Why did IBM choose the 8088?

[bluejay]

I'm still not entirely sure why IBM chose the 8088 for their PC. For its price I don't see why it would have been a problem if it used an 8086... Or was it some stupid reason like the engineer that was in charge of the CPU couldn't get ahold of an 8086 in time? A real 16 bit processor could have made the original PC better.

[carlsson]

If you like to believe Wikipedia, there is a chapter Design process that talks about the choice of CPU.

https://en.wikipedia.org/wiki/IBM_Personal_Computer

 

I have a feeling it was brought up here on AA fairly recently as well.

 

Beyond the CPUs mentioned in the article (TMS9900, 68000, 8086/8088 and IBM 801), elsewhere it has been mentioned that Zilog Z8000 could've been in the run as well if it wasn't for the fact that Exxon Enterprises, single owner of Zilog, aimed to be a competitor to IBM and according to the then CEO of Zilog it was reason enough for IBM to not consider them. I'm not sure if that is an accurate explanation though.

[youxia]

Quote

IBM chose the 8088 over the 8086 because Intel offered a better price for the former and could supply more units.^[15] Another factor was that the 8088 allowed the computer to be based on a modified 8085 design, as it could easily interface with most nMOS chips with 8-bit databuses, i.e. existing and mature, and therefore economical, components. This included ICs originally intended for support and peripheral functions around the 8085 and similar processors (not exclusively Intel's), which were already well known by many engineers, further reducing cost.^[g]

https://en.wikipedia.org/wiki/Intel_8088

[Bruce Tomlin]

The way I hear it, what they wanted was a 16-bit CPU with more than 16 bits of address bus, and 8-bit data bus. The extra address bits were needed to go past 64K without fiddly bank switching, and a 8-bit data bus meant that you need half as many RAM chips for the minimum configuration. It's important to keep in mind that DRAM was getting cheaper, but still very price sensitive. So they were talking to Intel about the 8088 and apparently also to Motorola about the 68008. That they specifically wanted the 68008 is less certain, but it fits with the time frame.

 

There is a lot of speculation here, but this was a time when the sales guys at Motorola really only wanted to sell the 68000 in hundreds and thousands quantity for use in expensive Unix systems. (Read DTACK Grounded for more about Motorola's attitude. Before 1983 or so they really didn't want the 68000 to be used for home computers or embedded applications.) IBM had a date by which the chip would have to be ready, and Motorola refused to make a commitment to that date. I guess it wasn't a big priority at Motorola, and they were just making the 68008 to fill out their product line. Apparently it was actually ready by IBM's deadline (Wikipedia says "introduced in 1982"), but IBM had already gone with the 8088.

 

That Wikipedia article about the 8088 mentions something that was also important: the bus interface. The 68000 had a synchronous bus by default, because the higher clock speed meant it needed wait states depending on what it was talking to, letting them use existing 6800 peripheral chips made for 2MHz clocks. The difference was adding two lines: DTACK and BERR*. Motorola's idea was that every address would send back a signal (DTACK) when the memory or I/O device was ready, then you could mix and match different speed devices. But if you grounded it (hence "DTACK Grounded"), it would run at full speed, becoming a 6800-style bus, assuming everything you had could keep up with the faster clock. There were already minor differences between the 8080-style bus and the 6800-style bus, but this was an extra pain in the ass if you wanted a simple system.

 

(*BERR would tell the CPU that the requested address was invalid. If neither DTACK nor BERR ever happened, the 68000 would wait forever for it. This is why reading certain address ranges on the Sega Genesis will lock up the console. You could probably implement it with a 4040 counter and a gate or two, but it wasn't useful for finished games.)

Edited November 4, 2020 by Bruce Tomlin

[Krebizfan]

The histories of the Sinclair QL suggest that the 68008 was not available in volume until 1983 making it a challenge to include in the IBM PC. Motorola couldn't manufacture enough 68000 chips to meet demand so there was no excess capacity to place in a lower cost packaging. 

