uk>, ia...@prismtech.co.uk (Iain
> : > : Displays of at least 1280 x 1024 in 32 bits should
> : > : be possible (implies 8Mbyte video RAM)
>
> : > I make it 5M for that.
>
> : a low end system will be sub-equiped with 2M.
>
> I'm sure it could be done. Recently I saw a review of a PCI graphics
> adapter
> that had 2.25MB on board. This allows 1024x768 at 24 bpp without rounding
> up
> to 4MB.
Yep, as far as I understand it, if you have very wide access to the video
memory, that implies quite 'small' chips. EG for 4Mb of 128 bit wide stuff,
you could use 16 devices each 8 bits wide, each containing 256Kb - hence
the ability to have 'non-round' quantities of video RAM.
I'd be quite happy with 5242800 bytes of video RAM. :-)
cheers,
Patrick Herborn
In message <54l8re$i...@prism.prismtech.co.uk> Iain Anderson wrote:
> Patrick Herborn (p...@nanobyte.demon.co.uk) wrote:
>
> : 1280x1024 is precfectly usable on a 17" monitor. Anything higher and VIDC
> : falls over, since it can't handle the PIXEL rate... I have some high-res
> : modes that are limitied not by memory bandwidth but by the VCO. I get to
> : about 145 Mpix / sec and that's about it. Pity, since my MT-9017E will do
> : 1600x1200 @ 76 Hz....
>
> 1600x1200@76 implies a pixel rate of about 200MHz.
Eh? 1600x1200x76 = 145920000 pixels / second. Now add about 10% for the sync
and porch widths and you get 160512000 pixels / second. MT-9017E's analogue
bandwidth is 160 MHz, what a surprise... the figure is real.
> Even the MT-9017E has not got that sort of bandwidth.
Not 200 MHz, but since the bandwidth required is 160 MHz, and MT-9017E has a
160 MHz bandwidth, it will work nicely... even the line rate is OK... 91.2
kHz, which the MT-9017E can handle without breaking sweat.
> It also hasn't got a fine enough AG to display all of those pixels.
Probably not... let me just check... yeah, it's designed for 1280x1024.
> Although I suppose it would be worth a go
I already have / use 1600x1200 and 1600x1280. The refresh rate is the
problem... a bit flickery at 140 Mpix / sec :-(
> - it might be like hardware anti-aliasing (?!)
Haven't you ever displayed 1024 x 768 on an AKF18 ? Yes, you get hardware
antialiasing.
> (time to warm up those ramdacs I think).
Well, VIDC20 is out. Now, I remember seeing some lovely 25" monitors at a
pre-press exhibition I went to... analogue bandwidth of about 300 MHz... very
very serious piece of kit... and very very expensive. But then they were in
good company... other goodies on display included an Onyx, A0+ imagesetters,
ultra high resolution imagesetters (6000dpi+, but who needs it?), very big
RIPs (raster image processors) with many CPUs and hunders of megabytes of
RAM, laser printers that did 120 (yes 120) pages per minute.... get the
idea? Very serious kit...
> ____
> Iain
So if it isn't Iain, who is it? (logical negation using overscore) :-) :-)
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. Last Christmas I got a computer for my Girlfriend - Good Trade!
Stuart Halliday
Iain Anderson wrote the following...
>
> I'm sure it could be done. Recently I saw a review of a PCI graphics
adapter
> that had 2.25MB on board. This allows 1024x768 at 24 bpp without rounding
up
> to 4MB.
Remember Acorn's 24bit display actually is 32bit.
--
Stuart Halliday - Web Manager of the
___ ___ _ __ ___ _ _
/ _ \ __ ___ _ _ _ _ / __| _| |__ ___ _ \ \ / (_) | |__ _ __ _ ___
| _ / _/ _ \ '_| ' \ | (_| || | '_ \/ -_) '_\ V /| | | / _` / _` / -_)
|_| |_\__\___/_| |_||_| \___\_, |_.__/\___|_| \_/ |_|_|_\__,_\__, \___|
|__/ |___/
ZFC HM
http://www.cybervillage.co.uk/acorn/
Iain Anderson
Patrick Herborn (p...@nanobyte.demon.co.uk) wrote:
: In message <54l8re$i...@prism.prismtech.co.uk> Iain Anderson wrote:
: > 1600x1200@76 implies a pixel rate of about 200MHz.
: Eh? 1600x1200x76 = 145920000 pixels / second. Now add about 10% for the sync
: and porch widths and you get 160512000 pixels / second. MT-9017E's analogue
: bandwidth is 160 MHz, what a surprise... the figure is real.
Looks like I'm allowing too much for the porches. I got the figure by
extrapolating from a 1280x1024@85 mode that I use. Must be it uses hopelessly
conservative timings. Or yours are too aggressive. Or both. When I get a
working xfree I'll see what it actually needs (but isn't speculation great!)
: Haven't you ever displayed 1024 x 768 on an AKF18 ? Yes, you get hardware
: antialiasing.
Yeah. And 1152x480 regularly, and 1280x1024 once. It managed to sync okay, but
the entirely rubbish horizontal size control on the AKF18 wasn't quite enough
to keep the entire screen in the viewable area.
: > ____
: > Iain
: So if it isn't Iain, who is it? (logical negation using overscore) :-) :-)
Good grief. I've only been signing off like that for a couple of years and
people are starting to mock it already. Cheers.
====
Iain
- happy now ? :-)
Stuart Halliday
David Gamble wrote the following...
> Future proof?
> What possible use do you have for a screen mode larger than 1280 x
> 1024?
> (except straining your eyes *badly*, or connecting to cinema screens)
Well, widescreen 16:9 TV will be here within a decade.
So 1820x1024 will be required for a start to equal 1280x1024 in equivalent
height.
Thomas Boroske
In message <54o76n$p...@netty.york.ac.uk> Dickon Hood wrote:
> Owen M. Astley (oma...@thor.cam.ac.uk) wrote:
> : In article <54l99t$i...@prism.prismtech.co.uk>,
> : Iain Anderson <ia...@prismtech.co.uk> wrote:
> : >M R Atkinson (M.R.At...@nortel.co.uk) wrote:
>
> : >: Storage
> : >: Some backup system capable of saving all hard disc in one go
> : >Tricky. All the good backup systems (ie DAT on SCSI) seem to be still way
> : >expensive.
> : Agreed.
