• Please review our updated Terms and Rules here

Happy Birthday IBM PC!

hargle

Veteran Member
Joined
Nov 30, 2007
Messages
1,397
Location
minneapolis, MN
"The IBM Personal Computer, commonly known as the IBM PC, is the original version and progenitor of the IBM PC compatible hardware platform. It is IBM model number 5150, and was introduced on August 12, 1981." -wikipedia

So it is now old enough to run for president of the United States.
 
Sigh, 35 years already?

I recall my profound disappointment after seeing the announcement of the CS 9000 shortly before that of the 5150. Many people had hoped for a 68K-based machine--had that happened, I wonder where we'd be today. Would Apple have bothered bringing out the Macintosh? What would PC software look like?

The 5150 was really a big letdown. Slow with an 8-bit bus. Expensive for what it was.

It's interesting to speculate.
 
It is pretty funny that it was not designed with a 64K DRAM option when it was one of the first popular machines based on the 8088 processor. BTW, the 16K DRAM shortage seems to date back to 1979 and ended in 1980.
 
That was the other laughable thing. An 8088 machine with 64K of DRAM that you could purchase with only 16K installed. Way behind the times. I suspect it's the PC that saved the 8086 from obscurity.
 
That was the other laughable thing. An 8088 machine with 64K of DRAM that you could purchase with only 16K installed. Way behind the times. I suspect it's the PC that saved the 8086 from obscurity.

I don't mean 64K DRAM as default though, which may be too expensive back in 1981.
 
A 68000 CPU cost about $1000 in early 1981. 64k of RAM was about $500 at the same time. Going through resellers as IBM intended and adding floppy drive and case would make the minimum price of a 68k IBM PC about $4000. I believe Motorola didn't manage to make 250,000 68k chips in 1981 and support chips were still sampling in 1982. The 68000 could not meet IBM's rather restrained goals for a PC and would not until about 1984.
 
Considering that the Commodore 64 rolled out in 1982 at a price of $595? Nope.

I believe that was only officially released later in 1982 though.

Trivia: The first 64K DRAM chips by Fujitsu in 1978 (MB8164) had a +7.5V and a -2.5V power supply. Of course, by 1980 it was already replaced by the single +5V power supply version before it even reached beyond sampling volumes.
 
So, not even an 8086 and a 16-bit bus?

Remember that a 5150 with 64K, monochrome adapter and a single floppy drive set you back about $3K back in 1982. I think the CS9000 was about $6K back then, but then IBM dinged customers for every little thing--I'm surprised that they didn't charge for the shipping box.

In 1981, our engineers thought that a 68K was the next logical step after the 8-bitters. We even worked up our own prototype CPU card. Bill Davidow, who was on our board essentially said that it would be a cold day in hell before one of our computers used a competitor's CPU. We were forced into the 80186, which hadn't been released yet, but was sampling--bugs and all, and set our development back over 6 months trying to get something working. We had to make initial shipments without the 80286 because it wasn't yet ready for prime-time in 1983.

The 68K used a new 3-micron HMOS process that was amazing for the time. In a couple of years, 68K prices plummeted and so did 64Kb DRAMs. Had IBM waited a year and done it right, the world would look very different. The operation down in the Rat's Mouth struck me as a very amateurish endeavor.

FWIW, the NEC APC, introduced a couple of weeks before the 5150 offered much more bang for the buck.
 
Last edited:
Oh, I would have liked lots of differences in the IBM PC. The problem is I would have to find out what the status of parts production and second sourcing was in 1980. I know that was the problem with double-sided drives; they were scarce when IBM started signing contracts but quickly IBM was the major reason single-sided drives continued to be made. The IBM PC was unusually nimble for them but it still was an aircraft carrier in a field of helicopters.
 
I have always hoped IBM had introduced a PC which was machine code compatible with their mainframes. That would have allowed all the existing IBM assembler programmers (and all their years of knowledge) to do a seamless transfer to the PC. And I think a good part of that knowledge would have been to design far beyond today's technology.

For example, the IBM 370 class mainframes used 24 bit addressing. After all, the thinking at the time was who could possibly need more than 16mb of memory? It didn't take long before 16mb wasn't enough and the designers had to go back and increase the addressing to 31 bits (the 32nd bit was used to identify the addressing mode of either 24 or 31 bit). Yet despite this major change, all 24 bit programs run just fine on the 32 bit machines WITH NO CHANGES! I left the mainframe world before the change to 64 bit addressing but I am sure all the old programs still run just fine on the new hardware and OS.

As a result of their decision, they started from scratch and as the processors got more and more powerful they needed to Rube Goldberg the internals structure. Had they replicated the mainframe architecture, programs designed to run on the first release of the processor would still be able to be used on later processors...exactly as programs written on the mainframe in the 1970s can still run on today's mainframes.

