From SMACKINLA@cc.curtin.edu.au Sun Jun 27 10:34:39 1993
Return-Path: <SMACKINLA@cc.curtin.edu.au>
Received: from gossip.pyramid.com by dutecai.et.tudelft.nl (4.1/1.34JP)
          id AA21138; Sun, 27 Jun 93 10:34:35 +0200
Received: from sword.eng.pyramid.com 
	by gossip.pyramid.com (5.61/OSx5.1a Pyramid-Internet-Gateway)
	id AA05347; Sun, 27 Jun 93 01:34:23 -0700
Received: from goss.pyramid.com
	by sword.eng.pyramid.com (5.61/Pyramid_Internal_Configuration)
	id AA04719; Sun, 27 Jun 93 01:33:01 -0700
Received: from cc.curtin.edu.au 
	by gossip.pyramid.com (5.61/OSx5.1a Pyramid-Internet-Gateway)
	id AA05332; Sun, 27 Jun 93 01:32:55 -0700
Received: from cc.curtin.edu.au by cc.curtin.edu.au (PMDF #3256 ) id
 <01GZVM88DN3CFUMJ82@cc.curtin.edu.au>; Sun, 27 Jun 1993 16:32:40 +0800
Date: 27 Jun 1993 16:32:40 +0800
From: Pat Mackinlay <SMACKINLA@cc.curtin.edu.au>
Subject: More ramblings...
To: riscy@pyramid.com
Message-Id: <01GZVM88DN3EFUMJ82@cc.curtin.edu.au>
X-Envelope-To: riscy@pyramid.com
X-Vms-To: R3KPC
X-Vms-Cc: SMACKINLA
Mime-Version: 1.0
Content-Transfer-Encoding: 7BIT
Status: RO
X-Status: 


Ok, there's been lots of talking and I think we're getting closer to what 
is going to be practical for the final system. Although this isn't in any 
way final, what I think looks best at the moment is the following.

* IDT R3051/52 CPU at 40MHz: I think we're pretty much hooked on the CPU 
_type_ now, there are good reasons for using this one, and none of the 
other potential devices are really cheap enough when you include the cost 
of support chips. Obviously, we could use the '81 if the whole board cost 
ends up ok.

* IDT 3730 I/O controller: I like the sound of this chip, even if we're not 
going to make use of it's fancy "video DMA" hardware. Perhaps this feature 
could be used to drive a DAC?

* Dual or quad UART: gives us important motherboard-based serial ports for 
mouse/terminal/modem.

* Keyboard controller: I haven't actually been able to locate much info on 
these even after looking quite hard. The chip used in the PC is a simple 
8051 while the AT used an i8041/2 microcontroller. Could someone in the US 
give AMI a call about these? I know they manufacture them, but have no data 
or pricing info on them. I suspect they'd be pretty easy to interface to 
the 3730 I/O bus.

* Video controller: I'm still for a TMS34010-based solution. Here, the 
problem is mainly software. I think using this chip would help to cut total 
system costs quite remarkably. Steve's probably going to flame me about the 
NSC chip, but I think I prefer this one.

* Ethernet controller: Providing it's cheap enough, it looks like the NSC 
8390 and it's interface chip might be the right way to go here. If the 
devices are expensive, I'm in favour of leaving this section of the board 
unpopulated.

* SCSI controller: I think we've pretty much settled on the 53c94? Again, I 
don't have pricing info, but I think it's reasonable. This chip is pretty 
vital to the whole system, and I personally think the 53c94 is the best 
chip for the system.

* Boot EPROM: This is pretty straightforward. How much are we shooting for? 
I'd suggest 64k or 128k - I don't think we can really use much more than 
that. I'd like to have a complete system configuration/testing program in 
this along with the code necessary to read a boot image from SCSI, Ethernet 
and serial interfaces. I am willing to spend quite a lot of time on this to 
make it right <grin>.