>
> Rubbish. I paid #250 (ish) for my DAT drive. Very useful, too.
Well, #250 (ish) is about quite a lot of money, don't you think so ?
Remember, ATM you can get an SA RiscPC for #100 extra, and this
surely is a #100 well invested.
I would love a backup medium, though.
--
Thomas Boroske
.. What are you doing?!? The message is over,GO AWAY!
David Higton
In message <ant231957ab5k5z&@gdavid.enterprise.net>, David Gamble
(gda...@enterprise.net) wrote:
>Future proof?
>What possible use do you have for a screen mode larger than 1280 x
>1024?
>(except straining your eyes *badly*, or connecting to cinema screens)
1600 x 1200 screen mode. The monitor is my son's, which he bought
(second hand) against my advice that it was a waste of money. Now I'm
using it, I really like it. I'll miss it badly when he takes the
monitor away!
I've also used Sun workstations with a *lot* of active pixels (I don't
know exactly how many, but I'm sure it was lots more than 1024 x 768),
and they're also nice. It's how much information and how much detail
you can present on a screen that makes it good. IMHO, the more pixels
you have, the *less* you strain your eyes - assuming, that is, that you
make text a reasonable size. Huge numbers of pixels on a tiny screen
require more pixels per character, but the result is characters which are
easier to read than the same size characters written in larger pixels.
Dave
--
___ ___ ___ ___ ___ ___
| / \/ \/ \/ \/ \/ \ | The Archimedes Bulletin Board in London |
| A R C A D E | +44(0)181 654 2212 (4 lines) Croydon, UK|
| Email & News @arcade.demon.co.uk | +44(0)181 655 4412 Fidonet#2:254/27.0 |
Richard Walker
In article <326CFDD0...@nortel.co.uk>, M R Atkinson
<URL:mailto:M.R.At...@nortel.co.uk> wrote:
>
> IMHO the order of priority for a new motherboard is
>
> Memory bandwidth
> Video capability
> Expansion bus
> Interfaces
> Flash ROM for RISCOS
> Sound
I'd agree - except for the last two, which I think should be
swapped round!
> Memory bandwidth
>
> This should be capable of supporting current StrongARM cards, a
> cached StrongARM card at higher frequencies and a hydra made from
> (cached) StrongARM cards.
Yes. It should allow a current SA card to run at full speed and
a Hydra-full to run (almost?) perfectly!
> Video capability
>
> Processor to video RAM bandwidth should be similar to that to
> main memory. Displays of at least 1280 x 1024 in 32 bits should
> be possible (implies 8Mbyte video RAM), higher resolutions desirable.
> Upgradable to use 3D graphics chip (GLINT, or similar).
Yes - up to 8Mb VRAM, in existing 2Mb cards, say? I don't think
it'll be upgradable, though :-(
> Video in and genlock as standard.
Video in?!?! You mean a digitiser? Why? Why not just add a suitable
expansion card for the (few) who'll need it?
> Expansion bus
>
> 2 slot backplane for current bus as legacy (i.e. no enhancements).
> 2 slot backplane for PCI bus.
IMHO: exactly the same DEBI and 2-slot backplane, but develop (or
give Atomwide to do!) a PCI adapter. (development of ISAdapter).
The idea is, DEBI would be _so fast_ it could handle a 'PCI bridge'.
Anyway? What do you want PCI as standard for? For the few who need
it, an optional 'system' would be fine. Anyway, PCI will be in ARTs
CHRP system! (the XMPC)
> Interfaces
>
> IDE - as standard, using high speed DMA.
> SCSI III - as standard, using high speed DMA.
> Serial - as standard, 2 high speed serial (better than 230kbits/s)
> Parallel - as standard, bidirectional
> PMC-CIA - as standard, NC personality card
> MIDI - as standard?
> Network - upgrade
SCSI III, is there any need? PMC-CIA, whassat? - PCMCIA? There is
a PCMCIA podule, which would be fine.
The motherboard must fit into the existing case! Where would these
extra ports go?
For 'sound' the 3.5" jack could be replaced with a mini-DIN or something
which splits out (with a special 6" cable - supplied) into the 3 x MIDI,
3.5" sound out, 3.5" sound in sockets... and _surround sound_!
motherboard
| 'sound' |
---------- O ----------------
|
|
| extension cable
/ \----\ (supplied)
/ \ \
/ \ \
3xMIDI in out
+ surround connectors!
> Flash ROM for RISCOS
>
> 8 (16?) Mbyte of flash ROM for RISCOS. Hardware protected from
> overwriting. Upgradeable without opening case (probably puting
> key in lock and upgrading from CD ROM is the best option).
Is this really essential? Perhaps so! I suppose once the NC is in
production, we could buy them and re-flast them to 'real' RISC OS and
use them as A7000s (after all, they'll have the same PCB won't they?!?!)
> Sound
>
> CD quality stereo sound output as standard.
> CD quality stereo sound input as standard.
Yup, inc. MIDI and surround sound!
> Case
>
> Although the present one is OK, it could do with improvements, you can't
> get at the motherboard without taking the whole thing apart, its quite
> hard to fit disc drives, etc.
How often do you take the motherboard out! ;-) The existing case it
fabulous!
> keyboard & mouse
>
> Good quality ones essential.
The existing ones are fine!
> Speakers
>
> Good quality stereo speakers.
Yes, definately!
> Monitor
>
> Let the user choose.
Exactly - supply without monitor!
--
Cheers, Richard.
-----------------------------------------------------------------------------
Richard Walker mailto:mano...@campus.bt.com --> BT CampusWorld is the
~~~~~~~~~~~~~~ http://www.manorsch.demon.co.uk --> worst ISP, bar AOL ;-)
-----------------------------------------------------------------------------
Ian Molton
David Gamble <gda...@enterprise.net> wrote:
> Future proof? What possible use do you have for a screen mode larger
> than 1280 x 1024? (except straining your eyes *badly*, or connecting to
> cinema screens)
Uhm i use 1600x1200x16cols @ 56hz *A LOT* It is very useful, and not a bit straining on a _DECENT_ monitor.
I for one would love this mode in 24 bit!!
Wether or not i could afford the VRAM is a different matter.
