Bill Gates really did claim that the 640K barrier was due to his decisions - Is there any other actual evidence that this was not the case?

[cj7hawk]

So in another thread hidden away in CP/M world, this came up... And I was surprised that I received immediate criticism for making this observation, after the 640K limit came up in the thread, attributed to IBM.

Me: "I thought the idea of the 640K limit came from Microsoft didn't it?"

Responses:
"This is horse sh*t."
"In short you are picking on Microsoft for building a product to the specifications of the customer that follows the rules of the BIOS they targeted."
"Anyway, the whole 640K thing is a pretty darn stupid criticism."

Well, I come here for the robust debate and some good stuff came out of it, but I remembered hearing something about this from Microsoft back when I was a technology journalist and had frequent meetings with Microsoft to hear what they were doing... And the CP/M forum is not the appropriate place to continue the discussion.

Interestingly, the moment I started using an AI to help me track it down, I got much the same response from an AI, suggesting that the Internet is very anti when it comes to my statement. I even had the AI respond with the information I was looking for, only to suddenly erase the page it was printing and replace it with a trigger warning ! Which means just asking about this tripped a guardrail in google Gemini.

But still I eventually found the original claims and Bill gates did say it was Microsoft's idea (his decision) and he highlighted that he was wrong about his assumptions when he did this, and since it's a really cool MP3, that does seem to bring up a different perspective to what is commonly heard, I thought I'd share it in the correct part of the forum.

The best recording I could find was here: https://csclub.uwaterloo.ca/resources/tech-talks/1989-bill-gates-talk-on-microsoft/

It's at the 21.30 minute mark that he begins to discuss this -

It goes something like this;

Well, in 1981, the shift to the 16 bit design gave us a megabyte of logical memory.
The 8086 has a 20 bit address - it's kind of interesting, it's really a sixteen bit data path, they have these things that they call segment registers, they are really index registers, but they shift them four bits to the left and add that to an offset to create a twenty bit address.
And so that's a one megabyte address space and in that original design I took the upper 384K and decided that a certain amount for the video memory and a certain amount for the rom and IO and that left 640K for general purpose memory.
And so that leads to today's situation where people talk about the 640K barrier. the limit to how much memory you can put to these machines.
I have to say that in 1981, making those decisions, I felt like I was providing enough freedom for 10 years. That is, a move from 64 K to 640 K felt like something that would last a great deal of time. Well, it didn't - it took about only 6 years before people started to see that as a real problem.
Fortunately there is a reasonable solution to this which is that as Intel has moved forwards with it's chip families, the 286 chip that you find inside the AT introduced in 1984, moves us up to a 24 bit address.

So what's the general thinking on this - Was Bill Gates lying? I'm pretty happy to take him at his word for this, but are there other quotes from IBM that would actually counter this?

Also, was there ever a video recording of this talk that exists? I'm surprised it was so difficult to find when I went looking, but it's an interesting claim Bill Gates made that counters the common narrative.

Also, just to avoid it - I am not suggesting "640K should be enough for anyone" was Bill Gates quote since that is very strongly refuted - Just that according to Bill Gates, the 640K barrier decision came from Microsoft.

 

[vwestlife]

In 1981, the 640K barrier was only theoretical. The most the first version of the IBM PC BIOS could address was 544K of RAM. Although, Gates did claim Microsoft convinced IBM to move the barrier up from 512K to 640K:

The ‘640K’ quote won’t go away — but did Gates really say it?

In 1981, when the IBM PC was introduced, Bill Gates supposedly said that 640KB of memory "ought to be enough for anybody." The quote has followed him through the years, despite a lack of solid evidence that he actually said it.

www.computerworld.com

Gates, who is retiring from his day-to-day role at Microsoft Corp. on June 30, also insisted in a 2001 interview with U.S. News & World Report that he hadn't made the comment. "Do you realize the pain the industry went through while the IBM PC was limited to 640K? The machine was going to be 512K at one point, and we kept pushing it up," he told the magazine. "I never said that statement -- I said the opposite of that."

