\documentclass{article}

\begin{document}
\title{The intel Professional Workstation\\or the panther motherboard}
\author{Ard van Breemen\\{\normalsize\tt ard@murphy.nl}}

\maketitle
\begin{center}
\begin{verbatim}
$Id: pws.tex,v 1.9 1999/01/14 11:42:12 ard Exp $
\end{verbatim}
\end{center}

\tableofcontents

\section{Introduction}

Let see what intel has to say about it (This is their standard reply...)
\subsection{Intels Standard Reply}
Information on the Professional Workstation motherboard commonly referred to
as 'Panther' can be from our Faxback Service at  800-525-3019 or 503-264-6835
in the US or  +44(0)1793-496646 in the UK.
Press option 2 for
"components, boards,  platforms, and tools for OEMs and developers "
and follow the prompts.  Request a 'SYSTEMS' catalog.  From this, you can
reference documents and their associated FAXBACK document number. A BIOS
upgrade to version 1.00.07.S0 can be obtained at URL
ftp://ftp.intel.com/pub /bios

The ECU or EISA Configuration utility file '\$PWS2\_3.EXE' is obtained in the
following manner from the Intel BBS:

\begin{itemize}
\item Dial the Application Support BBS (503)264-7999 in the US or +44(0)1793-432955 in the UK. 
\item After Logon, press "2" for "Intel Application Support". 
\item Press until you come to the "Main Menu". 
\item Press "F" for "File Area". 
\item Choose the entry for the "EISA Configuration Utilities". 
\item Read the License Agreement. 
\item At the END of the License Agreement, there are instructions for creating the ECU diskette. Follow those instructions. Make sure your diskette label is "SYSTEMCFG". 
\item After reading the instructions, press "Y" to continue. 
\item Fill in the "User Information" template. 
\item Press "D" to DOWNLOAD a file. 
\item Type in the name of the file you want to download. 
\item Choose a transfer protocol. 
\item Download the file and logoff after download.
\end{itemize}

The Professional Workstation motherboard, commonly referred to as 'Panther'
will not support NT 3.51, NT 4.0, or Windows 95. It will support NT 3.0 and
Win 3.11. This product went to End of Life status prior to  the advent of these
Operating systems and subsequently no drivers exist nor will any drivers be
manufactured. The workaround is to disable the onboard SCSI controller and
install a 3rd party SCSI card or network interface card that supports these
newer operating systems.
 
\subsection{Well, we don't think so...}
The panther has been reported to work with native SCSI support on Windows 95, Windows NT, DOS and Win3.*(?). The ethernet drivers work reported up to and including Windows 95 (although driver installation is a little bit harsh), and linux.

So here is our faq (what I have assembled so far...)

The first Professional Workstation faq...
\subsection{My System}
First let me tell you more about my system (cstmel.nl.eu.org) : it is an
intel professional workstation (panther motherboard)
with 4x16MB 72 pin FP Parity simms,
an adaptec EISA 2742A twin SCSI controller connected to
an IBM 1GB hard disk,
a GIGA GIGATAPE DAT tape streamer,
some weird SCSI cd player without parity (on the second channel),
an AMD 486 running at 133 MHz
and an ISA NE2000 compatible ethernet card.
Of course all of this is running linux.