 

The 8086/8088 were the only chips that could have been used to get a system developed to manufacture in about a year and would not need a complex memory banking scheme before the 5 years of planned production finished. The 8086 would have been slightly better but the support chips cost more and the need to have 16-bit memory access doubles the minimum number of required chips. Getting a second bank of 16k RAM would have added about $100 and the 16-bit peripheral chips would have increased cost by another couple of hundred dollars. 

[bluejay]

I dunno, the 8086 was used in later IBM computers, and considering its $1500 base price the complexity(and the further cost increase due to it) must have been the biggest problem.

 

The original PC could have been better with a true 16 bit processor. But looking at the 5150's specs I think it was clearly aimed at the home market, with its 16k(base) RAM and cassette port. (that is, other than its price) It really makes me wonder how in hell it was such a huge success. Outrageous price tag, lackluster performance, and what must have been amazingly boring adverts.

[Krebizfan]

The models and prices seemed similar to the Tandy Model III. The 16K no drives 5150 was a match to the Model III with 16K no drives and a cassette port. The IBM listed for more but if someone shopped around, getting a 30% discount on a 5150 was possible. IBM offered great margins for resellers. Moving up the food chain, the top of the line Model III with 48k, 2 drives, and the high resolution graphics board (roughly equal to monochrome CGA) cost about $2,900 which isn't that far off the price of a 64k dual drive 5150.

 

Cassette port proved necessary for initial testing. The disk controller took a few months to get working. The cassette port was also a requirement for some school districts which used it to steer contracts to preferred bidders.  

 

IBM's only real flaw with the design was going for single sided drives though production of double sided drives was a bit limited on 1981. IBM did release a driver for DOS 1.0 that allowed the use of double sided drives as pairs of single sided drives in 1982 for those that upgraded drives before DOS 1.1 was released. 

 

IBM's big advantage for the 5150 was the quality of support. One was unlikely to get a failed 5150, much better results then when ordering from other companies. If there was a problem, IBM went to great lengths for resolve it. I have read stories of IBM flying multiple engineers to look at a defective 5150. Profit margins on the 5150 couldn't sustain that so its good that the 5150 was so reliable. 

 

Prices plummeted fast. About 2 years later, the 8086 based Advance-86 PC compatible was released with 128 K, 2 joystick ports, parallel, serial, CGA, and a cassette port. No drives. That was priced at 400 pounds which if converted to dollars and removing the VAT should have meant a US price of $450. 

[mr_me]

IBM chose the 8086 cpu.  However, the 16-bit support chips were not completely ready.  IBM had a deadline so intel made the 8088 cpu that could be used with their old 8-bit support chips.  It's still a 16-bit cpu, 100% software compatible with an 8086.

Edited November 5, 2020 by mr_me

[carlsson]

Many businesses which up to then had not entered the computerized era, finally felt confident to do so when the well renowned brand IBM entered the personal/home market, both because their reputation of high quality and support, and knowing that the software market (*) would follow like ducks follow the first thing they saw when they were hatched. Apple, Tandy and to a lesser degree Commodore had tried but didn't convince as many businesses as IBM supposedly did.

 

(*) Or at least IBM made sure they would have their killer apps ready for the launch, approaching companies with products like VisiCalc, WordStar etc ahead of release. I've read somewhere how they analyzed the best selling software packages on Apple (and perhaps Tandy) to determine which ones they needed to have onboard in order to secure sales of the PC. If it had been launched without that initial 3rd party support, and perhaps a different approach towards that market, it probably would not have flown as well as it did.

 

 

[Keatah]

2 hours ago, bluejay said:

The original PC could have been better with a true 16 bit processor. But looking at the 5150's specs I think it was clearly aimed at the home market, with its 16k(base) RAM and cassette port. (that is, other than its price) It really makes me wonder how in hell it was such a huge success. Outrageous price tag, lackluster performance, and what must have been amazingly boring adverts.