-Ian
Mr K Parker
In article <ant231957ab5k5z&@gdavid.enterprise.net>, David Gamble <gda...@enterprise.net>
wrote:
> In article <na.17c7784720...@argonet.co.uk>, Stuart Bell
> <URL:mailto:sab...@argonet.co.uk> wrote:
> > [snip]
> > I don't think that asking a new motherboard, three years on from
> > the first RPC, to at least support 4Mb VRAM is unreasonable, especially
> > if it is to be in any way 'future proof'.
> What possible use do you have for a screen mode larger than 1280 x
> 1024?
> (except straining your eyes *badly*, or connecting to cinema screens)
How about 1280 x 1024 x 16.8M colours - a bit more than 2Mb?
Me! I'd like 1600 x 1200 x 16.8M colours - but there again, I'm greedy :-)
Keith
--
----------------------------------------------------------------------------
kei...@argonet.co.uk PRINT and DESIGN
ART RiscPC - Wintel Outside no project too large
no commission too small
============================================================================
Patrick Herborn
In message <n21A...@supergas.demon.co.uk> John Martin wrote:
> M R Atkinson <M.R.At...@nortel.co.uk> wrote:
>
> > Speakers --------
>
> > Good quality stereo speakers.
>
> I'd much sooner route the sound via a decent hifi system.
Yes. Most definitely. There is no way that a cr*ppy set of multimedia
speakers can hope to compete. But then we don't all have to the rediculous
extremes I do... I just like to know that I /can/ generate 120dB at 20Hz...
> I do not _need_ another pair of speakers to clutter up my study.
Most definitely!
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. Mental Floss prevents Moral Decay.
Patrick Herborn
In message <54ltln$n...@mercury.dur.ac.uk> D P Macdonald wrote:
> I would have thought that at least the spec of a Diamond Stealth VRAM or
> Matrox Millenium would be needed
My tripple VIDC solution would blow those cards away, IMHO. How does
1280x1024 in true colour at 80Hz sound? Plus support for multiple screens and
hardware gamma / black correction... you view the original on one screen and
you view the gamma corrected on another... :-)
> - most new PCs come with one or the other.
Most of the higher end models, yes.
> Something like expansion to 8Mb VRAM,
Well, 6MB would be enough, IMVHO. 1600 x 1200 in true colour would fit into
that using my design. You could add another 2MB if you wanted opcaity
information or Alpha (?) channels...
> maybe 1Mb on board to cut costs - as this is really a minimum standard in
> the PC world.
Agreed.
> Interfaces : :
> ----------
>
> : EIDE. Supporting whatever those new fast DMA modes are.
>
> Mode 4?
Mode 4... I'de have to check the 665 datasheet to know for sure. I think it's
part of the ATA2 or ATA3 spec so wouldn't be supported. Somehow I don't think
that IOMD could keep up.. I would love to be proven wrong, of course. I think
it mught just get there using Type D cycles...
> someone recently mentioned that RPC has EIDE
That's a load of BULL. RPC has a bog standard IDE bus. EIDE allows access to
4 peripherals, and the 665 does not support that, IIRC.
> but that it's high speed modes have been disabled, any truth in this?
None, IIRC
> : : SCSI III - as standard, using high speed DMA.
>
> Is SCSI III available on a card yet (is that Fast and Wide?)
yYes and no, respectively.
> Data Bus
> --------
>
> Probably the most important ATM. With a 233MHz SA now available a 33 or
> 66 MHz bus is really a necessity.
The latter. Why only go half way?
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. (beep) Help, I've fallen and can't reach the beer.
Patrick Herborn
In message <n21B...@zetnet.co.uk> Ian Molton wrote:
> ia...@prismtech.co.uk (Iain Anderson) wrote:
[Re: 1600x1200@76 Hz]
> If you can get this mode @76 Hz, Gimme a modedef
Sorry. I can't get it yet. I haven't plucked up the courage to do the board
mod to enable this mode to work. I have to change a timing capacitor in the
VCO used the generate VIDC20's pixel clock. It has a current limit of about
150 Mpix/sec on a good day. I could extend that up to about 180 MPix/sec but
VIDC20 may fall over... it is rated at 110 MPix/sec
> - Mine does 1600x1200@56
I get 1600x1200@59Hz before VIDC20 throws a wobbly. Still trying to figure
out whether it's the VCO reaching it's limit of compliance causing clock
jitter or whether it's a data bandwidth problem. It could also be that at
these extremely high pixel rates the output from the 74ACT04 is just not
large enough to allow VIDC to lock on.
My 'standard' mode is 1280x1024@79 Hz :-)
> Oh yes. The AG is plenty fine enough - Clear as a bell.
Not exactly... but you still get a wonderful picture.
> -Ian
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. "Bother", said Pooh, as the pin fell out the grenade
Patrick Herborn
In message <19961022....@cheesey.demon.co.uk> Rik Griffin wrote:
> > IDE - as standard, using high speed DMA.
> > SCSI III - as standard, using high speed DMA.
>
> Both as standard? Again, who is going to use this?
CDROMs on ATA2 and HDs on UltraSCSI sounds OK to me....
> > Serial - as standard, 2 high speed serial (better than 230kbits/s)
>
> Serial ports usually get used for mice and modems.
Of which the latter will need more than 115k200 when you use ISDN channel
aggregation / compression.
> I suppose ISDN / cable modems would need the faster ports,
Oh yes. But wouldn't USB be better. How about Firewire ?
> and 2 ports would be very useful.
Indeed. Agreed. RPC's IO controller (37FDC665) has two serial ports, but
Acorn in their infinite wisdom, decided to connect only one. You could pop in
a mezannine board with the required 1488 and 1489 EIA RS232D receivers /
transmitters...
> > Parallel - as standard, bidirectional
>
> Of course.
One that does ALL the 1284 modes... SPP, Nibble, Byte, EPP, ECP, ECP+RLE....
> > PMC-CIA - as standard, NC personality card
> > MIDI - as standard?
>
> More things that I think the average user would have no interest in.
The 665 can do MIDI.... just needs a driver, IIRC.
> > 8 (16?) Mbyte of flash ROM for RISCOS. Hardware protected from
> > overwriting. Upgradeable without opening case (probably puting
> > key in lock and upgrading from CD ROM is the best option).
>
> How expensive is flash rom?
Expensive, compared to DRAM. Cheaper than EEPROM, though. I wouldn't even
dream about putting 8MB of the stuff on the board!
> If it's not cheap, I think a beter option would be normal ROMs for the main
> version of RISCOS, with a half meg (maybe) to load patches, replacement
> modules etc into.
