CPU family tree - help needed

[kuro68k]

The criteria for inclusion is that they were used in computers. As far as I can tell the Toshiba and NEC parts were never used in computers, only embedded applications.

The Electronic Arrays 9002 seems to have been used in a single board computer that was only ever used as a demonstrator, so doesn't really count.

If you know of machines they were used in though, please let me know.

 

[Chuck(G)]

If that's your criterion, then the 4040 and 4004 need to be removed. At best, they were used in calculators, not microcomputers. Same for the TMS1000 and the PPS4 (You left out the PPS8). Did the NSC PACE ever appear in a microcomputer, other than National's own development system?

 

[DeltaDon]

One reference says that the K6-3+ was released 3rd 1998: https://www.anandtech.com/show/42/amd-k6-review.

I don't know if the K6-3+ (and the cache reduced 2+) would be considered a rehash of the K6-III or if a clean sheet new redesign. I believe it was a step towards the next CPU - the K7 Athalon, but I say that because just word of mouth info I got at the time from a friend that knew an AMD designer. There's some info that leads me to believe that it was a whole new die design because of all the changes added compared to the K6-III. My thought is based upon this: https://en.wikipedia.org/wiki/AMD_K6-III and: https://www.anandtech.com/show/707/super7-upgrade-guide-the-k6-2-and-k6-3-processors/2

 

[kuro68k]

If that's your criterion, then the 4040 and 4004 need to be removed. At best, they were used in calculators, not microcomputers. Same for the TMS1000 and the PPS4 (You left out the PPS8). Did the NSC PACE ever appear in a microcomputer, other than National's own development system?

That's a fair point. I guess the additional criteria is that they can be included if they are historically significant in some way, like the 4004 is the first commercially produced microprocessor in a single chip.

So based on that, I will reconsider your suggestions. In fact looking at my latest revision I already had the Electronic Arrays 9002! I have now added the other two as well.

The NEC part has historical significance as their first single chip microprocessor. The Toshiba part is more questionable, but there was room for it and it does appear to be a microprocessor rather than a microcontroller.

I suppose you could say I'm more open to very early microprocessors being included because they are historically important.

Updated document attached. It now has a key as well. It needs cleaning up a bit and making pretty. Not sure why the years on the right are cut off either, Inkskape shows them on the page.

 

[kuro68k]

One reference says that the K6-3+ was released 3rd 1998: https://www.anandtech.com/show/42/amd-k6-review.

I don't know if the K6-3+ (and the cache reduced 2+) would be considered a rehash of the K6-III or if a clean sheet new redesign. I believe it was a step towards the next CPU - the K7 Athalon, but I say that because just word of mouth info I got at the time from a friend that knew an AMD designer. There's some info that leads me to believe that it was a whole new die design because of all the changes added compared to the K6-III. My thought is based upon this: https://en.wikipedia.org/wiki/AMD_K6-III and: https://www.anandtech.com/show/707/super7-upgrade-guide-the-k6-2-and-k6-3-processors/2

Thanks... This raises more questions

I have the K6-III down as 1999, with the K6-II down as 1998. Your link talks about the K6-3D which I think is a different product. I will keep digging for more information.

 

[Chuck(G)]

The TMS1000, for example, was very much a microcontroller, as it was mask-programmed as manufactured. I had a microwave oven that used one.
To bolster your point about embedded MCUs, however, I note that my 1992 F150 ECU uses an Intel microprocessor that I'd never heard of--the 8065. Apparently, Intel made a raft of ICs dedicated to automotive engine control.

 

[kuro68k]

The TMS1000, for example, was very much a microcontroller, as it was mask-programmed as manufactured. I had a microwave oven that used one.
To bolster your point about embedded MCUs, however, I note that my 1992 F150 ECU uses an Intel microprocessor that I'd never heard of--the 8065. Apparently, Intel made a raft of ICs dedicated to automotive engine control.