Joe
 
Last edited:
Oh Joe, IBM couldn't do that across their range of same-series mainframes. For example, is the S/360 model 20 compatible with the model 30? Model 50 with the model 44?
 
Oh Joe, IBM couldn't do that across their range of same-series mainframes. For example, is the S/360 model 20 compatible with the model 30? Model 50 with the model 44?

Hi Chuck,
I don't know the answer to your question, although I'm guessing the answer is no.

I entered the workforce around 1980 (on a mainframe) and started to learn assembler programming on the S/370 mainframe maybe 2-3 years later. Despite a couple job moves over the years, the code I wrote at one job (whether assembler or a high level language) would always work at the next one without changes. Sometimes the mainframes were IBM, other times they were clones (I think Amdahl was one of them). Having never worked in the computer room itself, I was often only slightly aware of the exact model of the mainframe and there were many upgrades over the years. But it didn't matter to me as a programmer or database systems guy because they all worked the same.

I guess I always worked in companies where they used a descendant of the IBM 360/370 architecture and never worked on any other line of IBM hardware. It has just always seemed to me that the microprocessor chip wasn't intended to be the heart of a computer but rather the heart of a single purpose electronic device. I base that on comparing the ease (and logic) of assembler programming a mainframe and trying to understand (at this juncture of my life) assembler language on a microprocessor.

I don't know how many have successfully made the transition from low-level programmer on the mainframe to low-level programmer on the PC. I know I tried and have failed miserably.

Joe
 
The 5150 was really a big letdown. Slow with an 8-bit bus. Expensive for what it was.

Was it fair to call it slow in the landscape of personal computers it was released in? Most of the competition (Apple, Commodore, Atari, TI, Sinclair, etc.) were running CPUs with an 8-bit external bus too, but slower.
 
That's an excellent question--and goes to the root of the matter, I think. Did IBM intend to create a "home computer" or one that would see proliferation in the business world? I suspect that it was the former view that was held by IBM and accounted for the Florida operation. Had it been with a business aim in mind, the development most likely would have been carried out in New York or San Jose (Santa Theresa). With that view, speed wasn't a concern--and neither was expandable memory. Recall that the 5150 debuted with a memory configuration of 64KB or less.

The "business side" probably surprised IBM; but it shouldn't have. VisiCalc put an Apple into many an executive's office. We offered the standard suite of MCBA minicomputer applications (on an 8085) as well as spreadsheet and word processing. A couple of friends had set up an operation to produce high-end bicycle carrier racks and were looking for an office computer. They wisely bought a Morrow MD3, which came complete with printer and software applications for much less than the cost of a comparable 5150 system--and this was at the same time that I was purchasing my 5150. (I had considered buying the NEC APC, which again, was a much better deal than the 5150, but could find little on the system--it was a "closed" system).

Then there was IBM. They pretty much had to sell the thing through Sears and Computerland, as they were completely unequipped to sell to minor customers. I recall going to the IBM Sales office on Arques and trying to place an order for five 5150 systems. I was turned over to a sales engineer who didn't have a clue and was informed that we could expect to take delivery of the 5150s in six weeks. We bought them from Computerland the next day.

Another sign of conflicting vision was the Displaywriter (using an 8086), that was viewed by some IBM management as the ultimate business personal computer. But again, IBM couldn't read the PC world correctly. this article shows the problem. Third parties couldn't sell the Displaywriter and IBM couldn't sell general-purpose software for it.

As far as buses, we used a modified Multibus architecture, with a few extra lines to enable better communication between certain cards. The rule was that no card should have a cable hanging off of it. All cables to various peripherals terminated at the backplane. But we were 16-bit capable right from the start. S100, being a technically inferior bus, still managed to progress to 16 bits during that time. It was obvious that the 8-bit ISA bus was never going to host 16 data bits.

My view is that IBM never intended to market anything other than a "home computer". The success of the 5150, 5160 and 5170 convinced them that there was a real untapped business market and that led them to bring out the PS/2 series., which really was a tour-de-force, but again, a "closed" system, which was too bad.
 
Interesting insight as usual, thanks Chuck!

It is amazing to track the occurrences of accidental success in this industry. The IBM PC was likely meant to be a home computer but it proliferated through the business world despite its high price. 2 years later, the PCjr was intentionally designed to be a successful home computer you could use for business, but tanked -- because of the price. Looking back, I would have made the same decisions IBM did and would have been equally surprised by both outcomes.