To answer that, just ask the people who bought one at the time. What was their reasoning?

 

When the 5150 came out it was not aimed at the home user.

 

I was too young to understand all its wonderful and incredible benefits. Reliability, support, versatility, and "generic-ness". The 5150was introduced at a time when desk-size "micros" were seriously being miniaturized. It was a brave new world. And if a proven company from the old world was making modern machinery for the new world, well, it's clear whom customers were going to go with.

 

And it's a business machine first and foremost. It got things done reliably. Don't expect exciting graphics or sound or advertisements. It wasn't a hobby/toy like the other micros from Commodore, Atari, Coleco, Mattel. Price tag wasn't a big deal to businesses. Nothing outrageous about it. Flashy game ads weren't required. All you needed was text explaining the features and specifications.

 

Its documentation rivaled what was available on the Apple II. And in some cases far exceeded it. There are many parallels between the Apple II and the IBM PC.

 

IBM liked Intel chips because of their documentation. Not to mention all the support and peripheral chips to go with them. And this vast selection of chips (and compatible chips from other mfgs) was already familiar to design engineers of the day.

 

The PC was a culmination and natural extension of forces already underway in the industry

[mr_me]

3 hours ago, bluejay said:

I dunno, the 8086 was used in later IBM computers, and considering its $1500 base price the complexity(and the further cost increase due to it) must have been the biggest problem.

 

The original PC could have been better with a true 16 bit processor. But looking at the 5150's specs I think it was clearly aimed at the home market, with its 16k(base) RAM and cassette port. (that is, other than its price) It really makes me wonder how in hell it was such a huge success. Outrageous price tag, lackluster performance, and what must have been amazingly boring adverts.

That $1500 base model wasn't complete for a home computer.  All application software including games were on floppy not cassette.  And to hook it up to your TV would require you to get an external RF adapter.

 

That base model was popular for many years.  Resellers probably added third party floppy drives, hard drives, and third party ram at considerable savings over an all ibm equivalent.

 

I never considered an ibm pc or compatible as a home computer until the late 1980s.  Clearly ibm positioned it as one in some of their 1981 brochures and made sure games were available at launch.

 

They would have also been after the educational market.  My high school had the commodore pet.  How did the ibm pc do in schools in the early 1980s?

Edited November 5, 2020 by mr_me

[carlsson]

In America, didn't Apple corner the educational market with great discounts early on? As for home computer, we must not forget the purposely crippled PC jr a few years later, still priced well above the gaming computers from other brands. But as noted, it never was IBM's market or intention to sell to home users/gamers. Forty years later in retrospect, the difference between them all may not seem that huge, and considering the PC platform nowadays is within everyone's reach, it can be easy to mistake the IBM PC for a well built, high end but overpriced home/personal computer.

 

Personally, I found the entire PC platform to be priced as unobtainium until at least 1993/94, though I understand that in America it had a breakthrough earlier on, like at least the late 80's. You have the difference to success of machines like the Amiga and Atari ST baked into this as well, without the intention to go entirely off-topic.

 

In any case, even with the 8088 and the price tag, IBM seemed to sell more units than they could produce which explains why the European launch of the IBM PC was delayed by some 1 - 1.5 years and actually opened a market gap for similar systems like Victor 9000/Sirius 1 not to mention all other brands. IBM didn't wait to get into Europe for its own sake, but because they had their hands full of delivering on the American markets.

[mr_me]

The ibm pcjr happened precisely because they weren't doing as well in the home market as they hoped.  IBM thought people were intimidated by their computers, so they dummed it down but kept the high price.  By that time you had 100% compatibles from third parties that were much more competitive.  They were competing against Amiga and Apple for the home market.  All were too expensive for most families.

[bluejay]

Because from what I've read it seems to suggest IBM was going to make their Personal Computer, well, a personal computer.

Obviously no average home user bought the PC, and various businesses that didn't like huge minicomputers bought them with MDA, the 5151 monitor, and a floppy drive or two. (the PC had an option for 2 floppy drives at launch, right? Or did that come later?)