Eh? The proposed solution was clean, simple and reliable. Why change it?
> > CD quality stereo sound output as standard.
>
> We've got that already.
Don't be silly! Having a DAC in the same box as the CPU is a no-no as far as
high quality audio goes. I would welcome AES-EBU or SP-DIF in/outputs. You
did know that IOMD already supports CD quality audio input, didn't you? All
these lovely things that ART could have donr, but didn't.
> > CD quality stereo sound input as standard.
>
> PCs do it cheaply, so I don't see why Acorn can't.
See above. All you need is an ADC..... Try the PCM78P... I think it's a
monolithic fully-monotonic converter IC with I²S compatible output...
> > Monitor
> > -------
> > Let the user choose.
>
> Agreed!
Oh yes. Most definitely. I thought it was absolutely REDICULOUS that I had to
join the Clan to get a RPC base without a monitor. Not that I minded
joining... I had already joined. I asked for a base unit without an HD (I had
a big HD). No go, not even as a Clan member; my dealer was VERY good, and
bought the standard drive back off me :-) Why this nonsense, though ?
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. Illiterate? Write for a free brochure!
Patrick Herborn
In message <19961023....@y0001006.tu-bs.de> Thomas Boroske wrote:
> What about using some sort of very wide, very fast but cheap DRAM type ?
They don't exist :-) Very fast and cheap are mutually exclusive. You might
get away with using very wide DRAM banks... say 256 bits wide, and then using
a MUX. But you'de need custom SIMMs again; I doubt people would be overly
enthusiastic about having to pop in 4 SIMMs a time!
> Of course that would mean that processor -> video ram bandwidth is slower,
> but
Use a cache controller to do write-behind cacheing... a bit of a hack, but it
might work..
> 1. In the desktop, the remaining bandwidth will be enough anyway, and
> memory bandwidth isn't everything here (the picture has to be calculated
> first).
Yes. True.
> 2. For games, the screenmodes used will almost certainly not use the full
> video bandwidth, so most of the bandwidth will be there for those
> highly optimised plotting routines.
Probably.
> 3. In the current RiscPC, the processor can't write to VRAM with the
> maximum speed anyway
Eh? Define 'maximum speed' please.
> (first, VRAM is clocked higher,
The SAM side is.. yes.
> but for write access the DRAM speed is used,
Yes. That's right. Any CPU access to VRAM uses MCLK.
> second, the data bus is only half as wide as the video bus).
Yeah. That's a consequence of the SAM side being twice as wide as the CPU's
data bus. On the DRAM side, the bus is only 32 bits wide.
> I think this will also be the case with the new motherboard.
And why not? Unless you put in more Video DACs... then you could go to 128,
192 or 256 bits wide on the SAM side... This is what I have proposed in the
past, and ART have allegedly looked at something along these lines.
> Even if they make the data bus 64bit,
Then you'de have 32 bits from the CPU to the L2 controller, and from the L2
to DRAM 64 bits wide... The L2 controller would need to control a MUX, of
course.
> they'll have to use 128bit for video (how else can you use those 4MB or
> even 8MB of VRAM everybody wants ?)
Clock the SAM port on the VRAM faster. That would require that VIDC20 could
keep up, of course. There is no reason why you can't clock SAM ports amore
than the 21.33 MHz that they currently run at. Of course you'de need to make
sure that VIDC's data bus could handle it...
> In this case, it would be probably better to put a sort of buffer
> just before the video memory, wich will be written 64bit wide by
> the processor and will then write 128bit wide to the video memory.
But that's what the VRAM does anyway. It's processor side data bus is only 32
bits wide! Yes the VIDC side is 64 bits wide. This achieved by using several
banks of VRAM.
> I'm normally all for the power approach.
Good to hear :-)
> The reason I suggest this is that 4MB or more of VRAM will be far too
> expensive to justify in any way,
We're talking about giving the board the /capability/ of more than 2MB. That
doesn't in any way imply that you /need/ more...
> I think (8MB of SDRAM would be cheaper.
Perhaps, yes. But SDRAM is still a bit on the expensive side...
> Leaves the problem to get it in small enough DIMMS to get the data bus wide
> enough).
DIMMs have a 64 bit data bus. You'de just need two of them to get the
proposed 128 bit bus...
> Another solution probably would be to go the PC route entirely and use
> a PCI graphics card.
Ouch! They are cr*p compared to VIDC20! Try doing 1200x870 ! :-)
> Or use some sort of standard-VRAM, this Acorn-only solution is just too
> expensive.
There is no such thing as standard VRAM. The only standard thing is perhaps
the JEDEC (?) packaging... but do you really want to install 16 chips rather
than a single DIMM ?
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. To shoot a mime, do you use a silencer?
Big Mad Drongo
David Gamble wrote:
> Future proof?
> What possible use do you have for a screen mode larger than 1280 x
> 1024?
> (except straining your eyes *badly*, or connecting to cinema screens)
None. Yet. That's why it's called 'Future proof'.
Besides, why would a 2560x2048 screen mode strain your eyes? It doesn't
mean you *have* to have everything very small - you could just have
higher resolution graphics.
Adrian
Ben Dooks
In article <54o76n$p...@netty.york.ac.uk>,
dah...@york.ac.uk (Dickon Hood) you scribbled:
> : Agreed.
>
> Rubbish. I paid #250 (ish) for my DAT drive. Very useful, too.
Dickon, you bought that Drive of me, and I got it at a clearance sale.
--
Ben Dooks, b...@doggysoft.co.uk | We'll take our hearts outside. Leave
http://www.fluff.demon.co.uk/ | our lives behind, and watch the stars go out.
http://www.oaktree.co.uk/ |
Theo Markettos
In message <19961024....@y0001006.tu-bs.de> Thomas Boroske wrote:
> In message <19961022....@cheesey.demon.co.uk> Rik Griffin wrote:
>
> > In message <326CFDD0...@nortel.co.uk> M R Atkinson wrote:
> >
> > Serial ports usually get used for mice and modems. I suppose ISDN / cable
> > modems would need the faster ports, and 2 ports would be very useful.
>
> Yes, very inportant. I guess more and more people will use their serial
> port for a modem / isdn 'modem' and that leaves no port free for
> a graphics tablet / joystick / whatever.
There are already 2 serial ports on RPC boards, just Acorn didn't bother to
take any lines off the Super IO chip. They didn't even take them to somewhere
where a header could be put in.