 

[bakemono]

Putting the MDA and CGA cards into the Bxxxx range effectively limited the (consecutive) memory to 704KB. I don't know what the first peripheral to use Axxxx was, which would have brought the cutoff then to 640KB. B.G. sounds like he is taking credit for this address map, but I've never heard that he made any contribution to the hardware design.

 

[cj7hawk]

Putting the MDA and CGA cards into the Bxxxx range effectively limited the (consecutive) memory to 704KB. I don't know what the first peripheral to use Axxxx was, which would have brought the cutoff then to 640KB. B.G. sounds like he is taking credit for this address map, but I've never heard that he made any contribution to the hardware design.

A0000 is the EGA card and future VGA, but he definitely is taking the credit for this, and he also takes credit for other aspects of the IBM project, such as simultaneously supporting both Project Go ( IBM assigned to Matsushita ) and Project Chess with the same OS as Matsushita supposedly source this from him, which would go a long way to explaining why the Japanese were so far ahead in graphics if there was cross-contamination of ideas there.

He also suggests that the original goals of Project Chess were to see if IBm could successfully build a computer in 2 years - not to build a competitor to the systems of the era, in which case, it's entirely possible Microsoft had an invisible hand in most if not nearly all of the early IBM PC architecture - which of course would be owned by IBM due to the agreement if the case...

 

[cjs]

This was asked and answered over half a decade ago on retrocomputing.stackexchange.com, which is always a good place to look for answers to questions like this. The TLDR is:

• Stephen Kit hauls out the original IBM PC techref manual and points out that they explicitly marked address space from 640K onward as reserved.
• Patrick Schlüter points out that he was running an (apparently standard) MS-DOS or PC-DOS on a NEC V30 system that used his Atari ST's RAM, and he was able to get it to use the RAM that the system could place in the $A0000-$AFFFF range, and even in the $B0000-$B7FFF range (giving 736 KiB DOS RAM) if he was willing to give up MDA emulation and other graphics modes that used that range.

I have no idea why Gates says in that talk that this was his decision; obviously the final call was IBM's, since they were designing the hardware. I wonder if he's just misremembering, or if he at least had some input on that, and he's taking more credit than he deserved.

 

[cj7hawk]

This was asked and answered over half a decade ago on retrocomputing.stackexchange.com, which is always a good place to look for answers to questions like this. The TLDR is:

• Stephen Kit hauls out the original IBM PC techref manual and points out that they explicitly marked address space from 640K onward as reserved.
• Patrick Schlüter points out that he was running an (apparently standard) MS-DOS or PC-DOS on a NEC V30 system that used his Atari ST's RAM, and he was able to get it to use the RAM that the system could place in the $A0000-$AFFFF range, and even in the $B0000-$B7FFF range (giving 736 KiB DOS RAM) if he was willing to give up MDA emulation and other graphics modes that used that range.

Let's assume all of that is true - none of that counter's Bill's claim.

Unless IBM wrote the technical reference manual AFTER microsoft was engaged - I doubt that is in dispute - and if form follows function, then the architecture may well have been created by Bill Gates as he claimed and subsequently documented... And now that this claim is clear and reasonable, I can't find anything that is evidence to the contrary.

Seems everyone just assumed that Microsoft developed the hardware and told Microsoft what it was going to look like, but according to Bill Gates, IBM wanted to build an 8 bit machine - I'm going to guess something like a CP/M system, and it was him that convinced IBM to shift their thinking to the new x86 architecture.

His argument revolves around software challenges Microsoft was facing around 64K machines at the time, which seems reasonable.

This being the case, he may well have been the spiritual architect of the IBM PC - and Not IBM ( or it was IBM via their agreement with Bill Gates for another perspective ).

Nothing in anything IBM has said since seems to counter this scenario.

 

[commodorejohn]

Curious tidbits, and I'd love to know where the truth lies. I admit I'm not well-versed in the lore of the PC project, but do we know exactly when MS was looped in by IBM? Per Wiki they turned to Gary Kildall at Gates's suggestion, so it must've been before that 1980 meeting...

 

[cj7hawk]

Curious tidbits, and I'd love to know where the truth lies. I admit I'm not well-versed in the lore of the PC project, but do we know exactly when MS was looped in by IBM? Per Wiki they turned to Gary Kildall at Gates's suggestion, so it must've been before that 1980 meeting...

