This is only a preview of the May 1993 issue of Silicon Chip. You can view 51 of the 96 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Items relevant to "A Nicad Cell Discharger":
Items relevant to "Build The Woofer Stopper":
Items relevant to "Remote Volume Control For Hifi Systems; Pt.1":
Articles in this series:
Articles in this series:
Items relevant to "Alphanumeric LCD Demonstration Board":
Articles in this series:
Articles in this series:
Articles in this series:
|
COMPUTER BITS
BY JOE ELKHORNE
Upgrading to a 386 - more on
kludging a computer
Last month, I had lots of fun getting the
hardware of my cobbled up computer to work.
But there was more fascinating frustration to
come when I realised how beautifully simple it
is to purchase a completely new computer.
The hardware is only half of a platform, of course. Having achieved a
working computer, I now had to pursue state-of-the-art software. Fortunately, I'd picked up Windows 3.1 at
PC92 at a significant cost reduction $95 is a lot better than the $225 I'd
the 3.0 version to get the CGA driver
and then install my 3.1 "over the top".
That evening, we did exactly that and
both discovered how clunky Windows
looks on a CGA monitor. Well, it
couldn't be helped.
The whole exercise was a process
"A quick skim gave me enough information
to be dangerous & I started the installation.
To my dismay, the first problem I encountered
was that a CGA driver is no longer provided."
seen at a discount house when I began this exercise! I also would have
bought MS-DOS 5 but they'd sold out.
The dealer, however, promised to let
me have one from the next shipment
at the show price. Good show!
I opened the white box containing
the 3.1 goodies and had a quick look
at the "Getting Started" booklet. A
quick skim gave me enough information to be dangerous and I started the
installation. To my dismay, the first
problem I encountered was that a CGA
driver is no longer provided. Blast! It
was pointless to proceed further and I
telephoned my friend Graeme.
His suggestion was that we "install"
78
SILICO N CHIP
of getting a machine up and running
as cheaply as possible. This included
using the existing inadequate 40Mb
hardcard and the Amstrad CGA monitor. Incidentally, somewhere along the
line, an expert told me that said monitor would never work "since they're
non-standard". Having h eard the
"Amstrad's aren't compatible" canard
before, I took that with a grain of salt.
Empirical tests proved that the monitor did function in a normal fashion attached to an ordinary CGA card.
One of the curiosities I glossed over
earlier was the hardcard installation
itself and how we simply ignored the
CMOS Setup. Well, since the compu-
ter was working, we continued to ignore the anomaly. Having dropped it
into a slot and found that it worked,
regardless of the CMOS Setup report,
there were more important matters to
attend to. Since normal disc operations (including defragging) worked,
the system obviously was happy to
recognise it.
Despite the limitations of the CGA
monitor, we completed the Windows
3 .1 installation. There were some frustrations - the system "fell over" several times, although I am certain that
this was due to my kludged hardware
rather than the software.
In the course of subsequent daily
operations , the system has clagged a
number of times when using file manager or even the MS-DOS Prompt. I've
learned not to attempt a file copy from
the A: drive to the B: drive, though
admittedly it sometimes works for
short files.
File size seems to be a factor in
whether or not the system will hang.
Marking a group of files is a recipe for
disaster, usually, but those same files
can often be transferred one by one.
Again, these problems must be located somewhere in my hardware or in
the system setup.
Incidentally, this is a good place to
point out that - if you have anomalous behaviour - don't make assumptions about the cause. Do not blame
Windows for something that might be
endemic to the system. The best test
when something falls over is to stay at
the DOS level and try to duplicate the
behaviour.
Drivers, drivers everywhere
Right - at this point, we have a
more or less working piece of hard-
ware. We've achieved disc operations
at the fundamental level, even if some
of the bells and whistles don't quite
work. Given the previous difficulties
with the floppy drives, this is hardly
surprising.
The monitor is definitely marginal
but is usable - just. The keyboard
does what it's supposed to, except
when it-doesn't. Well, what would
you expect from a definitely used keyboard? An occasional missed character is no big deal and the problem can
be solved by throwing a little more
money at it. More on this later.
Since the serial card has already
been tested with the modem, I know I
can continue to communicate with
the outside world. Ripper. The final
and perhaps most important peripheral operation is hard copy. I've been
using a colour printer for several years
and blithely assumed it would be supported by Windows 3.1. There's a
Microsoft Windows Version 3.1 Hardware Compatibility List included with
the software package. This covers compatible computers (my Kludge Special wasn't listed, of course), displays,
printers, networks, pointing devices
and keyboards.