--
Theo Markettos th...@marketto.demon.co.uk
Liphook HardBack's home page:
Hampshire http://www.marketto.demon.co.uk/hardback/index.htm
UK http://www.townpages.org/diary/Liphook/index.htm
Ian Molton
th...@marketto.demon.co.uk (Theo Markettos) wrote:
> There are already 2 serial ports on RPC boards, just Acorn didn't bother
> to take any lines off the Super IO chip. They didn't even take them to
> somewhere where a header could be put in.
How is it accessed, and where (on the PCB) do i put my soldering iron???
-Ian
Ian Molton
Richard Walker <mano...@campus.bt.com> wrote:
<stuff about soundsystem snipped>
How about instead of you're Idea, a system where a SIC (sound interface card) plugs into the computer just behind the nine pin minidin, so you can choose what you want?
Output Options:-
1) None???!!!
2) Stereo 16bit
3) Stereo Surround
In the (hypothetical) slot behind this one:-
Input options:-
1) Streo 16bit in
2) Stereo 16bit surround in
3) 1 with dual midi
4) 2 with dual midi
Pete Sykes
In article <54ltln$n...@mercury.dur.ac.uk>, D P Macdonald
<D.P.Ma...@durham.ac.uk> writes
>other. Something like expansion to 8Mb VRAM, maybe 1Mb on board to cut
>costs - as this is really a minimum standard in the PC world.
Not quite true really - 2Mb is now the standard. It's unusual to see a
machine advertised with >2Mb VRAM and there are still not all that many
PC graphics adaptors that can take 8Mb.
--
Pete Sykes
Mike Curtis
>
> Future proof?
> What possible use do you have for a screen mode larger than 1280 x
> 1024?
> (except straining your eyes *badly*, or connecting to cinema screens)
19 or 21 inch monitors, even good quality 17 inch ones (like mine :-).
Things like this give the machine a professional "feel" and it's the
"professional" market that is the one (IMHO) that Acorn needs to impress.
--
_____ _
|_ _|__ _ __ | | ____ _ to...@argonet.co.uk
| |/ _ \| '_ \| |/ / _` | Mike Curtis - running at 200MHz
| | (_) | | | | < (_| | ZFC A-
|_|\___/|_| |_|_|\_\__,_| Chester UK
Patrick Herborn
In message <54o70m$p...@netty.york.ac.uk> Dickon Hood wrote:
> Yes, but it'd have to be at about 100fps
Could cause a problem with pixel rate....
> (I can see flicker above 75 :-(
Perhaps you have a mon with a short persistence phosphor?
> I have mine at 1280x1024x75fps and that's still too low
I use 1280x1024@79Hz.... VIDC limit :-(
> I have a DOS mode which I use for my PC card at 640x480x120fps - makes using
> Linux consoles very pleasant.
My fastest refreshing mode runs at a cool 160Hz....
> Hardly. The OS generally needs to know about these things at a higher
> level than that.
No, I don't think so. As they say... all problems can be solved by adding
another level of abstraction (or something like that...). I don't see any
reason why you can't fit all your device drivers into 256k
> : Hmmm :-( How about a JEIDA (sp?) card? You can get 8MB DRAm ones, so it
> : must be possible to make 4MB ROM ones... Just insert your OS like a
> : disc....
>
> A flash upgradable OS would be nice, 'tis true.
Well, I'de go with Flash upgradable hardware dependent bits, but ROM cards
fot the bulk of the OS.
> It would remove part of the need for my Cumana SCSI-II podule
Perhaps.
> (currently I have RO 3.5 and three 1GB hard drives - new FileCore is on the
> podule).
I have RO 3.5, and 850MB drive, and I boot off the 850 :-)
> : Probably no more than for PCs (ie UKP <not very much>). Current(ish) price
> : for a 512kB pipeline cache module is UKP 17+VAT so it's not a significant
> : cost, compared to the total cost of the machine.
>
> Making it compatible with standard PC L2 caches would be useful (if
> there s a standard, of course. I suspect not, but then I'm like that).
L2 cache is SRAM. Surely making it compatible with SRAM isn't that
difficult... and if you can't then you don't have L2 cache at all since there
is no DRAM that runs that fast.
> : > Will there *be* any DEBI slots?
>
> : One would hope so....
>
> Very definitely.
Well... losing DEBI isn't the worst thing I could think of...
> : > How will this work in single slice machines?
>
> : Presumably by using a backplane... but that rather defeats the object :-(
>
> Badly?
Oh yes.
> : > It that '(?)' possibly, or '(?)' probably?
>
> : Wouldn't you like to know... :-)
>
> Yes, please. I don't like CHRP and it seems to have died somewhat.
It was 'on hold' last I heard....
> : Well, why not? If it's Fast-Wide SCSI ][ then you'de expect a (very) large
> : performance boost over your Cumana. It is choked by the DEBI bus :-(
>
> It's not a new idea, either. AFAIK the Medussa (RPC prototype) machines
> had SCSI on the motherboard.
Yes. I have heard that one too. My Medusa schematic doesn't show any SCSI
device, but I probably have a later revision. But it does show that ART
(Acorn) were toying with the idea of 4MB floppies and true DMA to the FDC.
> : Hmmm... or /maybe/ DEC have decided to produce an SA-100 CPU. But if that's
> : the case, how the hell do you get the signals for the FPA off the SA
> : processor card? There aren't enough pins left :-( Or will you just use a MUX
> : on the card, and a memory / not_fpa line ?
>
> I wouldn't be surprised if DEC produced an SA with an FPU on chip fairly
> soon.
SA1500FE....
> They are known to be interested in the hand-held embedded market
> (ie. Newtons and the like), which would benefit somewhat with FP.
To an extent, yes.
> But you're advocating a BIOS: Basic Input/Output System; the OS would be
> loaded in from disc in the situation you're describing. A BIOS just
> handles the hardware and does very little else.
OK. A little bit of a misnomer on my part. I meant that hardware specific
code should be in the FlashROM. Whether this equates to a BIOS is a bit
vague... having just vdu, kbd, parallel, serial, floppy, ide, scsi and
<whatever else> drivers in FlashROM is actually less than a BIOS. It is
useless without the rest of the OS. This would be on ROM and map into
0x3840000 (?) upwards.... with the 'BIOS' between 0x3800000 and 0x383FFFF
> But I don't think you could. Certain routine need to know what hardware
> they're unning on, and they'd be in the wrong level of the OS if they
> were in the BIOS.