Although this is 5 years old, I found this relevant: http://forwardthinking.pcmag.com/chips/286228-why-the-ibm-pc-used-an-intel-8088
Miller gets two different stories from two different camps; Gates and Allen claim they convinced IBM to go 16-bit, while the design team and BIOS writer claim that was already the decision before Gates and Allen were approached.. That's always the challenge of recording history; you have to get multiple viewpoints to hone in on the truth.
 
Well, in truth, everyone was looking for a 16-bit CPU. It was pretty clear that the bigger word size and the possibility of expanding addressing was going to be mandatory if we wanted to run applications typically run on minicomputers. Early 16-bitters (e.g., GI CP1600, National PACE) weren't really any faster than the 8080-class CPUs and required a lot more circuitry. Initially, Intel viewed the 8086 as somewhat of an evolutionary dead-end, with the heavy lifting going to the 432. As time went on, it was clear that the 432 project was either going to be prohibitively expensive, complicated, low-performing or all three.

The Fairchild 9440 or MicroNOVA might have been a possibility, but the 9440 was targeted at the defense market and DG wasn't about to sell the MN to outside vendors.

In 1980, we were talking with both Motorola and NSC about their plans for the New World. Motorola had this monstrous 64 pin DIP but with a really good design (absent provisions for virtual memory), great address space and better than anyone, supervisor and user modes (on a multitasking system, you really do need some sort of security). NSC had a bang-up design being done by an outside firm, but it was a story of "Real Soon Now" forever--and NSC's reputation for a short attention span when it came to CPUs. I don't think they floated the 32016 until 1983 and it had the reputation of being quite buggy.

Add to all of this, that in 1980, MOS DRAMs were slow. 150 nsec. in MOS was about the fastest you could buy--beyond that you had to go bipolar (low density, high power). I recall an engineer agonizing about using the Intel 8202 DRAM controller and worst-casing the timing numbers and discovering that applying those times, the DRAM would need a negative access time.

Fortunately, the period between 1980-83 saw huge improvements all around in technology.
 
That's an excellent question--and goes to the root of the matter, I think. Did IBM intend to create a "home computer" or one that would see proliferation in the business world? I suspect that it was the former view that was held by IBM and accounted for the Florida operation. Had it been with a business aim in mind, the development most likely would have been carried out in New York or San Jose (Santa Theresa). With that view, speed wasn't a concern--and neither was expandable memory. Recall that the 5150 debuted with a memory configuration of 64KB or less.

The "business side" probably surprised IBM; but it shouldn't have. VisiCalc put an Apple into many an executive's office. We offered the standard suite of MCBA minicomputer applications (on an 8085) as well as spreadsheet and word processing. A couple of friends had set up an operation to produce high-end bicycle carrier racks and were looking for an office computer. They wisely bought a Morrow MD3, which came complete with printer and software applications for much less than the cost of a comparable 5150 system--and this was at the same time that I was purchasing my 5150. (I had considered buying the NEC APC, which again, was a much better deal than the 5150, but could find little on the system--it was a "closed" system).

Then there was IBM. They pretty much had to sell the thing through Sears and Computerland, as they were completely unequipped to sell to minor customers. I recall going to the IBM Sales office on Arques and trying to place an order for five 5150 systems. I was turned over to a sales engineer who didn't have a clue and was informed that we could expect to take delivery of the 5150s in six weeks. We bought them from Computerland the next day.

Another sign of conflicting vision was the Displaywriter (using an 8086), that was viewed by some IBM management as the ultimate business personal computer. But again, IBM couldn't read the PC world correctly. this article shows the problem. Third parties couldn't sell the Displaywriter and IBM couldn't sell general-purpose software for it.

As far as buses, we used a modified Multibus architecture, with a few extra lines to enable better communication between certain cards. The rule was that no card should have a cable hanging off of it. All cables to various peripherals terminated at the backplane. But we were 16-bit capable right from the start. S100, being a technically inferior bus, still managed to progress to 16 bits during that time. It was obvious that the 8-bit ISA bus was never going to host 16 data bits.

My view is that IBM never intended to market anything other than a "home computer". The success of the 5150, 5160 and 5170 convinced them that there was a real untapped business market and that led them to bring out the PS/2 series., which really was a tour-de-force, but again, a "closed" system, which was too bad.

I think they did predict that small businesses would buy it too.
 
Sigh, 35 years already?

I recall my profound disappointment after seeing the announcement of the CS 9000 shortly before that of the 5150. Many people had hoped for a 68K-based machine--had that happened, I wonder where we'd be today. Would Apple have bothered bringing out the Macintosh? What would PC software look like?

The 5150 was really a big letdown. Slow with an 8-bit bus. Expensive for what it was.

It's interesting to speculate.

Checkout the base price on that CS 9000 - close to 6K without the add-ons.

IBM9000.jpg
 
Last edited:
Back
Top