* RTC + NVRAM: Don't know much about these. Can someone call Dallas Semi. 
and find out what they have available? I guess we're looking for about 1 or 
2k of NVRAM and some reasonable RTC.

* I/O bus: I have personally given up on the idea of an ISA bus. No one 
seems to have come up with an easy and cheap solution to implement one. I 
do like Steve's simple/proprietry 8 bit I/O bus, but I'd prefer not to have 
to have the 3041. I think this interface could be stuck onto the 3730 
relatively easily, and think it would serve all our needs as far as simple 
expansion goes. I really can't think of an ISA-based I/O device that anyone 
would want to add to this system that couldn't also be done with the 
proprietry interface. Again, I'd vote for leaving the I/O connectors 
"unpopulated" if they are a significant cost.

Does this sum up the current thinking ok? I'll listen to arguments about 
the video system, but I think the 34010 is really the best option at this 
stage, having the necessary attributes: low cost and low complexity. I 
think having our own 8 bit I/O bus is the best option, but I think it would 
be better to drive it with the 3730 than the 3041, reducing the complexity 
in the memory mapping and bus structure. The other parts are pretty much 
fixed, as far as I can see...

For the moment, the devices I can see being put on the I/O bus are: more 
serial ports, a DAC for sound, a parallel port and a floppy drive 
controller. I think that would pretty much cover everything, and having 
seen Steve's design before, I'd have to say it would be pretty easy to 
design and build the necessary cards. I really can't see the point of an 
ISA bus at this stage...

Comments?

Pat -- "There's only one thing left to do Mama, I got to ding a ding dang
	my dang a long ling long" (Jesus Built My Hotrod -- Ministry)
GCS d* -p+ c++ l++ m--- s+/- !g w- t- r

 
From a080700@hp750a.csc.cuhk.hk Sun Jun 27 16:10:40 1993
Return-Path: <a080700@hp750a.csc.cuhk.hk>
Received: from gossip.pyramid.com by dutecai.et.tudelft.nl (4.1/1.34JP)
          id AA21576; Sun, 27 Jun 93 16:10:29 +0200
Received: from sword.eng.pyramid.com 
	by gossip.pyramid.com (5.61/OSx5.1a Pyramid-Internet-Gateway)
	id AA16623; Sun, 27 Jun 93 07:10:16 -0700
Received: from goss.pyramid.com
	by sword.eng.pyramid.com (5.61/Pyramid_Internal_Configuration)
	id AA08036; Sun, 27 Jun 93 07:09:09 -0700
Received: from hp750a.csc.cuhk.hk 
	by gossip.pyramid.com (5.61/OSx5.1a Pyramid-Internet-Gateway)
	id AA16615; Sun, 27 Jun 93 07:08:55 -0700
Message-Id: <9306271408.AA16615@gossip.pyramid.com>
Received: by hp750a.csc.cuhk.hk
	(16.8/16.2) id AA07032; Sun, 27 Jun 93 22:07:01 +0800
From: Stephen Siu-ming Wong <a080700@hp750a.csc.cuhk.hk>
Subject: Re: More ramblings...
To: SMACKINLA@cc.curtin.edu.au (Pat Mackinlay)
Date: Sun, 27 Jun 93 22:07:00 HKT
Cc: riscy@pyramid.com
In-Reply-To: <01GZVM88DN3EFUMJ82@cc.curtin.edu.au>; from "Pat Mackinlay" at Jun 27, 93 4:32 pm
Mailer: Elm [revision: 70.30]
Status: RO
X-Status: 

Hi Risciers,

>* Dual or quad UART: gives us important motherboard-based serial ports for 
>mouse/terminal/modem.

I heard it once, and would like to bring it out again, will the Zilog SSC
be a better choice?  As I remember, it is a dual USART chip, and capable
to do sync and async.

>* Video controller: I'm still for a TMS34010-based solution. Here, the 
>problem is mainly software. I think using this chip would help to cut total 
>system costs quite remarkably. Steve's probably going to flame me about the 
>NSC chip, but I think I prefer this one.

