Re: Seeing VERSIONINFO under Vista?

See below...
On Mon, 28 May 2007 12:49:55 +0100, Daniel James <wastebasket@xxxxxxxxxxxxxxxx> wrote:

In article news:<b6dj535hhse5iah6aacvg5nb7qk8akg8jf@xxxxxxx>, Joseph M.
Newcomer wrote:
Remember, the PC was a bootleg project within IBM. It had very limited
budget, not the billion dollars or so it takes to create a new fab line!

That proves my point -- IBM had no concept of how successful the PC would be,
*****
No, that proves nothing. They could not predicate a billion dollar investment in a
problematic market that they did not know existed.
*****
nor of how it would shape the development of personal computing for at least
the next couple of decades. They also had a mainframe business to protect, and
no desire to impact upon the marketplace for their mainframe computers.
*****
That was not a relevant consideration. The PC was seen as a "niche product" that would
NOT be a serious threat to the main business. The fact that it became so was totally
unexpected. You are asking a company to have made different judgments in retrospect, and
that is not possible.
****
To them, the PC was just something to offer to the mainframe customers who kept
asking for personal desktop computers for office tasks, and who might
otherwise have gone to another manufacturer.
****
But that's what happened anyway. They saw the PC as something for secretaries to do
letters on, at best. They did not envision killer apps. The machine was FAR too weak to
be a serious product! And that would have been true no matter WHAT chip they had chosen.
Remember, we are looking at chips that had maximum clock speeds of around 2MHz, but the
PC, using a 4.77MHz chip, actually EXCEEDED almost anyother "personal computers" of its
day. Remember that the Mac, with its 8MHz chip, was still several years in the future.
The Apple II used a 6502 chip, which I think was about a 1MHz 8-bit machine.
****

Of course they didn't want to spend the money to produce a /good/ PC, it
wasn't in their own immediate business interests -- but that's why I do blame
IBM, which is what we were talking about. Had IBM wanted to do the job well
they could have done so.
*****
You are well into the "conspiracy nut" domain here. You, and a lot of other people, seem
to think that IBM deliberatly crippled the PC. This is simply not true, and I am tired of
this kind of paranoid ranting. There is no basis for it, other than some kind of
delusional system, and you really need to get in touch with reality. They wanted to do
the job well, and they actually DID do the job well. Just because you don't like their
choice of CPU, based on some bizarre notion that there was some miracle CPU already in
existence and which there was a vendor willing to meet their needs of volume shipment,
does not change the actual facts: there was virtually no choice. They took the best
machine that fit the available parameters, which included price and availability.
Availability dominated. Had the PC actually taken off and used a 68000, Motorola would
have been unable to meet the commitments (and had said so!) The fact was that IBM was
overly conservative in their estimate of the number of chips they would need, and Motorola
was unable to meet their original estimates. When it really took off, Intel had both the
capacity to meet the needs AND was willing to create more capability.

Next thing you know, you'll be telling us that the Evil Petroleum Companies have
suppressed the carbuerator that lets cars run on water. (This ignores certain realities
of the energy density of water...but not to worry, fuel cells are coming. They don't
produce energy from water, but they do produce energy from hydrogen and oxygen with water
as byproduct. But I digress)
*****

In the event, they did the job well enough to threaten their own mainframe
business anyway, but not well enough to allow them to profit so much from PCs
that the loss of mainframe business didn't matter, and not well enough to
spare us the pain of intel's crappy chips.
****
That was not an effect they expected. In fact, it nearly destroyed the company, because
they tried to maintain the mainframe business long after it was fundamentally dead. The
result was that in 1990 they reported a $5,000,000,000 loss, the largest loss suffered by
any American company up to that time (it was overshadowed a few months later by General
Motors posting a $7,500,000,000 loss, but the difference was that the GM loss was actually
a paper loss caused by a change in tax law that required writing off certain investments
that year, instead of depreciating them. The IBM loss, however, was a REAL loss of REAL
dollars. The result was that the $1,000,000 research contract that had been paying me
that year was canceled, and I became a consultant again, and have remained so ever since).