Which brings up another unanswered question - If IBM needed Bill Gates to show them what DR was doing - and to introduce them to Gary Kildall, then they probably hadn't thought too deeply around what they wanted to do and were still in discovery mode - quite possibly they hadn't fully considered what the PC architecture would look like at that point.

It seems like a detailed timeline might be in order to look deeper...

 

[cjs]

Nothing in anything IBM has said since seems to counter this scenario.

In "Bill Gates, Microsoft, and the IBM Personal Computer" (Infoworld, 1982-08-23), there's a quote from IBM:

IBM won't say much about Microsoft's input into the design of the PC. IBM spokesman John Tweddell says: "We went to Microsoft because we know that their BASIC interpreter is widely used and is very good. IBM designed the PC. They [Microsoft] looked at the design and thought we had a nice machine."​

 

[cjs]

Per Wiki they turned to Gary Kildall at Gates's suggestion....

That sounds rather unlikely to me, and is far from what Wikipedia says:

In the summer of 1979, Ron Mion, IBM’s Senior Business Trends Advisor for entry-level systems, proposed a plan for IBM to enter the emerging microcomputer market. At that time, the likes of Apple and Tandy were starting to encroach on the small-business marketplace that IBM intended to dominate. Mion believed that that market would grow significantly and that IBM should aggressively pursue it. However, he felt that they wouldn’t be successful unless IBM departed from its long-standing business model.​

​

Mion’s plan called for three major departures from how IBM traditionally did business. Mion felt that, if IBM wanted to compete in the microcomputer market, it would need to...​

​

b) Use a low-cost, third-party operating system. Mion felt that this was imperative in order to foster a cottage industry that could develop a broad array of applications that would help small businesses justify the purchase of a computer. Mion recommended Digital Research’s CP/M and a new O/S called MS-DOS from a little-known company named Microsoft.​

​

...That plan made its way up the chain of command but was ultimately rejected in the fall. The top IBM executives reaffirmed that all “IBM” computers, and their major components, must be developed, manufactured, sold, and serviced by IBM.​

​

In January of 1980, Tandy released their Annual Report and, as was predicted in Mion's plan, it confirmed that their 1979 shipments had exceeded 100,000 TRS-80s (about $50 million worth). IBM quickly dusted off Mion’s marketing plan.​

That said, I notice that Wikipedia has no references for this. Then again, the "Gates told IBM about DRI" story seems to come from Big Blue p.18, and that book too is even more short on references. That said, there's a strong implication there (from directly quoted text before and after that section) that this was derived from information from Jack Sams, an IBM executive working under Bill Lowe on software strategy, who had several early meeting with Bill Gates.

 

[Zare]

I'm not really sure what we're talking about here.

Has Bill Gates supposedly claimed that they aided or were heavily involved in PC design?
I fail to see where, in what part.

5150 goes back to 5100 and some 'wrong' choices that otherwise pretty successful PC arch made are due to the lineage, like DMA controller. So here blame is solely on IBM and Intel.

20-bit address space is not negotiable. What is, is the mapping.

Obviously there have been differences in opinion between IBM and Microsoft how PC family is going to end up in the long run and Microsoft was right; most of the 1MB address space should be left to RAM.
Keep in mind three ways of accessing peripherals - PIO, DMA and dual ported memory. Perhaps, this is my conjecture, IBM thought a lot more peripherals are going to be directly memory mapped like graphics card.

 

[cj7hawk]

In "Bill Gates, Microsoft, and the IBM Personal Computer" (Infoworld, 1982-08-23), there's a quote from IBM:

IBM won't say much about Microsoft's input into the design of the PC. IBM spokesman John Tweddell says: "We went to Microsoft because we know that their BASIC interpreter is widely used and is very good. IBM designed the PC. They [Microsoft] looked at the design and thought we had a nice machine."​

That is fascinating. Technically, anything Microsoft told IBM was IBMs property as per the NDA, and even if Microsoft influenced this, it's still absolutely true that IBM designed the hardware - But the source of ideas and decisions is the interesting detail.

Bill Gates also made clear he wanted to part of designing or working with hardware - that was something he said was Paul Allen's preference. He only wanted microsoft to write software.