The first panther I ever saw was way back in 1990/1991 at an intel office in
the netherlands. At that time P5 and PCI were still "kept" a secret. The system
was supposed to be a example model of how computers should be in the future:
low hardware maintenance costs by a click and replace model.
\section{hardware}
\subsection{VIDEO}
\subsubsection{overview}
The video controller consists of a WD90c31 with 1MB of RAM.
Older models contain a WD90c30 with 512KB of RAM expandable to 1MB.
The controller is supported by all windows,
include XFree 86 using the SVGA driver.
\subsection{SCSI}
\subsubsection{overview}
The pws is equiped with an NCR 53c700-66.
At that time this was a state of the art SCSI controller
\subsubsection{drivers:Win95}
\subsection{networking}
\subsubsection{overview}
The pws is equiped with an intel 82596. This is a very intelligent controller
which runs its own micro-code. Communication with the hostprocessor is done
through linked lists of commands and buffers in the hostprocessors memory.
A complete description of the 82596 is available from intel. Search for
a file called "29021806.pdf". The complete URL:
\par
\begin{center}
ftp://download.intel.com/design/network/datashts/29021806.pdf
\end{center}
\par
It is a complete description of the chip itself.
To use it for the pws some additions are needed regarding generation of
the PORT and CA signal, and the interrupt glue needed for a pc.
I/O map:
\begin{tabular}{|l|c|l|l||}
\hline
PORT  & SIZE & ACTION  & MEANING \\
\hline
0xCB0  & 2&WRITE& Lower 16 bits for PORT command \\
0xCB2  & 2&WRITE& Upper 16 bits for PORT command, and issue of PORT command \\
0xCB4  & 1&WRITE& Generation of CA signal \\
0xCB8  & 1&WRITE& Clear interrupt glue \\
\hline
\end{tabular}
All other communication is through memory!
\subsubsection{drivers:Win95}
These are available. At the moment you have to e-mail
markus.kniewel@reuters.com. He will mail them to you. In the future it probably
can be found on ftp.murphy.nl or www.murphy.nl, but I'm not sure how to
organize it yet. Especially regarding licenses.
\subsubsection{drivers:DOS/NOVELL}
There is at least an ipx object file driver available. This can only be used 
in conjunction with the wsgen disks.
Novell HAS an odi driver called lan596 or something like that. It does not
work!
\subsubsection{drivers:linux}
The current driver is considered stable. You can get it from my web page:
http://www.murphy.nl/~ard (primary mirror), or hot from the press:
http://cstmel.nl.eu.org/~ard (development site).
\subsection{Memory}
The required memory are 72 pin simms, 36 bits fast page. According to intel
16MB should work but is not tested (16MB did not exist yet). And I guess all
speeds will do, cause usually 72 pin simms are faster than required.
If all four banks are filled the memory access is much faster, probably due
to memory interleaving. This means a theoretical memory access speed up of
almost 4 times comparing a single simm and 4 simms...

Tested configurations:

\begin{tabular}{|l|l|l|}
\hline
Configs tested & totals & success & remarks \\
2x1Mx36 FP     & 8 MB   & yes     & original \\
2x1Mx32 EDO    & 8 MB   & no      & beeps   \\
4x4Mx36 FP     & 64 MB  & yes     & expensive :( \\
\end{tabular}
\subsection{I/O map}
\begin{tabular}{|l|l|l|}
\hline
I/O Address (Hex) & Function & I/O Device \\
\hline
0000-000F & DMA controller 1                                    & 82357   \\
0020-0021 & Interrupt controller 1                              & 82357   \\
0022-0023 & DRAM controller registers                           & 82359   \\
0026-0027 & CLASIC indexed registers (relocatable to 0C26-0C27) & CLASSIC \\
0040-0043 & Interval timer 1                                    & 82357 \\
0048-004B & Interval timer 2                                    & 82357 \\
0060,0064 & Keyboard/mouse controller                           & 8742 \\
0061      & NMI status register                                 & 82357 \\
0070      & NMI enable register (write only)                    & 82357 \\
0070-0071 & Real-time clock                                     & DS 1287 \\
0078      & CLASIC BIOS counter                                 & CLASIC \\
0080-008F & DMA low page register                               & 82357 \\
0092      & CLASIC control port A                               & CLASIC \\
00A0-00A1 & Interrupt controller 2                              & 82357 \\
00C0-00DF & DMA controller 2                                    & 82357 \\
00F0      & Numerics coprocessor                                & i486/i847SX \\
00F8-00FF &                                                     &        \\
01F0-01F7 & IDE interface ports                                 & IDE drive \\
02F8-02FF & Serial port 2                                       & WD 16C552 \\
0378-037F & Parallel port                                       & WD 16C552 \\
03B0-03DF & Onboard video controller                            & WD 90C30 \\
03F0-03F7 & Floppy disk controller                              & 82077AA \\
03F8-03FF & Serial port 1                                       & WD 16C552 \\
0400-040F & DMA controller 1 extended registers                 & 82357 \\
0461      & Extended NMI and Reset control register             & 82357 \\
0462      & Software NMI register (write only)                  & 82357 \\
0464-0465 & Last Bus Master granted (read only)                 & 82357 \\
0480-048F & DMA high page register                              & 82357 \\
04C2-04CF & DMA controller 2 extended registers                 & 82357 \\
04D0-04D1 & Interrupt edge/level control                        & 82357 \\
04D4-04D7 & DMA controller 2 extended registers                 & 82357 \\
04E0-04EF & DMA controller stop registers                       & 82357 \\
04F4-04FF &                                                     & \\
0C02      & CLASIC control port B                               & CLASIC \\
0C03      & CLASIC control port C                               & CLASIC \\
0C26-0C27 & CLASIC registers (optional, see 0026-0027)          & CLASIC \\
0C80-0C83 & System board ID bytes (read only)                   &   \\
0CB0-0CB7 & LAN control registers                               & 82596CA     \\
0CB8-0CBF & LAN glue registers                                  & GLUE    \\
0CC0-0CFF & SCSI control registers                              & NCR 53C700 \\
\hline
\end{tabular}