What, by the way, is "crappy" about the chip? Right now, Intel produces the fastest
consumer chip on the planet. And I don't mean raw clock speed, either. Tricks like
speculative execution, dynamic register renaming, pipelining, superscalar instruction
dispatch, an FPU that can do a 32x32 floating multiply in 1 clock cycle (that's about 360
picoseconds for a 2.8GHz machine), TLBs, L1/L2 caches, vector instruction sets, etc. Some
of these features did not exist in mainfraimes, some were only available on
multi-tens-of-millions-of-dollars mainframes. Now I can buy a chip with these features
for pocket change. Yeah, segment registers were a pain, but since they've been gone for
nearly 20 years, I think this is largely an irrelevant consideration. Or would you like a
/360 architecture, which makes the Intel architecture look downright sophisticated? (I
programmed in assembler in the /360 for many years, and I'd rather program a bare 8088).
Were you aware that the Macintosh required careful management of its Data Segment Register
to write software (oh, you didn't know about the Mac's DS register? Just because they
called it R5 didn't change the problems of managing it. I also programmed Macs).
****

I repeat: I do blame IBM.

The decision to build the PC with off-the-shelf parts was a given. Any
major R&D effort would have resulted in an up-front investment that could
not have been recovered unless the computer was sold at noncompetitive
prices.

I don't accept that. Look how much money the PC has made IBM, over the years,
****
You really, really don't get it, do you? When the PC project was conceived, NOBODY, not
in IBM, not outside IBM, had the FOGGIEST IDEA this was going to happen. OK, I've got
this deal for you. I've got a new idea, one that is really cool. To buy into it, you
need to invest a billion dollars. No, there's only a few little companies, most running
out of people's garages, that are doing something like this right now, and only a few
specialists are doing it, but hey, once our idea catches on and they see our product, we
are going to make BILLIONS AND BILLIONS of dollars. All I need, upfront, is a measly
billion dollars. Come on, look at how important this is going to be.

Can you spell "dot-com"?

NO sane company is going to make that kind of investment in what is a small business
sideline designed to support secretaries.
*****
despite being an over-engineered, under-designed piece of rubbish build around
a chip designed for a completely different purpose
****
Exactly what purpose was that? The 808x chips were general-purpose computers!
*****
and marketed by a company
who had a major financial interest in its NOT being successful.
*****
Really? Intel wanted the 808x line to fail? Gee, you could have fooled me. My
impression was that Intel had a commitment to making its microprocessor line succeed. One
of the reasons they jumpted at the chance to meet IBM's delivery requirements was that
this would generate MASSIVE revenue. That sounds pretty far from a desire to see a
product fail. Or is this more conspiracy-theory ranting?
*****
If IBM had
designed their own CPU from scratch and built a new factory to produce they'd
have had their investment back in a few years.
*****
But they didn't know that! NOBODY knew that! See my above proposal--all you need to do
to get fabulously wealthy is invest a billion dollars in my idea. They saw themselves
competing with the Altair and Apple II. The Mac and Lisa were not out yet. They knew how
many Apple IIs and similar machines had been sold. It was a SMALL market. You don't make
a billion-dollar investment to enter a market that small.

Besides, if IBM had designed the CPU, it probably would have looked like the /360, since
that would have been the right business decision for them. Have you ever programmed a
/360? You want "crappy design", it fits the bill perfectly! The maximum addressible
offset without a segment register change was 4095, not 65535. The fact that any of the
general-purpose registers could be used as a segment register was moot; by convention, R12
was the CS register, R13 was the SS register, the architecture had no stack, and calls
were done by BALR 14,function, which stored the return address in R14. If you wanted to
push parameters or the return address, good luck, because R13 pointed only to the current
stack frame, not the actual stack (actually, it was more like the EBP register, now that I
stop to think about it). I'm sure they knew to the penny the cost of designing a
360-on-a-chip, and it made no sense for a market as tiny as the personal computer market
in 1979.