Good enough, but still better to use VRAM as frame buffer for TMS and a
24-bit RAMDAC.

>* Ethernet controller: Providing it's cheap enough, it looks like the NSC 
>8390 and it's interface chip might be the right way to go here. If the 
>devices are expensive, I'm in favour of leaving this section of the board 
>unpopulated.

If we are not going to implement ISA bus, 8390 may NOT be the best choice,
some new LANCE chip should have a much better performance than 8390.
Moreover, 8390 is a rather dumb device, why not using a better chip?

One point I would like to raise, since we aim at PC board form factor,
does anyone think about how to put the ethernet connector on the mother
board and still accessible from a normal PC case?  There's a hole near
the rear right hand side for keyboard connector, but no other convenient
opening, how to deal with it, drill a hole on the case?

>* SCSI controller: I think we've pretty much settled on the 53c94? Again, I 
>don't have pricing info, but I think it's reasonable. This chip is pretty 
>vital to the whole system, and I personally think the 53c94 is the best 
>chip for the system.

Is it SCSI-2?

>* Boot EPROM: This is pretty straightforward. How much are we shooting for? 
>I'd suggest 64k or 128k - I don't think we can really use much more than 
>that. I'd like to have a complete system configuration/testing program in 
>this along with the code necessary to read a boot image from SCSI, Ethernet 
>and serial interfaces. I am willing to spend quite a lot of time on this to 
>make it right <grin>.

Do you mean boot from SCSI device (disk AND tape), Ethernet (bootp and tftp)
and some proprietry serial booting interface?

>* I/O bus: I have personally given up on the idea of an ISA bus. No one 
>seems to have come up with an easy and cheap solution to implement one. I 
>do like Steve's simple/proprietry 8 bit I/O bus, but I'd prefer not to have 
>to have the 3041. I think this interface could be stuck onto the 3730 
>relatively easily, and think it would serve all our needs as far as simple 
>expansion goes. I really can't think of an ISA-based I/O device that anyone 
>would want to add to this system that couldn't also be done with the 
>proprietry interface. Again, I'd vote for leaving the I/O connectors 
>"unpopulated" if they are a significant cost.

If no ISA bus, then we have to adjust our minimal device list on board,
such as floppy disk controller (much much cheaper than SCSI floppy),
bi-directional parallel port (almost a must), sound support (like 16-bit
stereo AD/DA, FM, PCM).

--
*><*><*><*><*><*><*><*><*><*><*><*><*><*><*><*><*><*><*><*><*
v Stephen Wong Siu-ming    v internet: stephenwong@cuhk.hk  v
^ Computer Services Center ^ bitnet  : a080700@cucsc.bitnet ^
* Chinese University of    * phone   : (852) 609-8904       *
v Hong Kong                v fax     : (852) 603-5001       v
*><*><*><*><*><*><*><*><*><*><*><*><*><*><*><*><*><*><*><*><*

 
From SMACKINLA@cc.curtin.edu.au Sun Jun 27 16:33:14 1993
Return-Path: <SMACKINLA@cc.curtin.edu.au>
Received: from gossip.pyramid.com by dutecai.et.tudelft.nl (4.1/1.34JP)
          id AA21635; Sun, 27 Jun 93 16:33:10 +0200
Received: from sword.eng.pyramid.com 
	by gossip.pyramid.com (5.61/OSx5.1a Pyramid-Internet-Gateway)
	id AA19738; Sun, 27 Jun 93 07:33:00 -0700
Received: from goss.pyramid.com
	by sword.eng.pyramid.com (5.61/Pyramid_Internal_Configuration)
	id AA09731; Sun, 27 Jun 93 07:32:52 -0700
Received: from cc.curtin.edu.au 
	by gossip.pyramid.com (5.61/OSx5.1a Pyramid-Internet-Gateway)
	id AA19692; Sun, 27 Jun 93 07:32:46 -0700
Received: from cc.curtin.edu.au by cc.curtin.edu.au (PMDF #3256 ) id
 <01GZW092D0ACFUMLVJ@cc.curtin.edu.au>; Sun, 27 Jun 1993 22:32:29 +0800
Date: 27 Jun 1993 22:32:29 +0800
From: Pat Mackinlay <SMACKINLA@cc.curtin.edu.au>
Subject: Re: More ramblings...
To: a080700@hp750a.csc.cuhk.hk
Cc: riscy@pyramid.com
Message-Id: <01GZW092D0AEFUMLVJ@cc.curtin.edu.au>
X-Envelope-To: riscy@pyramid.com
X-Vms-To: IN%"a080700@hp750a.csc.cuhk.hk"
X-Vms-Cc: R3KPC,SMACKINLA
Mime-Version: 1.0
Content-Transfer-Encoding: 7BIT
Status: RO
X-Status: 