But I mean, IBM must have expected it to be used as a home computer playing games at some point because if they didn't they wouldn't have made a joystick controller card.

Also the base model had pretty much exactly what the average home computer at the time came with. 16-64k RAM, crude color graphics(CGA), BASIC in ROM, cassette I/O. But it also had the ability to expand far beyond that, along with something called "build quality" which a lot of home computers lacked. I mean, if I dropped a VIC-20 off a 10 story building it would be shattered beyond recognition but a 5150 would be totally fine.

Also it had the Model F.

 

It was versatile.

It could be a normal home computer, and it could be expanded to be a (not so) great game machine, or a powerful business computer, with the right kind of add ons.

The off the shelf components must have made it more expensive and less efficient, and that IBM badge adds $300. But it's what made the PC so successful, wasn't it?

[Keatah]

It was those off-the-shelf components and their familiarity to engineers worldwide that allowed for some fantastic expansion options. Maybe a little more costly, maybe less efficient in terms of number of parts needed, but the blank slate versatility was staggering.

 

It was also interesting to watch the lock-step upgrade patterns of, say, the DOS software itself, BIOS, HDD interfaces, and HDDs themselves. There would be a limit on the size of storage supported, then there's be a breakthrough in one area. The other three would catch up at different paces. Then we'd be bottlenecked again, till another breakthrough elsewhere. Along the way there were profits to be made, performance barriers to be broken, and customer pride & happiness & satisfaction were made. Steady progress up a rocky road that still continues today.

 

Unfortunately I had developed a wanting interest in the PC around the 286-12MHz time. Late in the game compared with the Apple II. So I missed the early 8088 and 8086 machines when they were new. Oh I was aware of them alright. But their cost was an instant turn off and I didn't fully understand or appreciate the extraordinary versatility and complexity of its ecosphere. I was way too entrenched in 8-bit consoles and the upcoming 16bit Amiga. For a while I seriously thought the Amiga was going to take over! How naive! And that set me back further. A distraction and money-sink throughout the 386 epoch.

 

I put a stop to the Amiga when word processing on it became more trouble than it was worth. And there was very little technical and astronomy software for it anyways. So double-whammy. It didn't take too long, but I got a 486 and never looked back.

 

I was always told the 486 was binary compatible with stuff from the 8088. And it turned out to be true. And that was impressive in and of itself. I slowly began to appreciate how things were progressing.

 

So my first PC had 4MB of RAM, Windows 3.1, 486 DX2/50, 200MB HDD, 1MB 16-bit ISA graphics, 5.25 & 3.5, 2x serial, 1x parallel, gameport, 15" monitor. And in short time a SoundBlaster 16.

 

It was cool. I could put a modem in there, and get this.. A SECOND PARALLEL PORT! WoW! Eventually I would get 2 more hard disks, MIDI daughtercard, CD-ROM, the custom memory board for up to 64MB, network card, Snappy digitizer, ZipDrive, and other miscellanea.

 

I was completely blown away that the 486 had a math-coprocessor - was like a holy grail at the time. And it worked well with some astronomy stuff of the day, like Dance of the Planets.

 

It should also be noted that as I was building up a Pentium II 266 in the late 90's I had temporarily borrowed the graphics board and soundcard (from the 486) till I could afford a proper Riva-128 and SoundBlaster AWE64. Then I returned the borrowed hardware back to its rightful place.

 

I wasn't disappointed and only wished I got into PC sooner.

[Keatah]

55 minutes ago, bluejay said:

It was versatile.

It could be a normal home computer, and it could be expanded to be a (not so) great game machine, or a powerful business computer, with the right kind of add ons.

The off the shelf components must have made it more expensive and less efficient, and that IBM badge adds $300. But it's what made the PC so successful, wasn't it?