They did not understand how much money was at stake ... that's their fault,
and they deserve the blame.
*****
Oh, and I suppose this was the result of their defective crystal ball? What part of
history have you failed to understand? You think that because the PC took off and changed
the world that this should have been anticipated. I hope you have lots of money, because
you've just labeled yourself "sucker!" to a huge number of potential investors. All they
need to do is tell you they have a lock on the future, and you'll give them whatever they
need.
******

I suspect you've never actually participated in a business in which
hardware fabrication played an important role and time-to-market and
competitive pricing were relevant considerations.

I have, actually, but that's not important, here. I don't believe
time-to-market was considered important. What was important was to be seen to
be doing something so that their mainframe customers wouldn't but a shedload
of PCs from their competitors. I think those customers would have been happy
to wait -- these were the days when nobody ever did get sacked for buying IBM.
*****
Actually, it was one of the critical parameters. The original charter to the PC group was
to bring out a computer in a small number of years...I forget if it was 2 or 3.

There were no competitors. There was just the Apple II and a bunch of wannabes.

IBM also did not understand the potential impact on their typewriter business. But that's
a different story.

By 1979, Digital Equipment Corporation (DEC) was the powerhouse. Minicomputers were the
rage. The VAX was not yet out when IBM was talking to Gates (1980), and that was a bigger
threat to IBM than Apple. Essentially, the plethora of S-100 boxes was the "competition",
and you don't invest a billion dollars in a marketplace that sells a few thousand units a
year. The PDP-11 was the best-selling computer available, and the LSI-11 (PDP-11/03) was
the "personal computer" equivalent. It could run Unix, and troff, but it was, in the
marketplace, not a wild success for that purpose, besides being rather expensive per
compute cycle. It had a maximum address space of 64K, but with separate I- and D- spaces
you could actually have a program that used 64K bytes of instructions and still have
another 64K available for your stack and data (oh, like, wow!) so the Intel architecture,
which allows 1MB of combined instruction and data, was a SIGNIFICANT improvement over this
architecture.

So we had the 68000, the /360, the PDP-11, and what miracle architecture available? Oh, I
forgot, your solution was to invest a couple billion into entering a marketplace whose
total size in that time was a few million at best. And do this in a timeframe that
allowed a product to come out in 2-3 years.

But that's OK. The ability to see perfectly into the future is what you are assuming, so
OF COURSE, knowing the next 20 years of the industry with perfect clarity means that all
1979 decisions you disagree with were made with malice.
*****


The IBM PC was never competitively priced. The original PC cost two or three
times the price of a good desktop CP/M system but had just 16k of RAM, ROM
BASIC, and a cassette deck (no floppies, no hard drive).
****
The difference is that it was a "turnkey" system. I had a 6502 system I built, and a Z80
S-100 system that would have run CP/M if I could have bought a floppy controller for an 8"
floppy, which I wasn't able to.
*****

The "in the past" probably refers to the PowerPC chip, which Motorola
manufactured under license from IBM. This agreement was reached in the
mid-1990s ...

I was thinking more of the System370 on a chip, which was essentially a
tailor-made 68000 with a different instruction set ... that predates the PPC
by some years.
****
See my above comments on the /360 (and /370) architectures. Obviously you never
programmed these, or you would know they really were crap. It was NOT a 68000, and there
is absolutely no resemblance between the two architectures. The 68000 actually allowed
full addresses in instructions; there was not a single instruction of the /360 series that
allowed a full address in the instruction. None. Not one. EVERY operand required a
segment address register to access it. So you would load up one segment register with the
source, one with the target, and execute an instruction (e.g., MVC, Move Characters). And
it could only move 255 bytes, and this was a compile time limit unless you used the
Execute instruction, which would then OR in the value in a register (in the range 1..255)
with the length field of the instruction. Oh yes, this was a simply GREAT architecture,
MUCH better than the 68000, and INFINTELY better than the 808x family. Or are you
thinking of the MVCL instruction of the /370, which required ESI, EDI, and ECX registers
(at least it had a destination count to avoid buffer overruns). Oh, they weren't called
ESI, EDI, and ECX, but they were even stranger, since you named the source and target by
naming the first register of an even-odd pair for the source, length and target, length.
values. And one of the fields used the high-order 8 bits to specify the fill character,
so you could only work in 16MB of virtual memory. Oh wow, how COULD I have overlooked the
architectural superiority of this machine over the 808x family?