So if his ideas became IBMs then he had no reason to keep good ideas from IBM and this brings up yet another question about IBM that never got answered - What exactly was it that IBM liked about Bill Gates and why did they go back there after his reference wasn't the success they hoped?

I get the feeling there's a LOT IBM took from microsoft... One of the true legends of early computing who understood the market like no other, who had no reason to hide potential hardware ideas, and was in a position to influence every aspect of design and did so freely...

That sounds rather unlikely to me, and is far from what Wikipedia says:

In the summer of 1979, Ron Mion, IBM’s Senior Business Trends Advisor for entry-level systems, proposed a plan for IBM to enter the emerging microcomputer market. At that time, the likes of Apple and Tandy were starting to encroach on the small-business marketplace that IBM intended to dominate. Mion believed that that market would grow significantly and that IBM should aggressively pursue it. However, he felt that they wouldn’t be successful unless IBM departed from its long-standing business model.​

​

Mion’s plan called for three major departures from how IBM traditionally did business. Mion felt that, if IBM wanted to compete in the microcomputer market, it would need to...​

​

b) Use a low-cost, third-party operating system. Mion felt that this was imperative in order to foster a cottage industry that could develop a broad array of applications that would help small businesses justify the purchase of a computer. Mion recommended Digital Research’s CP/M and a new O/S called MS-DOS from a little-known company named Microsoft.​

​

...That plan made its way up the chain of command but was ultimately rejected in the fall. The top IBM executives reaffirmed that all “IBM” computers, and their major components, must be developed, manufactured, sold, and serviced by IBM.​

​

In January of 1980, Tandy released their Annual Report and, as was predicted in Mion's plan, it confirmed that their 1979 shipments had exceeded 100,000 TRS-80s (about $50 million worth). IBM quickly dusted off Mion’s marketing plan.​

Well, we can either discount this as a reliable source - because by all accounts, there was no such thing as MS-DOS in 1979

Or we can consider it a reliable source and then it means they were talking to Microsoft about this *well* before they talked to Digital Research - and it also infers Ron Mion was very close to Microsoft at this point, in which case, it still looks like Bill Gates is telling the truth. He made these claims on more than one occasion, so I don't think it was misspoken.

If Microsoft was this heavily involved in the PC early design, that they were already considering MS-DOS which didn't exist and for which the problem of "not existing" was yet a long way away from manifesting, it suggests much earlier contact with IBM than we are aware of - that this "2 year project" and Microsoft's involvement entirely began in 1979 and it wasn't until 1980 that they had their meeting with Gary Kildall. This would go a long way to explaining their confidence in Microsoft.

 

[cj7hawk]

I'm not really sure what we're talking about here.

There's an inconsistency in the story that is assumed when the PC was created, and while the inconsistencies are often questioned, they have never been explained.

This could very well explain those inconsistencies and potentially makes Bill Gates, not IBM, the true starting point of the IBM PC.

Damn it Bill, you're not dead yet... It's been 45 years! Tell us what really happened.

20-bit address space is not negotiable. What is, is the mapping.

Interestingly, Bill Gates has also claimed they were originally going to go with an 8 bit system - likely a z80 or 8080 though he never really elaborated - and that this outcome was one of the things he influenced. As well as was noted earlier that it was going to be a 512K RAM machine.

Obviously there have been differences in opinion between IBM and Microsoft how PC family is going to end up in the long run and Microsoft was right; most of the 1MB address space should be left to RAM.
Keep in mind three ways of accessing peripherals - PIO, DMA and dual ported memory. Perhaps, this is my conjecture, IBM thought a lot more peripherals are going to be directly memory mapped like graphics card.

I think this was emerging already with cards like the Matrox, and IBM probably solved the minor details and would have still made the final decisions on architecture. Maybe they couldn't decide whether Text or Graphics were the new frontier and hedged their bets? Or maybe they just felt the markets were too different to converge until EGA came along?

 

[vwestlife]

Remember, Gates isn't a hardware guy, and never has been. If he said "IBM wanted to build an 8 bit machine", he might simply be talking about the Datamaster, an 8-bit 8085 machine upon which the PC was heavily based, and was released almost in parallel with the PC (just one month before it, in July 1981).