Perusing the list, I then discover
the model I have is not included.
Bummer! Lots of Brand X printers,
just not .mine. Now, one of the beauties ofWindows is that you need install a printer once and the graphical
environment handles the talking between applications and hard copy
hardware. This is, however, contingent on having the right printer driver!
We did try several likely possibilities,
including a couple of Epson bogstandard types, without success.
Well, the X-42 (name changed to
protect the guilty) is a bit long in the
tooth but it is an 18-pin printer and
has pr,oduced some reasonable hard
copy from packages like Digital Research's GEM. I thought it would simply be a matter of getting the right
driver; either Microsoft or the printer
vendor was a possibility, I thought.
Accordingly, I called my local Brand
X dealer the next working day. They
heard of anyone else trying this. Let
me just check the master list on the
computer".
There was a long pause. It must be a
big list! As the silence became uncomfortable, the technical support
person (or whatever he was) said, "I'm
just unzipping the file now." Then he
added, idly, "Of course; three point
one only supports 24-pin printers, you
know. He then confirmed that the X42 wasn't on the list and suggested
that what I really needed was a new
"The monitor is marginal but is usable - just.
The keyboard does what it's supposed to,
except when it doesn't. Well, what would you
expect from a definitely used keyboard?"
referred me to the Sydney office, giving me a 008 number. I tried that.
The telephonist understood my initial query easily. She put me straight
through to the customer support officer. I explained the situation to him.
"Uh, Windows?" he mumbled.
"Yes."
"Three point one?"
"Yes."
"What model printer again?"
"X-42."
"One of ours?"
Well, I didn't get very far there.
Next, I decided to call the Microsoft
office in Sydney. A voice answered
and I quickly explained the situation.
"Umm, tricky. I don't think I've
printer. Well, I hadn't argued the point
but I knew after looking at the compatibility list that there were a number
of 9-pin printers. Maybe his curious
statement actually referred to Brand
X printers and not all printers. Who
knows? Who cares?
By now, I was thoroughly frustrated.
Just at that point in time , I got a call
from the software dealer. I raced down
to pick up DOS 5.0 and pessimistically took my compatibility list with
me. The sales rep was not surprised at
my tale of woe. I_ said I was about
desperate enough to buy a new printer.
He thought that a bad move, since I
was essentially satisfied with my existing hardware, if I could get it to
Protect your valuable issues
Silicon Chip Binders
These beautifully-made binders will protect your copies of SILICON
CHIP. They feature heavy-board covers & are made from a distinctive
2-tone green vinyl. They hold up to 14 issues & will look great on your
bookshelf.
* High quality with heavy board covers
* Each binder _holds up to 14 issues
* 80mm internal width
*
logo printed in gold-coloured lettering on spine & cover
SILICON CHIP
Price: $A11.95 plus $3 p&p each (NZ $6 p&p).
Just fill in &mail the order form in this issue; or fax (02) 979 6503; or ring (02) 979 5644 "e your credit card number.
MAY
1993
79
COMPUTER BITS - Upgrading to a 386
work. He offered further suggestions
to trace a source for an appropriate
printer driver. I was impressed with
this since it really meant no money
for them. Naught came of this, however, and in the end, I-went back to
the dealer and bought an inexpensive
Epson.
Will it work?
The rep said he used one himself. I
looked askance , knowing the bane of
compatibility problems and pointed
out that this particular model was not
on the Microsoft list. "Use such-andsuch," he said. "It works". Oh, before
I continue, of course I was still able to
get hard copy at the DOS level. And
even under Windows 3.1, the "generic" driver functioned properly. It
only meant the little advantages like
TrueType fonts and other graphical
bells & whistles weren't available.
"Ifl have any trouble, I'll be coming
back with fire in my eye," I warned.
The plastic money got bent a bit
further that arvo and I raced home.
Installing the driver under the Win-
out of the ordinary there. Without even
going into Windows, I did a straight
DO£ copy of an ASCII file to the
printer. It had errors.
Not dramatic ones, to be sure. It
took me a second to recognise what
was wrong: no upper case characters.
Just to double check, I did a hex dump
and repeated the same process on the
old Brand X printer. The conclusion
was inescapable: a stuck bit in the
interface electronics. Bah humbug, so
much for the self-test earlier. At least
I'd been able to eliminate the guesswork by a systematic approach.