On a technicality I will completely and 100% disagree here. While true, the 5150 itself wasn't great for gaming. Its later siblings like the 386/486 and up were incredible game machines, surpassing all that came before. Especially with fast 2D and new 3D cards, and FM/MIDI music boards.

 

The base architecture was versatile enough to accept the new powerful add-ons. And have its internals pushed faster and farther and made bigger than ever. To do that kind of expansion on an ST or Amiga or anything else required a redesign of the whole machine. Everything PC was modular.

 

[bluejay]

1 minute ago, Keatah said:

On a technicality I will completely and 100% disagree here. While true, the 5150 itself wasn't great for gaming. Its later siblings like the 386/486 and up were incredible game machines, surpassing all that came before. Especially with fast 2D and new 3D cards, and FM/MIDI music boards.

 

The base architecture was versatile enough to accept the new powerful add-ons. And have its internals pushed faster and farther and made bigger than ever. To do that kind of expansion on an ST or Amiga or anything else required a redesign of the whole machine. Everything PC was modular.

 

With a VGA card and a Sound Blaster, although rather inappropriate, it can run games like Prince of Persia and Lemmings beautifully from what I heard(I don't own a 5150 so I don't know if that's the case for sure) but it runs fine on my 80286 powered Compaq Portable II with SB sound so I think it's safe to assume both would run on the 5150 with VGA and SB.

But yeah, the 5150 kinda sucks for gaming, especially with the Model F. The Model F is great for typing on(again I've never used a Model F seriously, only on a similar keyboard on an IBM typewriter) but I can't imagine it being in any way good for games.

I completely agree on the fact that 386 or better PCs were what really made PC gaming great, along with VGA graphics and the Sound Blaster. It really made the personal computer competitive with video game consoles. Playing games on a 5150 would be kind of like trying to race a van. It's can be done, and with the right upgrades rather well, but it's really not designed for it.

[youxia]

Early PCs were meant as business machines first and foremost. But it doesn't meant gaming wasn't a factor as well, it's just that it was "computer" gaming, a vastly different concept from what we have now when it's all more unified and there is no big console/PC divide.

 

Back then there was such divide, and genres such as flight sims, text-> point'n click adventures, and strategy games were really big as well. That's why MS Flight Simulator and other Sublogic titles were such huge sellers, ditto Infocom's adventures and the likes of King's Quest. They all hail from pre-VGA/SB era and were marketed to different audiences. So people who bought these machines for serious tasks could also indulge in these "serious" games, as opposed to the arcadey consoles with their jump/shoot line ups.

[mr_me]

Early 1980s arcade action on the original ibm pc with cga graphics.  Audio was the only problem.  They even supported four button game controllers.  The biggest game developers/publishers were supporting the ibm pc before the crash.

 

 

 

[carlsson]

Were third party components really more expensive than if IBM had developed, tested and manufactured everything on their own? I mean with all the process steps and delays involved in producing IBM hardware, I'm not sure it would be any cheaper to only utilize inhouse technology.

[mr_me]

In 1980 ibm came out with the 5120, their lowest priced computer to date.  It had an ibm cpu and a starting price of $9340.  They knew they couldn't compete.

[bluejay]

That's a good point. Kind of off topic but wasn't the Apple II also built with all off the shelf components as well? What made the IBM PC overtake the Apple II? They were both popular and powerful computers with a great software library and similar form factor.

[mr_me]

The apple ii is 8-bit and replaced with the macintosh by apple.  With the apple ii being open architecture and the macintosh being closed architecture, the ibm pc and compatibles kind of picked up where the apple ii left off.

[Krebizfan]

The Apple II was nearing the design limits at the time. Apple had the Apple III as a follow-on but the Apple III had major problems freeing the IBM PC to secure the high end micro market.  IIe and later models kept selling but the IBM PC was established by then. 

 

The 5100 series were not good comparisons for price since it used static RAM and had a huge store of Read Only Storage. Much more expensive but did improve performance. 

[bluejay]

I wonder what the computer market would have become of had the Apple III actually been a good computer.