And if you've seen his testimony in hearings, he's good at taking credit for things and then not recalling he ever said such a thing when it later became inconvenient for him...

 

[Zare]

Interestingly, Bill Gates has also claimed they were originally going to go with an 8 bit system - likely a z80 or 8080 though he never really elaborated - and that this outcome was one of the things he influenced. As well as was noted earlier that it was going to be a 512K RAM machine.

Nah. I don't believe him at face value.

The IBM PC design was led by Don Estridge and 8 bit CPU was never under consideration. Intel was chosen over TI and Motorola, due to IBM's familiarity with Intel in Datamaster line.

 

[cjs]

The IBM PC design was led by Don Estridge and 8 bit CPU was never under consideration. Intel was chosen over TI and Motorola, due to IBM's familiarity with Intel in Datamaster line.

Not to mention that much of the Datamaster work could be re-used, since the 8088 had an 8-bit bus (and was deliberately intended to be compatible with the old 8-bit peripherals). I have the feeling some of the excesses of the PC (a separate dedicated DMA controller? Really?) might have come out of the Datamaster design. But I could be wrong about that; maybe the 8237 really is an addition worth its cost.

At any rate, looking at the Datamaster the need for more than 64K of RAM would have been quite evident, since the Datamaster itself supported 128K.

 

[cj7hawk]

Nah. I don't believe him at face value.

The IBM PC design was led by Don Estridge and 8 bit CPU was never under consideration. Intel was chosen over TI and Motorola, due to IBM's familiarity with Intel in Datamaster line.

This was never at question - Bill Gates did not involved himself in hardware.

But has Don Estridge ever published his own claims about this time?

Also, as a hardware designer myself, I would be very surprised if it was true that he never considered an 8 bit CPU and suspect there is retelling error in that and the Datamaster on which the PC was based somewhat was an 8085 8-bit CPU - a good designer should consider both the current technology as well as the emerging technology.

Also, there's usually a formative period before you sit down to the detailed design process during which you solicit a lot of ideas from other people and typically you don't consider this much once you've made your mind up, so again, Bill Gates may have influenced this before that time, but what I'm trying to find at the moment are clear records left behind by the original actors.

I don't doubt Don Estridge designed it, and he's often referred to as the father, but what if he was the mother, and Bill Gates is the daddy?

Did Don Estridge ever write a book on it? We know Estridge delivered a PC to Bill Games 27th November 1980 and they were in constant, frequent and regular contact once things were rolling. Again, none of this disproves Bill Gates claims.

 

[Quagmire]

In 1981, the 640K barrier was only theoretical. The most the first version of the IBM PC BIOS could address was 544K of RAM. Although, Gates did claim Microsoft convinced IBM to move the barrier up from 512K to 640K:

The ‘640K’ quote won’t go away — but did Gates really say it?

In 1981, when the IBM PC was introduced, Bill Gates supposedly said that 640KB of memory "ought to be enough for anybody." The quote has followed him through the years, despite a lack of solid evidence that he actually said it.

www.computerworld.com

Wait , . . wasn't that the other way around? I understand the 640K limit thing is because of the effects of the Intel intrinsic chip design. Am I wrong?

 

[jafir]

I understand the 640K limit thing is because of the effects of the Intel intrinsic chip design. Am I wrong?

It's more about the choices of where other things live in memory beyond 640K. the SCP machine could have a full 1MB of RAM:

The First DOS Machine

Seattle Computer Products (SCP) introduced their 8086 16-bit computer system in October 1979, nearly two years before the introduction of th...

dosmandrivel.blogspot.com

Of course, that makes me wonder if anyway has ever hacked PC hardware and BIOS to move the MDA or CGA closer to the top of RAM, to give more contiguous memory. Of course it would break all of the software that tries to write to those cards directly.

 

[JohnElliott]

Of course, that makes me wonder if anyway has ever hacked PC hardware and BIOS to move the MDA or CGA closer to the top of RAM, to give more contiguous memory. Of course it would break all of the software that tries to write to those cards directly.

Yes, there were add-on MMUs like the AllCard that could map out the display adapter and populate up to 952k with RAM.