\section{links}
\subsection{mailinglists}
Sent a message to majordomo<at>murphy.nl with the following body: \\
\begin{center}
subscribe elp486
\end{center}
\par
For linux specific driver testing: sent a message to: ard<at>murphy.nl
\subsection{Web-Site}
More information can be found at: http://www.murphy.nl/~ard
\section{untranslated}
\begin{verbatim}
========================Collected-Remarks=======================================
I put this here, since I have no time writing a beautiful faq...
--------------------------------------------------------------------------------
My immediate interest is in getting the network going under Win95. I
already have the SCSI working under that OS, although there might be some
issues around that (There may only be support for only one CD drive at a
time, which may explain why the Nakamichi MJ5.16 CD changer only sees one
disk rather than 5. This may seem esoteric but some of these machine will
be used for a hands-on science roadshow and I need access to more than one
CD per machine!). The Win95 driver is the NCR/Acculogic SDMS Miniport
Driver built into Win95, whil;st the chip is the 32-bit NCR 53C700 @ 33Mhz.
There is also a network driver for the Intel 82595 motherboard chipset but
we have not got this to work yet, hence the downloads from the US. The chip
on the PWS is a 32-bit Intel 85696CA @ 33Mhz. The binaries from the US are
for DOS and WinNT for the 82595 and Win311 and WinNT for the  82596. These
I have not yet tried and I am hoping I can use the Win311 driver. I am
happilly using the Win95 Western Digital driver for the video. The onbosrd
chip is the WD WD90C30 with 512k 60ns DRAM as standard.
[ The i82596 is software compatible with the i82586 in 82586 compatability
mode. For dos that probably is satisfactory, ard]

Other machine we will want to run Linux. These machines happily run with
the Kingston Turbochip 133 which uses the AMD 586 chip. So does the
Evergreen 486 upgrade. so I will be also trying that since it is cheaper.
Using the Kingston chip a benchmark utility rated the resultant machine as
equivalent to a 486 running at 133Mhz. I'll try and run a more
sophisticated benchmark program on this when the machine returns from an
event in Ireland!

This machine requires parity memory which is now rarer and as a result more
expensive. The claim in documents is the machine can only support 32Mb ram
and its most efficvient when all four SIMM slots are full 1.e. 4 x 4 is
better than 2 x 8.  I am told, but have not checked, that you can actually
use 2 x 16, although Intel documents do not  support this claim. I will try
this if I can find the right memory chips. The ones I have all came with 2
x 4mb only.

The cards some of you may have found in the machine are I believe
proprietary boards from Reuters whhci ran an operating system known as QNX.
I found a set of disks for this taped inside the cover of one machine. Also
one disk came upo with the message 'Can't find the QNX operating system".
Since the disk was ap[parently empty, this must have come from the boot
sectors. I believe they those I have were used for routing Reuters messages
within Secker-Warburg. One board might even be for remote booting. I
believe they are no value to anyone else, unless you are always short of
jumpers. There are plenty on these board!