I wrote one of the first interactive languages on a /360 (LCC), wrote a standlone debugger
for it, was one of the few programmers outside IBM who knew how to write an LPSW
instruction that enabled virtual memory, and we wrote a small operating system for the
/360 as a class project. What's a mere OUT instruction compared to the joys of a CCW
(Channel Control Word) chain with dynamic looping?

And who could forget the joys of the indefinite interrupt? (The interrupt happened
somewhere in the preceding sequence of arithmetic instructions, but we're not going to
tell you where...)

And I've got a flock of winged pigs here, and all I need is an investor. We can make
millions because we won't need to use trucks or trains to transport the pigs to the
slaughterhouse; we just train them to fly there themselves. This will revolutionize the
bacon industry. The price of ham will drop to where it becomes the meat of choice of
everyone whose faith allows it. So just invest a few million dollars, and my knowledge of
the future of the pork industry, being perfect, guarantees that we will become wealthy.

Oh yes, I do need some indefinite number of years to design a genetically stable strain,
and get wings actually strong enough to fly, but time to market doesn't matter, and I'm
sure you won't mind investing a few million into my porcine genetics institute to design
the new pig from scratch.
*****

I didn't mean to suggest that IBM and Motorola had any cross-manufacturing
agreements in place at the time that the original PC was developed, just that
IBM and Motorola are companies with a fairly common corporate direction ...
****
Which was exactly WHAT? Motorola was a semiconductor manufacturer. They were not in the
mainframe business. They manufactured CPUs, memory chips, I/O chips, and a lot of other
chips. They did not sell mainframes, they had no competitive position with IBM.

Oh yes, they were both in business to make money. Is that "common direction" enough?
*****
so it was not inconceivable -- even then -- that they could have reached an
agreement had they had the will.
****
Perhaps. But whatever you THINK should have happened, the truth was that Motorola
DECLINED to deliver the quantity of chips required in the timeframe specified, and Intel
AGREED to do so, That was ultimately the deciding factor. The PC group really wanted to
use the 68K; it really was a much nicer chip. But it didn't matter, because they couldn't
get it.
******

It is not coincidental that the Z80 instruction set and the 8080
instruction set were assembler-compatible. It was a deliberate design
decision taken by Zilog.

Naturally.

The bitwise representation of the opcodes were different ...

Not true. The Z80 ran 8080 binaries. The assembler mnemonics and syntax were
different but the bitwise operations were identical. The Z80 had more
instructions and more registers -- in particular it had the IX and IY
registers which could be used similarly to the HL register (called M in 8080
assembler) to address memory but which also allowed a 1-byte immediate offset.

The 8086 architecture added many things to the 8080 architecture, including
the SI and DI registers, which filled the same role as the Z80's IX and IY,
which made it really easy to port Z80 code to the 8086 family.

I do not recall major issues on 8080/8086 translation, but the 8085 was a
different beast.

From a software POV the 8085 was exactly the same as the 8080 except that it
provided additional hardware IRQs (three maskable and one non-maskable, as
opposed to one of each on the 8080). From an electronic and circuit-design POV
it was rather different (in particular -- and like the Z80 -- it ran from just
+5V supply and didn't require -5V or +12V).

http://en.wikipedia.org/wiki/Intel_8085

But this is not relevant, because IBM at that time labored under the
delusion that they would be writing all the application packages anyone
would ever need.

One of the stated reasons for choosing the intel chips was the ease with which
3rd-party software could become available. It was all part of IBM's failure to
understand the PC's potential that they saw it as just a box to tick on the
sales brochure (so that their mainframe customers could satisfy all their
needs from IBM, and not have to talk to competitors) and they had no wish to
have to get involved in the provision of software or support. They probably
thought that ROM basic was enough!
****
That appears to have been after-the-fact marketing hype. From the inside information I
got from an ex-IBMer, the real issue was the chip availability. Nothing else mattered.