>>* Video controller: I'm still for a TMS34010-based solution. Here, the 

>Good enough, but still better to use VRAM as frame buffer for TMS and a
>24-bit RAMDAC.

I was assuming this. I've since found out a bit more about the 34k. 
Basically, the chip has access to one memory bank. This bank may be either 
DRAM or VRAM, and is used to store both the chip's code and the video 
image. The host has access to all of this memory through an 8 or 16 bit 
"host access port" - so it can essentially "write through" the 34k and into 
the RAM on the other side. I think this would attach to the 3730 quite 
nicely. I'm assuming TI makes a neat little DAC/palette chip intended for 
use with the 34k, and if it's cheap enough, it'll be a neat solution.

I have some other good news: a mate of mine has C source to an assembler 
for the 34010, so we have _some_ software. I can't imagine writing an X 
server totally in assembler though <grin>.

>>* Ethernet controller: Providing it's cheap enough, it looks like the NSC 

>If we are not going to implement ISA bus, 8390 may NOT be the best choice,
>some new LANCE chip should have a much better performance than 8390.
>Moreover, 8390 is a rather dumb device, why not using a better chip?

Because we're more worried about price than performance (to a certain 
extent). One of the faster devices, the Intel thingo, is both quite 
expensive and SMT, making it unsuitable for our design. If you can give us 
some numbers on the new Lance, I'm sure it will be given due consideration.

>does anyone think about how to put the ethernet connector on the mother
>board and still accessible from a normal PC case?  There's a hole near

This is a good point. Perhaps on a little "extension cord" going up to a 
"slot filler" (those little metal strips on the back)?

>>* SCSI controller: I think we've pretty much settled on the 53c94? Again, I 

>Is it SCSI-2?

It's been said before, and I'll say it again - being "SCSI-2" is largely a 
_software_ issue, not a hardware one. Perhaps you mean "Can it do FAST 
SCSI?" If so, then the answer is no. There is a version of the chip that 
_can_, however, and I suggest this is the one to go for: the 53cf94.

>>* Boot EPROM: This is pretty straightforward. How much are we shooting for? 

>Do you mean boot from SCSI device (disk AND tape), Ethernet (bootp and tftp)
>and some proprietry serial booting interface?

Yes. The serial boot idea is mainly intended for use during very early 
kernel development (serial ports are usually the most trouble-free devices 
to get working reliably).

>>* I/O bus: I have personally given up on the idea of an ISA bus. No one 

>If no ISA bus, then we have to adjust our minimal device list on board,
>such as floppy disk controller (much much cheaper than SCSI floppy),
>bi-directional parallel port (almost a must), sound support (like 16-bit
>stereo AD/DA, FM, PCM).

You didn't read what I wrote. The idea is that we use the I/O bus for 
allowing this kind of expansion. There will still be an I/O expansion bus, 
the only difference is that it will be Steve's proprietry design rather 
than the ISA bus.

Pat -- "There's only one thing left to do Mama, I got to ding a ding dang
	my dang a long ling long" (Jesus Built My Hotrod -- Ministry)
GCS d* -p+ c++ l++ m--- s+/- !g w- t- r

 