No. You just define virtual device interfaces.
> : > Very wierd.
>
> : Not really...
>
> Yes, really.
Not really :-) :-)
> : Well, I for one won't be crying if we lose the DEBI bus....
>
> I will. technically, it's slower than the PCI bus, but in fact the
> devices that you see on a PCI bus tend to be slower.
Maybe. Maybe not. But this is more down to the fact that PCI peripherals are
built to a price not a quailty criterion. Most of them are cr*p, and those
that aren't are expensive, but blow away anything in the Arc.
> : > Also, what happens about the open bus?
>
> : Presumably it stays on non-CHRP machines.
>
> One would hope so.
Yes. Here I must agree.
> : > Will there be more processor slots?
>
> : Now /there's/ an idea! :-)
>
> There is no reason why more slots couldn't be on the present motherboard
> (AFAIK),
Apart from the nightmare of arbitration. Two CPUs is bad enough... Gareth was
a brave man to try to get more to work!
> that is the point about the Hydra card.
Yes.
> What they've done is to add a lot more functionality by adding the
> possibility of an L2 (or L3 - depends on the processors) cache to it.
Hmm... IMHO L2 cache doe NOT belong on the mainboard in parallel systems.
> : > Will there be support for 64-bit processors?
>
> : Problem with that is getting data to/from the CPU card. You'de need bigger
> : slots. That shouldn't be a problem, though. Ie, the default bus master is
> : the wide card, IF it's there, else it's narrow card....
>
> Yuk, yuk, yuk.
Better than sticking with 32 bits.... You'de only need one line that's pulled
high on the mainboard with a 4k7 resistor... it could be pulled low by
connecting the line to GND on the card....
> : > Will the PC card and Hydra work in it?
>
> : One would hope so.
>
> Indeed. Although there might be no need for the hydra card.
Yes.
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. I am Popeye of Borg. Prepare to be askimilgrated.
Patrick Herborn
In message <ant231957ab5k5z&@gdavid.enterprise.net> David Gamble wrote:
> What possible use do you have for a screen mode larger than 1280 x
> 1024? (except straining your eyes *badly*, or connecting to cinema screens)
CAD on big (21") screens. BTW 1600x1200 is fine on my MT-9017E....
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. I call my computer Hole in the Desk
Patrick Herborn
In message <4722101F7D%arc...@spuddy.mew.co.uk> Darren Salt wrote:
> In article <na.03d6d04720...@argonet.co.uk>
> Mr. R. Tierney <kra...@argonet.co.uk> wrote:
> [snip]
> > This is /my/ wish list (BTW)...
>
> > 2) Four SIMM slots. EDO would be nice, or better still SDRAM IMHO.
>
> How about setting jumpers to say whether you've got matched pairs of SIMMs?
The OS can determine that for itself...
> Or even *four* matched SIMMs?
:-)
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. LHC : Lacks Hard Currency
Patrick Herborn
In message <54q9ev$5...@prism.prismtech.co.uk> Iain Anderson wrote:
> Looks like I'm allowing too much for the porches.
Perhaps.
> I got the figure by extrapolating from a 1280x1024@85 mode that I use.
You are very lucky to get 85Hz out of you VIDC. What pixel rate are you using?
> Must be it uses hopelessly conservative timings.
Aha! I use hopelessly hacked timings just to get 79Hz :-(
> Or yours are too aggressive. Or both.
Well, they generally say that the number of pixels / seconds is roughly
(visible pixels)*1.1 ....
> When I get a working xfree I'll see what it actually needs (but isn't
> speculation great!)
Yup. :-)
> : Haven't you ever displayed 1024 x 768 on an AKF18 ? Yes, you get hardware
> : antialiasing.
>
> Yeah. And 1152x480 regularly,
Oh... that's taking me back a while...
> and 1280x1024 once.
Never managed that :-(
> It managed to sync okay, but the entirely rubbish horizontal size control
> on the AKF18 wasn't quite enough to keep the entire screen in the viewable
> area.
You were lucky to get anything at all! :-)
> : > ____
> : > Iain
>
> : So if it isn't Iain, who is it? (logical negation using overscore) :-) :-)
>
> Good grief. I've only been signing off like that for a couple of years and
> people are starting to mock it already. Cheers.
Sorry! There were /two/ smilies...
> ====
> Iain
>
> - happy now ? :-)
So it /IS/ Iain now (double negation) :-) :-)
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. Anything worth doing is worth overdoing.
Patrick Herborn
In message <19961025....@quantumsoft.co.uk> Stuart Halliday wrote:
> Remember Acorn's 24bit display actually is 32bit.
With their current hardware, yes. But if they were to use a tripple VIDC
solution, it could be dropped to 24 bits / pixel.
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. Claravoiant meeting canceled due to unforseen events.
Patrick Herborn
In message <961026...@arcade.demon.co.uk> David Higton wrote:
> I'm writing this on a RISC PC connected to a 21" monitor and using a
> 1600 x 1200 screen mode. The monitor is my son's, which he bought
> (second hand) against my advice that it was a waste of money.
Buying a good monitor is NEVER a waste of money, if you use your computer
regularly. The monitor must be the most underrated piece of computer hardware
around.
> Now I'm using it, I really like it. I'll miss it badly when he takes the
> monitor away!
'nuff said!
> I've also used Sun workstations with a *lot* of active pixels (I don't
> know exactly how many, but I'm sure it was lots more than 1024 x 768),
> and they're also nice.
I like Sun's BIG monitors.... :-)
> It's how much information and how much detail you can present on a screen
> that makes it good. IMHO, the more pixels you have, the *less* you strain
> your eyes - assuming, that is, that you make text a reasonable size.
Yes. Of course.
> Huge numbers of pixels on a tiny screen require more pixels per character,
> but the result is characters which are easier to read than the same size
> characters written in larger pixels.
Quite right.
> Dave
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. Our necessities are few but our wants are endless...
Patrick Herborn
In message <19961026....@marketto.demon.co.uk> Theo Markettos wrote:
> There are already 2 serial ports on RPC boards, just Acorn didn't bother to
> take any lines off the Super IO chip. They didn't even take them to somewhere
> where a header could be put in.
Yes. I never could understand that. Still there's nothing stopping you from
soldering wires onto the 665 and taking them to a mezannine board with the
necessary drivers...