Did you ever see IBMs software offerings? They had a payroll package, an inventory
package, an accounting package, etc. They were scaled-down mainframe software, every bit
as horrible to use as the mainframe software had been. Nobody bought their products
because they were so bad. Third-party vendors sprang up overnight to create software for
this new computer, and because it had "IBM" on it, it was a reputable machine. These were
usually written in BASIC or assembler, and the end user could have cared less about what
kind of chip was inside the machine. Computer architecture was irrelevant. What mattered
was that there was a payroll package, or a billing package, or an inventory package, and
THAT's what sold the PC. Not the chip. Nobody cares about chips. In fact, one of the
things that became fairly obvious was that NOBODY HAD THE SLIGHTEST INTEREST IN CPU
ARCHITECTURE ISSUES! And probably never had. People bought cost-effective FORTRAN
engines, or cost-effective COBOL engines in the 1960s. A few geeks pretended that CPU
architecture mattered, but what I had learned by the mid-1970s was that there was only one
thing that mattered: price/performance. Pretending a chip architecture (or any other
implementaiton of an architecture) matters was silly, and still is. It only matters
insofar as it allows compilers to achieve certain price/performance goals. Everything
else is irrelevant.

This is one reason the issue about the Intel architecture never bothered me. I knew it
didn't matter. I'd programmed the 1620, 1401, 1440, Honeywell 200, /360, /370, PDP-10,
PDP-11, PDP-20, VAX, 6502, 6809, Z80, 8080, and 68000 by the time the IBM PC came out,
read about the G20, 7090/7094, several Burroughs machines (including the 5500/6700
series), and the Stretch, and, shrug, the Intel chip was Just Another Computer
Architecture. Sometimes a bit of a pain to use, but what architecture hadn't been?
*****

This was not actually a consideration. The key problem was getting enough
chips available to meet the need, and Motorola simply refused to consider
that level of delivery.

I blame IBM! If they had correctly predicted the number of PCs that they would
sell Motorola might have been stunned into activity.
*****
Sure. And who, exactly, was the genius who predicted the personal computer revolution?
Even Bill Gates took a few years to figure out that the goal of Microsoft was to get a
computer to every human being in the world.

You keep saying "If we had perfect knowledge of the future, we would do things
differently". I presume that you just participate in this for entertainment, because with
your perfect knowledge of the future, you bought Microsoft sfock when it was cheap and you
are now a multimillionaire. (If you didn't, who do you blame for failing to see the
future?)
joe
*****

Cheers,
Daniel.



Joseph M. Newcomer [MVP]
email: newcomer@xxxxxxxxxxxx
Web: http://www.flounder.com
MVP Tips: http://www.flounder.com/mvp_tips.htm
.

Relevant Pages

  • Re: Seeing VERSIONINFO under Vista?
    ... I'm just explaining why I do blame IBM. ... There was no need to do a massive market study for a market as small as the ... It was going from mainframes to workstations. ... architecture could come within a couple orders of magnitude of mainframe ...
    (microsoft.public.vc.mfc)
  • Re: Patents, Copyrights, Profits, Flex and Hercules
    ... fore *new* entrants (I don't believe VCs are hot on the mainframe business, ... Maybe IBM is not the only entity that has limited the access to ... PWD allows for "Developing Products" in addition ...
    (bit.listserv.ibm-main)
  • Re: Seeing VERSIONINFO under Vista?
    ... that would NOT be a serious threat to the main business. ... I'm just explaining why I do blame IBM. ... going to have on the mainframe business ... ... Only hardware geeks care about architecture. ...
    (microsoft.public.vc.mfc)
  • Re: Why data serving on a mainframe
    ... Why Data Serving on a Mainframe ... A mainframe as overlay within the Cambodian government where all the ... share the data would provide a departure from business as usual. ... The IBM mainframe may not be available everywhere. ...
    (soc.culture.cambodia)
  • Re: Press Release: IBM Academic Initiative Boosts Mainframe Community, Helps Plug Projected Skills G
    ... > IBM Academic Initiative Boosts Mainframe Community, ... > eServer zSeries mainframe through its Academic Initiative. ... and training for students and faculty. ...
    (bit.listserv.ibm-main)