I have installed a Quantum 1 Gb IDE disk in one of these machines. To avoid
the 540 Mb IDE limit I used OnTrack manager whcih was supplied with the
disk. Everything is fine on this machine except the network!
--------------------------------------------------------------------------------
yes, Reuters still is installing the LP486E workstation for special purposes.
If QNX is running on the machines, the only cards fitted will be two network
(ethernet and/or TR) interfaces. Nothing magic to those cards, they just aren't
the brands you would usually choose. The product the "Panther" is used for is
called "Session Server", which connects the Reuters TCP/IP intranet with the
customer's network. Besides rudimentary security functionalities it does
address virtualisation to avoid address range conflicts.
--------------------------------------------------------------------------------
Files available from the intel Bulletin Board.
(Note:
If you have to fill out a form, fill out EVERY entry, relevant or not. It will
not accept a form if you do not 
Due to licensing reasons I cannot put these files (that I already have!) on an
ftp server. Intel is not willing to co-operate.
I have most of these files from the Bulletin Board:-

[ 1]  $10007S0.EXE  567k 10/05/94 PWS: 1.00 07.S0 Bios Update
[ 2]  $WD43DOS.EXE  919k 10/05/94 PWS:WDv4.3 video Driver & Utils for Dos
[ 3]  $WD43WIN.EXE  904k 10/05/94 PWS:WDv4.3 video Driver & utils for Wind
[ 4]  41VIDBUG.TXT   426 10/05/94 PWS,XPR,CLS S:WD V4.1 Video bug @10x7x16
[ 5]  EASE03.DOC     22k 10/05/94 PWS: Config. Inst. (Word 2.0 file)
[ 6]  LP486GEN.1      4k 10/05/94 PWS: SCSI connector sources, WORD 2.0
[ 7]  LP486GEN.2      5k 10/05/94 PWS: TurboCache Module Issues, Word 2.
[ 8]  LP486GEN.3      2k 10/05/94 PWS: Control-Alt-Delete anomoly WforW
[ 9]  LP486GEN.4      7k 10/05/94 PWS:   ITD Cache description, Word 2.0
[10]  OP_20469.TXT   802 10/05/94 PWS:Possible PS/2 Mouse/kybrd timing prb
[11]  OP_20512..TXT   593 10/05/94 PWS,Xpr,CLS S: WDV4.2 Video bug @10x7x16
[12]  OP_20591.TXT   638 10/05/94 PWS:No On-bd SCSI Support for Novell
[13]  OP_20611.TXT    2k 10/05/94 pws: Flash recovery fails with Dos 6.0
[14]  OP20651.TXT    607 10/05/94 PWS: Problem with Novell HELP Utility
[15]  OPUS323.TXT    18k 10/05/94 Notes(LP486E)
[16]  OPUS324.TXT     4k 10/05/94 DOS LAN Driver Inst./Config.  Notes(LP486
[17]  OPUS325.TXT     1k 10/05/94 SCSI Terminator Availability(LP486E)
[18]  OPUS326.TXT     5k 10/05/94 FLASH BIOS Upgrade Procedure(LP486E)
[19]  OPUS329.TXT     2k 10/05/94 Intel SIMMs fail POST memory size test(L 
[ 1]  OPUS331.TXT     1k 10/05/94 L) hangs system(LP486E,Xp,386)
[ 2]  OPUS333.TXT     1k 10/05/94 L) hang after WIN3.0(LP486E)
[ 3]  OPUS336.TXT     3k 10/05/94 LP486E LAN: Duplicate Ethernet Address(L
[ 4]  OPUS342.TXT     1k 10/05/94 ECU Version numbers(LP486E,Xpress)
[ 5]  OPUS343.TXT     3k 10/05/94 LP486E LAN:  Missing Ethernet Address(LP
[ 6]  OPUS344.TXT    769 10/05/94 2.2 Failures on Panther(LP486E)
[ 7]  OPUS345.TXT     1k 10/05/94 LP486E: Incorrect video scan frequency(L
[ 8]  OPUS349.TXT     2k 10/05/94 1.00 04.S0:  Incorrect BIOS for PWS(LP48
[ 9]  OPUS350.TXT     1k 10/05/94 HP ScanJet II scanner parity errors(LP48
[10]  OPUS351.TXT     1k 10/05/94 Incorrect ECU Information Screen(LP486E)
[11]  OPUS352.TXT     2k 10/05/94 IGC VM386 Keyboard failures(LP486E)
[12]  OPUS354.TXT     2k 10/05/94 CMOS may be corrupted by reset(LP486E)
[13]  OS213VID.ZIP  331k 10/05/94 pws:   OS/2 R1.3 Hi-res Video Drivers
[14]* PWSBCH10.ZIP   43k  6/25/95 pwsbch10.zip
[15]* PWSTPS20.ZIP   46k 10/05/94 PWS: Tech.Prod.Sum. R2.0 (Word 2.0
[16]* R14DOS.ZIP    970k 10/05/94 PWS: Dos Drivers V1.4
[17]* R14WIN.ZIP    1.1M 10/05/94 PWS: LP486 r1.4 Base software Kit Window
--------------------------------------------------------------------------------
- I/O Bus performance; Intel states that the enhanced option "will
improve the video performance by a factor of approx. 2". I didn't
measure any difference on a variety of benchmarks; why?

