Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!csd4.milw.wisc.edu!uakari!ctrsol!ginosko!uunet!lupine!infopiz!athertn!hpda!hplabs!decwrl!sun-barr!cs.utexas.edu!ut-emx!walt.cc.utexas.edu!nghiem
From: nghiem (Alex Nghiem)
Newsgroups: comp.sys.ibm.pc
Subject: Re: 80386 80286 XT components: preliminary summary of responses
Keywords: 80386 Motherboards BIOS Phoenix Award AMI
Message-ID: <8907230435.AA04479@vondrake.cc.utexas.edu>
Date: 23 Jul 89 04:35:15 GMT
Sender: news@ut-emx.UUCP
Reply-To: nghiem@walt.cc.utexas.edu (Alex Nghiem)
Organization: The University of Texas at Austin, Austin, Texas
Lines: 94
To: axaris@cs.Buffalo.EDU
Cc: nghiem
Status: R

[inquiry for clarification of responses]

The WD controller is the WD1006VMM2--Two floppys (all form factors)
and two hd's with cache. This controller is designed for 
fast 286/386 machines.

I had no responses regarding Microscience. I do know that the
MTBF performance of Seagats drives in now advertised to
be 100,000 hours.

I had no comments regarding Intel manufactured mother boards. They
do make an Inboard 386 and an Inboard/PC 386. I have no use for
the Inboard 386, although many people are happy with them,
and the Inboard/PC 386 will not do for my application 
because of the non-standard eight bit data path (bottleneck).

The standard 386 Motherboard seems to be the Micronics 386. They have 
several models. They are rather pricy--from 1995 to 2095 depending
on the vendor. There are several vendors in Computer Shopper who
sell Micronics. 

As for the other manufacturers, Mylex have nice features, but do
not run Xenix. There are tons of other manufacturers of other
boards. Generally, the ones that sell ultra cheap have some kind of 
problems. I have since found that many boards with AMI bios and
shadow bios do not work properly with many 16 bit video and
hard disk controller cards with the shadow enabled. 
If you disable the shadow feature, the 384k taken up by the 
shadow feature is not free to be used for
VDISK.SYS--you simply lose access to it. Obviously, these boards 
do not fit the Lotus/Intel/Microsoft standard for extended memory.
So far, I have not found a cheap board that implements the disabled
shadow feature properly. Right now I am trying to verify from
the manufacturer of a particular board whether or not 
the shadow works correctly with the 16 bit bus hardware I use in the
first place. That way, I won't have to worry about whether or not
I will lose 384K.

Problems with BIOS:

I have received comments that the Phoenix 386 bios version 1.10
seems to work properly for most people with almost all software.
However, there are known networking and bios shadowing problems
with the Phoenix Bios that have been resolved with replacement 
with an Award BIOS.

Award BIOS version R2 does not run Windows/386. Apparently
Award R3 runs Windows nicely. Award and Quadtel bios (derived from
the Award) seems to run nicely with netware. However, the
Compuadd machine with the Award bios had mediocre test results.
Hauppage Computer Works, on the other hand, uses the Award BIOS
and are considered expensive, but very reliable.

Most of the 386's reviewed by PC Magazine and given Editor's Choice
had the AMI bios. PC Magazine just gave Editor's Choice to
an Everex 386SX machine whose bios is derived from the AMI bios.
Micronics uses the AMI bios. It seems to me that for a 386 machine, 
the AMI is the way to go. The problem of determining which low-cost
board will work properly still remains.

When looking for the board, it appears the Chips and Technology
NEAT (New Extended AT) chip set is the standard. If the board uses propriety
discrete logic, be wary of incompatibilites.
I have not determined if the problem of losing 384k when the
shadow ram is disabled is a function of the chip set or not. If
it proves to be a result of the chip set, then the bug will
be "standard" on lot's of clones.

Regarding the Seagate SCSI drives and controllers: 
There were some manufacturing problems with some Seagate SCSI HD40's
used by Apple, but I think they were resolved. Apple had a 
huge headache with the Quantum SCSI HD40's though. Any problem with
Seagates are small in comparison.

Many people like the  ST01/ST02 controller. 
They are cheap and fast. There were some problems mentioned with
slow transfer rates with the Seagate controllers, but I 
believe that they were resolved with the combination of a BIOS
upgrade and documentation on how to set jumpers on the
controller card for an AT (XT is the default). But, it has been reported 
that if the HD fails or is not recognized for any reason, the computer will
not boot from a floppy--it will hang until some one corrects the hard disk
problem or removes the hard disk controller card. 

I decided not to go with SCSI because of the lack of diagnostic software
for SCSI drives on the PC--most everything currently written is for MFM
ST-506 formats. I think the same deficiency holds true for EDSI and RLL
formats. Of course, this lack of diagnostic software probably will change
in the future.

Thanks for the feedback. I really appreciate it. Send more!

nghiem@walt.cc.utexas.edu
!cs.utexas.edu!ut-emx!walt!nghiem