Yeah, Intel are terrible for that. I mean most microcontroller manufacturers were very proprietary in terms of architecture and tools, until ARM Cortex came along, but Intel were one of the worst.

 

[kuro68k]

A little update. Colours and layout improved. Unless anyone has any solid information I think it's close to complete now.

 

[pbirkel@gmail.com]

Nice chart! I notice that you have the DEC F-11/T-11/J-11, but don't start with the original LSI11? Which actually relates to the WD16 above the F-11, if I understand what you mean by the WD16. See this page for more info: https://gunkies.org/wiki/LSI-11_chip_set

And developmentally IMO the T-11 is a branch rather than being on the direct development path from the F-11 to J-11. Some info: https://gunkies.org/wiki/T-11

The entire family of DEC VAX VLSI implementations is missing as well, but I appreciate that you have to draw the line somewhere :->! A good place to start might be: https://en.wikipedia.org/wiki/DEC_V-11

 

[kuro68k]

Thanks, I think I misunderstood the MCP-1600. It's a flexible multi-chip CPU system, with a ROM for the microcode. As such it is the a building block that other CPUs are made of, rather than a CPU itself.

The WD16 seems to have been a reference implementation of the MCP-1600 chip-set. Alpha Microsystems used it in their AM-100 computer.

The WD16 implementation was modified into the WD-900, for the Pascal MicroEngine. While it was the same chip-set, the microcode was different so it's effectively a different CPU.

The LSI-11 was another modification of the WD16, to implement the PDP-11 instruction set.

Is that correct? Please check the attached file.


As for the VAX VLSI, my rule is to only include CPUs used in microcomputers, so no minicomputers or mainframes. There is the VAXstation series of microcomputers, but did any of them use a single chip CPU? Normally I don't include chip-set CPUs, with exceptions made where they are historically significant which generally means they are predecessor of a single chip CPU.

 

[pbirkel@gmail.com]

Hmm. What is the meaning of a dashed line vs. a solid line? Ought to put that in a legend somewhere.

The fun fact about the WD1600 (etc.) is that internally it's an 8-bit design that iterates to produce a 16-bit interface. Three chips get you an operational capability: CONTROL, DATA, MICROM. Read more starting at the bottom of page 22 in: https://archive.computerhistory.org/resources/access/text/2019/07/102738263-05-01-acc.pdf Read forward from there for a history of the F-11, T-11, J-11, and on into the VLSI VAX. Lots of grist for your mill :-}.

And more in: https://archive.computerhistory.org/resources/access/text/2019/07/102740181-05-01-acc.pdf

By your definition the https://en.wikipedia.org/wiki/MicroVAX_78032 is a single-chip CPU.
But I don't think anyone at the time would call it a "microprocessor" or the MicroVAX II a "microcomputer".

 

[DeltaDon]

The K6-IIIE+ (and the reduced cache IIE+) had "PowerNow" power management with speed stepping besides increased cache and the other improvements due die shrinkage. It was designed to be a low power laptop CPU and I'm sure that it was not a completely new core, but was modified with these improvements plus used smaller die size. I retrofitted these into laptops using Intel or AMP CPU's and used a modified BIOS to turn on PowerNow and for normal operations the CPU cooling fan ran only when max'ing out the CPU operations and battery life went from perhaps 1.5-2 hours to nearly 6 hours for normal use of running Windows with office use while also increasing computer through put too. It was too late with its introduction date to make much of splash as the ODM's as they were switching over to the newer Athlon and Pentium III CPU's as about the same time plus they were never released to to the general public for upgrading existing motherboards. I got my hands on some that found there way into the gray market as ODM's dumped their inventory, I guess.

Spec sheet found here: https://en.wikichip.org/w/images/3/...ed_Processor_Data_Sheet_(September,_2000).pdf

 

[seaken]

Very nice chart. It helped me learn more about the CPU's that were contemporary with the Intel CPU's.