Usual benchmark tests only test memory bandwidth, not BIOS speed or I/O speed.
The improved video performance is probably only available when using a lot of
I/O, which is usual the case in a very optimised 640x480x16 video driver.

- Console redirection to COMx; What should be used for?
You can use a fixed frequency moniter, or no monitor at all, just a terminal,
to boot your machines. Enabling this option usually mean that you do not run
DOS or Win95. Although it could be used by dos by putting ctty COM1 or COM2 in
the autoexec.
--------------------------------------------------------------------------------
As first message I resume my experience if can be useful to someone.
- My LP486 is equipped with an IBM 0663L12 1 Mb SCSI drive; I managed to
have the onboard SCSI adapter to work with Win 95 with the NCRSDMS.MPD
driver; this driver works also with Win NT 4.0, replacing with it on
installation diskette the NCR53C700 driver. I tried this on a trial
version of Win NT 4.0, but for a short time as the machine was too slow.
- I upgraded the CPU from DX33 to DX2-66 only, as overdrive processors
were otrageously expensive, and I haven't been able to find (in Italy)
any adapter of voltage CPU socket. Suggestions are welcome!
- My LP486 was equipped with 512 K of video RAM and the WD90c30 graphics
controller; later versions of LP486 were equipped with the WD90c31, and
I wonder if they can reach more than 256 colors. I recommend upgrading
to 1 Mb of video RAM, as I found the 256K chips almost free (surplus)
and they improve the performance at 800x600x256 (tested with ZD Winbench
96 under WfW 3.11). Intel requires 60ns chips, but I found no difference
also using 70ns (easier to find).
- I added an "home made external" Hitachi CDR7730 CDROM bringing out the
power cable and the IDE cable. It convives peacefully with SCSI, it is
only necessary to let in BIOS Setup both Onboard IDE and Onboard SCSI
enabled. It is also possible to use the IDE controller of an audio
board, that I'm currently using as it seem to perform slightly better.
Onboard IDE can be disabled in this case.
- I confirm that the LP486 works faster when all the four SIMMS are
populated, as Intel claims; I obtained a further small improvement in
performance by using an old QFRESH utility to set the optimal refresh
rate for the memory.
--------------------------------------------------------------------------------
I can confirm that a single 16 meg parity SIMM works OK.
Confirmed memory configurations:
# | SIZE | BANK
--+------+-----
2 |  4MB | (2)
--+------+-----
1 | 16MB | (1)
--+------+-----
4 | 16MB | All
--------------------------------------------------------------------------------
This is to confirm that the Intel Professional Workstation will run Windows
95, which recognises uses the Western Digital video chipset and the
NCR/Acculogic SDMs Miniport Driver for the SCSI port. I have successfully
run an external Apple CD300 CDROM with this configuration. I will be trying
later the following rmmovables:  Iomega ZipPlus drive, Xyratex MCD540Mb
drive, Syquest 4Mb drive and probably an Archive Viper QIC tape drive. I'll
post the results. I have not yet tried the Win95 ethernet driver provided
by Markus Knievel. This is the crucial component for us in the Department,
but not for home use.
--------------------------------------------------------------------------------
The Kingston Turbochip 133 AMD 586 upgrade works fine. Since this is just 4
x 33MHz there is no need to change the jumpers. In any case these are
documented in the faxback literature and on the label that is inside the
cover, although not all machines seem to have thesJust remove the Intel
DX33 and plug in the Turbochip board. I'll be trying the Evergereen 586
upgrade next week.
================================================================================
From: Paolo Iachettini
Subject: LP486 Benchmarks

Hi everybody,

I report some benchmark (Ziff-Davis WinBench 96); if anybody is
interested in that, we can include in the table other results with
different configurations and/or benchmarks.