I returned to the dealer with fire in
my eye. When I walked through the
doorway, several people dived for
cover. The wimps sent one of the
young women out to see me. I told her
that I had to see the sales rep and they
managed to drag him out of the back
room. I showed him the printouts with that inescapable evidence of
hardware malfunction, I received a
swap printer straight away.
Back home once again, I went
through the entire process from the
"I returned to the dealer with fire in my eye.
When I walked through the doorway, several
people dived for cover. The wimps sent one of
the young women out to see me."
dows operating system was not difficult. Foolishly, I didn't start at Square
One. Yes, I know better. Why should I
take my own advice? Instead, I went
to a Windows application, called up a
file I'd been working on and pointed
it at the printer. WYSIWYG? No way.
By now, I could recognise bit streams
being interpreted wrongly.
Was it the driver? Even the Epson
book verified what the sales rep had
said. Curses, foiled again! I turned off
the rig in disgust and watched Star
Trek and went to bed.
The following morning, I did a systematic diagnosis. I verified the selftest of the new printer. It was good. I
checked the software configuration of
the printer itself, just in case. Nothing
80
SILICON CHIP
ground up. And of course, there was
no problem at all. By the time I got
into Windows and produced my first
state-of-the-art hard copy, I felt like
I'd been through a war.
Yet another upgrade
Just at this point in time, "me good
ol' mate" decided he really needed a
high-resolution non-interlaced monitor. With appropriate video driver
card, of course. So I inherited one
SVGA monitor with Tseng Labs card.
It's a world of difference to see what
you're doing, rather than interpolating it from a CGA monitor whose dotpitch approximates S2 glasspaper.
Over the next month, I soared up
the learning curve of MS-DOS 5.0,
and the intricacies of Windows 3.1
and various applications. It was inevitable that missing keystrokes would
frustrate me to the point of buying a
new keyboard. I smirked to myself
that nothing could go wrong in this
regard. Sure.
After all, there 's only really a very
few possibilities: the original XT-style
(not mine) or the original AT style
(nor again), the less common 3270
type (definitely not mine), or the 101key or PS/2 or enhanced keyboard.
You definitely can't interchange XT
and AT types. Yes, your typical clone
keyboard usually has a switch with
XT I AT positions.
My existing keyboard did have a
switch, did have 101 keys and surely
any typical enhanced keyboard would
work. I discovered there was a
monthly computer swap meet on and
attended the venue. Several dealers
had attractive keyboards at even more
attractive prices. One word of advice,
here: do give a prospect a bit of a run
to see if you like the feel and sound!
There are wide variations.
I selected one and took it home.
Nothing can go wrong, I said ... off
with the old, on with the new. I fired
up the machine and knowing the
CMOS did have the keyboard test installed, waited through the 35-second boot-up procedure. No error report, good! I typed "WIN" and waited
whilst the operating system loaded.
When I first started to do something,
the system locked up.
Hey, what's this? I'd decided some
time back to coin a new acronym: "I
had to CAD out of Windows (again)".
CAD?-Control-Alt-Delete, of course.
Well, to my surprise, I couldn't even
CAD. A poke on the Reset button
solved that problem. I watched again
through boot-up. Then I stayed at the
DOS level. Tapping the keys got response. But when I hit either the Caps
Lock or, on a subsequent test, the
Scroll Lock, the appropriate LED came
on permanently and the keyboard was
locked up. Very, very strange.
The leaflet in the clone keyboard
box referred to "the switch" and
showed one of two possibilities, depending on whether it was "a type A
or type B" keyboard. The unit itself,
however, had no such switch; I'd already checked and assumed it was a
standard enhanced unit.
I called the dealer. He'd never had a
keyboard fault, naturally, but offered
to swap it. I drove a considerable distance to his location and watched in
dismay as he plugged "my" keyboard
into an ordinary 386 clone and it
worked perfectly. Since I was already
there, he gave me another keyboard as
a swap. Back home, it exhibited the
same symptoms!
At this point, I drove out to see my
favourite BBS sysop (hi Sandy !) and
she put this u-beaut new keyboard on
one of her many computers. It worked
just fine. I took a slightly used one
from her in exchange - it did have a
switch, set on the AT mode, of course.
And it works just fine on my system.
Why is 't his all so? I have no idea.