I would change the little text next to the 8088 and 80286. You should most definitely have "IBM PC" next to the 8088, not just the PCjr. Maybe even write it as "IBM PC (5150)". The 80286 should have "IBM AT" or "IBM AT (5170)".

I also don't think I would bother with PS/1 as it was not a first. It also started with the 80286, not the 8088. Maybe PS/2 could be in there on both 8086 and 80286 but no need for PS/1.

Seaken

 

[kuro68k]

A little update based on feedback, and grouping all the Motorola parts.

 

[tofro]

A little update based on feedback, and grouping all the Motorola parts.

You're still missing the Phillips 68070, which was the CPU of the Phillips CD-i console (and not a derivate of the 68060, but rather the 68000 plus some on-chip peripherals.)

 

[kuro68k]

You're still missing the Phillips 68070, which was the CPU of the Phillips CD-i console (and not a derivate of the 68060, but rather the 68000 plus some on-chip peripherals.)

Hmm, it's billed as a microcontroller, but as you say it was used as a CPU. I wonder if the microcontroller features are used in the CD-i at all. I2C could be useful for system management stuff.

I have a date of 1990 from the datasheet. Probably right, they likely developed it for the CD-i and then tried to sell it for other uses.

I don't have room to put the Philips logo in the key at the bottom. I'll think about how to move things around a bit, although it does say "Philips" in the logo itself.

 

[tofro]

Somehow I don't quite get why you have marked the SM83 with "Super GameBoy". That chip drove the original GameBoy as well, which I consider much more important.

 

[tofro]

And in the ARM family you might want to add the NVidia Tegra which powered the Ouya console.

 

[kuro68k]

I'll fix the GameBoy reference. I think the Tegra is excluded by just being a derivative of the main ARM design, and by being an SoC.

 

[Wildfire]

@kuro68k

I just came across this thread...

First of all



for your work !


Second a have some addendums:

You have the AMD AM2900 also in the list, only mentioned for Atari.
These ones have been used to build a huge load of different systems and/or controllers.

F.e. Data General used those to build their NOVA4 16 Bit CPUs with these.
DEC used them for add-on high-performance floating point units for the 11/34 /44 /23 systems.
I also have 3rd party Qbus controllers for PDP11 systems here which also did use am29xx chip to make their own cpus out of them,
sometimes bc it was cheaper than to buy f.e. an intel cpu; sometimes bc they had experience with those bit-slice things and already written software for other controllers with bit-slice.

Ken Shirriff wrote some information about those

Inside the Am2901: AMD's 1970s bit-slice processor

You're probably familiar with modern processors made by Advanced Micro Devices. But AMD's processors go back to 1975, when AMD introduced t...

www.righto.com

The Surprising Story of the First Microprocessors

You thought it started with the Intel 4004, but the tale is more complicated

spectrum.ieee.org

My opinion is also that there must be an Alpha section...
Why ?
They are history to remember:

There have also been workstations from the beginng to run winNT workstation,
f.e. the Alphastation 200 4/100, Personal Workstation PWS433 / 500 / 600, Compaq Professional Workstation XP1000
Once also Vobis Computer did try to sell it to the masses with an EV5 style cpu, also with NT.

One main reason is also that AMD used technology in their first gen Athlon CPUs from 500-1000MHz
the Alpha EV6 Bus protocol and also the floationg point units.
They hired / bought a lot of former Alpha developers and bc of that they were able to kick Intels asses at that time very hard and deep...

Also Alpha was used in supercomputing f.e. the Cray T3D and T3E systems of just with clusters of 2 or 4 CPU servers at Compaq/HP
- this was called Alphaserver SC (SC for supercomputing) 20 or 40 or 45 ( the used Alphaserver models) with Quadrics interconnect.
In its heydays operating in the top5 of the TOP500 supercomputing lists.
There are also VME bus style systems for real time applications.

I know you want to focus on workstation style, but just fyi