Cheers,
Pat.
--
=============================================================================.
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. Discoveries are made by not following instructions.
Chris Rutter
In article <urIDNHAM...@fship.demon.co.uk>,
pe...@fship.demon.co.uk (Pete Sykes) wrote:
> Not quite true really - 2Mb is now the standard. It's unusual to see a
> machine advertised with >2Mb VRAM and there are still not all that many
> PC graphics adaptors that can take 8Mb.
Which is _all_ the more reason to make the RiscPC take 8 mb! Let's say
Acorn agree in one years time to sport 4 megs -- most Acorn products normally
start getting properly sold and distributed about 2/3 years after Acorn
claim they will, so in the year 2000, Acorn will have 4 megs of VRAM. PC by
that time will be selling 16 meg VRAM upgrades probably -- Acorn _need_ to
keep in front with the time delay there is on producing new products.
Cheers,
Chris.
--
// Chris Rutter <ch...@collegium.co.uk>
/ QTMPlay v2.24b (http://www.doggysoft.co.uk/qtmplay/)
Ben Dooks
In article <n23A...@zetnet.co.uk>,
ha...@zetnet.co.uk (Ian Molton) you scribbled:
> th...@marketto.demon.co.uk (Theo Markettos) wrote:
>
> > There are already 2 serial ports on RPC boards, just Acorn didn't bother
> > to take any lines off the Super IO chip. They didn't even take them to
> > somewhere where a header could be put in.
>
> How is it accessed, and where (on the PCB) do i put my soldering iron???
The combo controller chip has the extra serial lines. You'll need to put
them through a buffer...
I tried this, and found that soldering those FPGA style devices is both
hard and fiddly... and I gave up in the end.
Simon Farnsworth
Another suggestion: Make EVERYTHING easy to upgrade. Exactly how I do not
care, but maybe boards, like the current processor boards, or socketed chips,
like in my A310!
--
Simon Farnsworth
Mal McClenaghan
p...@nanobyte.demon.co.uk (Patrick Herborn) wrote:
> Oh yes. Most definitely. I thought it was absolutely
> REDICULOUS that I had to join the Clan to get a
> RPC base without a monitor. Not that I minded
> joining... I had already joined. I asked for a
> base unit without an HD (I had a big HD). No go,
> not even as a Clan member; my dealer was
> VERY good, and bought the standard drive back off
> me :-) Why this nonsense, though ?
I didn't realise that this was still an issue.
Oz customers can buy RPCs and 7000s without monitors,
from any dealer, and most will be happy to spec the
machine as you want it. If a customer wants a new
RPC with SCSI II and a 1gig drive they shouldn't,
have to take the IDE that comes with the unit.
Mal
--
____________________________________________________
Mal McClenaghan the image factory
Tel: +61 3 9458 3599 PO Box 599,
Fax: +61 3 9458 3599 Rosanna,
Email: imag...@ozemail.com.au Victoria 3084,
Australia
http://www.ozemail.com.au/~imagfact/index.html
____________________________________________________
Sveinung W. Tengelsen
In article <19961025....@nanobyte.demon.co.uk>, Patrick Herborn
<URL:mailto:p...@nanobyte.demon.co.uk> wrote:
>
> In message <54ltln$n...@mercury.dur.ac.uk> D P Macdonald wrote:
>
> > I would have thought that at least the spec of a Diamond Stealth VRAM or
> > Matrox Millenium would be needed
>
> My tripple VIDC solution would blow those cards away, IMHO. How does
> 1280x1024 in true colour at 80Hz sound? Plus support for multiple screens and
> hardware gamma / black correction... you view the original on one screen and
> you view the gamma corrected on another... :-)
I've got three words, Patrick - "Thank you, YES!!!" :)))
--
Best regards,
Sveinung W. Tengelsen
a.k.a. pixeleyes
----------------------------------------------------------------------
| Dolphin Design is StrongARMed | +47 22 42 97 58 - pixe...@sn.no |
Stuart Halliday
Patrick Herborn wrote the following...
> In message <54ltln$n...@mercury.dur.ac.uk> D P Macdonald wrote:
>
> > I would have thought that at least the spec of a Diamond Stealth VRAM or
> > Matrox Millenium would be needed
>
> My tripple VIDC solution would blow those cards away, IMHO. How does
> 1280x1024 in true colour at 80Hz sound? Plus support for multiple screens
and
> hardware gamma / black correction... you view the original on one screen
and
> you view the gamma corrected on another... :-)
Now I hear rumours sometime ago that Acorn could do this with three VIDC20
is it really possible?
--
Stuart Halliday - Web Manager of the
___ ___ _ __ ___ _ _
/ _ \ __ ___ _ _ _ _ / __| _| |__ ___ _ \ \ / (_) | |__ _ __ _ ___
| _ / _/ _ \ '_| ' \ | (_| || | '_ \/ -_) '_\ V /| | | / _` / _` / -_)
|_| |_\__\___/_| |_||_| \___\_, |_.__/\___|_| \_/ |_|_|_\__,_\__, \___|
|__/ |___/
ZFC HM
http://www.cybervillage.co.uk/acorn/
Patrick Herborn
In message <n23A...@zetnet.co.uk> Ian Molton wrote:
> How is it accessed, and where (on the PCB) do i put my soldering iron???
Public Service Anouncement : Kids, don't try this at home!
They are on the SMC37C665.....
Signal Pin
RI 86
DCD 87
RXD 88
TXD 89
DSR 90
RTS 91
CTS 92
DTR 93
You will probably need to poke about with the 665 config registers. Point
your browser at htpp://www.smc.com/ and follow the links for the datasheet.
It is in PDF (goddammit!) but you can view it under 'doze....
The 665 sits at 0x3010000 and all offsets from that address must be multipled
by 4 (so offset 3F0 would become FC0). A[0] and A[1] are not connected to the
665...
You will need a 1488 and a 1489 to drive your new serial port... or you
coould go with a MAX232 which does the hard work for you... If you go with
the latter, you won't need to try to find the +/- 12 volts needed since the
MAX232 generates it's own +/- 10 volt supply using charge pumps from the +5
volt supply (much more accessible!)
> -Ian
Cheers,
Pat.