-----------------------------------------------------------------
WinBench 96	(DX33, 8Mb RAM, Win3.11 FW 32BFA, IMB 0663L12 HD,
800x600x256 colors)

Graphics Winmark	Disk Winmark	CPU Mark16	CPU Mark32

1.19			268		41.1		47.6

WinBench 96	(as above, CPU DX2-66)

1.61			290		59.7		64.6

WinBench 96	(as above, 16Mb RAM, RAM refresh rate set to 252)

1.72			365		64		70.3

WinBench 96	(as above, Win 95)

1.96			341		62.5		69.3		
-----------------------------------------------------------------

Paolo
================================================================================
From: Jörg Zöpnek
Newsgroups: cstmel.elp486
Subject: Links to drivers for older Intel Boards
Hello

I´ve tested most of the links but The House of Hall and YM430TX are only
links to the Intel Side but this links are very intersting
http://www.ioncomputers.com/text-only/Help.html
http://www.ioncomputers.com/text-only/Bios.html

On the help site you can find under the Point 1992 DeskTop Systems a
link called Professional. There you can find a very intersting help file
about the LP486E with informations about drivers and so on.
On the other side you can find a Bios update to Version 1.00.07.S0 for
the Lp486E under th same Point.
================================================================================
From: Jörg Zöpnek
<HTML>
Happy
<BR>Happy
<BR>I got the network adapter running under Novell DOS 7.0 with th lp486e.com
it works properly also I got the network adapter and the SCSI controller
running under WIN 95. I will make some Benchmark Tests in the next day&acute;s.
I will send you the results..

<P>Bye

<P>Joerg Z:</HTML>
================================================================================
From: Paolo Iachettini
Subject: LP486 Benchmarks

Hi everybody,

Two weeks ago at a local electronics fair I found an ominous "computer
scrapper" who had some surplus Panthers; they were equipped with DX33
CPU, 4 Megs of RAM, and a Maxtor 120 Mb drive; none of them had the
cache module. With some effort I bargained one of them (without case)
for 70 USD, as the motherboard was the last release with the 90c31 video
controller, tha is much faster than 90c30.

Meanwhile I replaced my IBM 0663L12 1 Gb SCSI drive with a new Samsung
WN32162U 2,1 Gb SCSI Drive; as its performance looked unsatisfactory, I
tried the "brother" IDE version WN321620A; the performance is similar
under DOS, Win311 and Win95, but the IDE requires Diskmanager for LBA
that I'm still waiting from Samsung. The odd thing about the SCSI
version is tha the drive is much faster on the second phisical half so,
provided that I created four partitions, the results are much better on
E: and F: than C: and D:. If somebody can explain me this ...

here are the benchmarks updated:

-----------------------------------------------------------------
WinBench 96     (DX33, 8Mb RAM, Win3.11 FW 32BFA, IMB 0663L12 HD,
800x600x256 colors)

Graphics Winmark        Disk Winmark    CPU Mark16      CPU Mark32

1.19                    268             41.1            47.6

WinBench 96     (as above, CPU DX2-66)

1.61                    290             59.7            64.6

WinBench 96     (as above, 16Mb RAM, RAM refresh rate set to 252)

1.72                    365             64              70.3



WinBench 96     (as above, Win 95)

1.96                    341             62.5            69.3            

WinBench 96	(as above, Win 95, Samsung SCSI, WD90c31

3.52			487		62.5		68.7	
================================================================================


\end{verbatim}
\end{document}

\section{changes}
\begin{verbatim}
$Log: pws.tex,v $
Revision 1.9  1999/01/14 11:42:12  ard
Added tested memory table

Revision 1.8  1999/01/13 14:11:27  ard
Primary mirror changed into murphy.
Local system has the name cstmel.nl.eu.org

Revision 1.7  1998/11/09 19:34:08  ard
Added some more version numbers

\end{verbatim}