All I can say is that it points up the
necessity for mix and match testing.
There's nothing like substitution to
sort out these little problems. Swapping one part and repeating a test is
the quickest and easiest answer.
Conclusions
What's the bottom line? It's been an
interesting time to be sure , with lots
of unanswered questions. At the end
of it, I have a reasonably inexpensive
and mostly working platform. Did I
save any money? If time is money,
then the answer is unequivocally
"NO! "
Using 20-20 hindsight to confirm
things I generally knew anyway from
past experience, here's a little list:
(1). Change one thing at a time.
(2). When in doubt, substitute.
(3). Cheap add-ons without documentation may be no bargain.
(4). Be systematic and thorough.
(5). Don't blame one thing when another may be at fault.
(6). RTFM - when all else fails, Read
The Flamin' Manual.
The last is particularly important
when confronted with little, niggling
annoyances like too-easily accessible
mains select switches on power supplies. Probably more units have blown
up from inadvertently flicked switches
than all the component failures or
supply spikes put together!
And finally, keep smiling!
Editor's note: readers would do well
to regard this and last month's story
as a cautionary tale. Upgrading hardware in this piecemeal way can be
frustrating, even to the most well-informed computer user. If you want a
new computer and can afford the
price, sell your old one, buy the new
one and sleep well at night.
SC
Intel Announces The Pentium
Intel Corporation has announced
the production version of the
Pei;itium® processor. Up to five times
as powerful as the 33MHz 486 DX
CPU, the fifth-generation Pentium
processor extends performance
while maintaining full compatibility
with existing software.
The new processor will benefit
areas such as scientific modelling,
computer-aided design and engineering (CAD/CAE), large-scale financial analysis, and high-throughput client/server applications. The
Pentium processor also will provide
the increased performance necessary for applications such as voice
recognition , imaging and real-time
video.
The Pentium processor will be
offered in 66MHz and 60MHz versions. The speed difference between
the 66MHz and 60MHz versions is
about 10%.
Manufactured using a 0:8µm
BiCMOS process and designed using a superscalar RISC architecture, the Pentium processor has two
5-stage execution units and can
process up to two instructions in a
single clock cycle . Both the Intel
486 and Intel 386 have one execution unit. The Intel 386 is a traditional CISC design that utilises several clock cycles per instruction. By
contrast, the Intel 486 CPU, designed with a RISC integer core ,
executes most instructions in a single clock cycle.
The Pentium processor features
two 8Kb on-chip caches, dramatically improved floating point performance, and a 64-bit burst-mode
external bus. It has 3.1 million transistors - nearly three times as many
as the 486 .
The powerful, fully compatible
floating point unit (FPU) incorporates optimised algorithms and dedicated multiply, divide and add hardware, with an 8-stage pipeline to
execute one floating point operation
per clock cycle. The FPU is capable
of running many applications 5-10
times faster than the same applications running on a 33MHz 486DX.
Other design techniques, such as
branch prediction, 256-bit internal
data buses and write-back caches,
all serve to improve performance.
Not only will current software run
on Pentium processors without
modification and with substantial
performance improvement, but new
high-performance tools and compilers are available that will allow commercial and in-house developers to
achieve even greater performance
enhancements through a recompilation process, also known as optimisation.
Many major software developers
have committed to optimising their
current applications for the Intel architecture, while others are porting
their high-end applications to the
Intel architecture for the first time .
Peripherals
Intel is providing system building
blocks to enable a variety of Pentium
processor-based systems designed
for high-performance desktop and
server applications. These blocks
include the 82496 advanced cache
controller and 82491 cache, the
82489 DX interrupt controller, and
the 82430 PCI set chip set.
The Pentium processor and second-level cache chip set, the 82496
cache control ler and multiple 82491
custom SRAMs, are a tightly combined group of components
optimised for high-performance
desktop systems and two to eightprocessor high performance servers. The 82489 DX, the first implementation of the advanced programmable interrupt controller (APIC)
architecture, provides multiprocessor system support.
The Intel 82430 PCI set provides
PCI local bus performance to
Pentium processor-based desktop
systems. It includes an integrated
cache/DRAM controller, a local bus
accelerator, and system logic with
an EISA or ISA expansion bus bridge
to enable a range of price/performance systems.
Intel has begun production of the
new processor and expects to ship
approximately 10,000 units in the
second quarter of 1993.
MAY
1 993
81
|