--
·=============================================================================·
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. REALITY.SYS Corrupted: Re-boot Universe (Y/N/Q)
Richard Walker
In article <n23A...@zetnet.co.uk>, Ian Molton
Hmm, but it would be an _external_ and _optional_ box!
I think it would all fit on the motherboard easily at no great cost,
and 6" of cable surely won't add much to the cost! (or add much
inconvenience!)
Just thought, this 'box' idea reminds me of the dongles... :-)
--
Cheers, Richard.
-----------------------------------------------------------------------------
Richard Walker mailto:mano...@campus.bt.com --> BT CampusWorld is the
~~~~~~~~~~~~~~ http://www.manorsch.demon.co.uk --> worst ISP, bar AOL ;-)
-----------------------------------------------------------------------------
Iain Anderson
Patrick Herborn (p...@nanobyte.demon.co.uk) wrote:
: In message <54q9ev$5...@prism.prismtech.co.uk> Iain Anderson wrote:
: > Looks like I'm allowing too much for the porches.
: Perhaps.
: > I got the figure by extrapolating from a 1280x1024@85 mode that I use.
: You are very lucky to get 85Hz out of you VIDC. What pixel rate are you using?
Ho, ho. Did I mention that my VIDC only runs up to 36 Mpixels/sec (and that
only works in 16 colours). 1280x1024@85 is what I use on my, um, _other_
computer. It ain't an Acorn (*blush*). Pixel rate for that mode is 160 MHz.
====
Iain
Thomas Boroske
In message <n246...@ozemail.com.au> Mal McClenaghan wrote:
> p...@nanobyte.demon.co.uk (Patrick Herborn) wrote:
>
> > Oh yes. Most definitely. I thought it was absolutely
> > REDICULOUS that I had to join the Clan to get a
> > RPC base without a monitor. Not that I minded
> > joining... I had already joined. I asked for a
> > base unit without an HD (I had a big HD). No go,
> > not even as a Clan member; my dealer was
> > VERY good, and bought the standard drive back off
> > me :-) Why this nonsense, though ?
>
> I didn't realise that this was still an issue.
> Oz customers can buy RPCs and 7000s without monitors,
> from any dealer, and most will be happy to spec the
> machine as you want it.
Yep, it's the same even here in Germany. Is this really still
an issue in the UK ?
--
Thomas Boroske
.. RAM DISK is NOT an installation procedure!
Patrick Herborn
In message <ant28154...@hasle.sn.no> "Sveinung W. Tengelsen" wrote:
> I've got three words, Patrick - "Thank you, YES!!!" :)))
I think I came up with the idea at about the same time as Acorn... I remember
that when I suggested it to Acorn a few weeks after I had the idea I got the
reply "we are looking at that at the moment" .... :-)
So, now you'll just have to persuade them! :-)
Cheers,
Pat.
--
·=============================================================================·
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. None of you exists, my sysop types all this in!
Patrick Herborn
In message <tip-122-19961...@fluff.demon.co.uk> Ben Dooks wrote:
> I tried this, and found that soldering those FPGA style devices is both
> hard and fiddly... and I gave up in the end.
Wimp! :-) Try a very fine iron, use low melting point solder (it has some
silver in it to improve flow)...
Cheers,
Pat.
--
·=============================================================================·
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. How's this for diplomacy? Shoot them all! --Kirk
Patrick Herborn
In message <n246...@ozemail.com.au> Mal McClenaghan wrote:
> p...@nanobyte.demon.co.uk (Patrick Herborn) wrote:
>
> > Oh yes. Most definitely. I thought it was absolutely
> > REDICULOUS that I had to join the Clan to get a
> > RPC base without a monitor. Not that I minded
> > joining... I had already joined. I asked for a
> > base unit without an HD (I had a big HD). No go,
> > not even as a Clan member; my dealer was
> > VERY good, and bought the standard drive back off
> > me :-) Why this nonsense, though ?
>
> I didn't realise that this was still an issue.
> Oz customers can buy RPCs and 7000s without monitors,
> from any dealer, and most will be happy to spec the
> machine as you want it.
Ah, well a lot of thing have changed since then.... :-)
> Mal
Cheers,
Pat.
--
·=============================================================================·
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. Just another inmate in this ASYLUM!!!
Patrick Herborn
In message <19961028....@quantumsoft.co.uk> Stuart Halliday wrote:
> Now I hear rumours sometime ago that Acorn could do this with three VIDC20
> is it really possible?
Yes. They have even been looking at the possibility....
Cheers,
Pat.
--
·=============================================================================·
| Patrick Herborn, | p...@nanobyte.demon.co.uk | Primary Mail Address |
| Electronic Engineer. | p...@sn2.ee.umist.ac.uk | Alternative Address |
'============================================================================='
Immensely inspiring tagline follows....
.. SSC : Spare any Small Change?
Dave Stratford
Another idea for you power freaks!
ATM memory is still measured in bytes, so we currently have a theoretical
max of 256 megabytes of ram. The ARM however is a 32 bit processor, so why
are we still working with bytes?
The next (but 1?) board should be working with 32 bit words, so without any
(serious) change to the os, we get 256 megawords max ram or 1 gigabyte max
ram.
Most s/w wouldn't care how big the individual 'units' are, so changing from
8 bits to 32 bits should be no problem for most things.
Dave
--
Dave Stratford ZFCA
da...@argonet.co.uk http://www.argonet.co.uk/users/daves
... The Cursed Cursor curses and having cursed...
Darren Salt
prism.prismtech.co.uk> <326F503B...@nortel.co.uk> <54o7su$t...@prism.prismtech.co.uk> <na.4d43564722...@argonet.co.uk>
Lines: 18
X-Newsreader: Messenger v0.27 for RISC OS
X-Posting-Agent: RISC OS Newsbase 0.56
In article <na.4d43564722...@argonet.co.uk>
Stuart Bell <sab...@argonet.co.uk> wrote:
[snip]
> I'd be quite happy with 5242800 bytes of video RAM. :-)
<pedant>
5242880 bytes.
</pendant>
;-)
--
+---------------------------------------------------------------------------+
| Darren Salt - arc...@spuddy.mew.co.uk - darre...@unn.ac.uk - ToonArmy |
| Acorn A3010 - Spectrum +3 - BBC Master - Wintel free zone - Season ticket |
+-01268-515441-free-email-&-Usenet---Retrocomputing:-a-PC-card-in-a-Risc-PC-+
Counting in octal is like counting in decimal, if you don't use your thumbs.