The largest computer

I bet, most don’t know what the largest computer is. Is it really Q7 of SAGE (pictured)? No. It is…

Maybe you can win any bet with this.
The largest computer is: The universe. No joke, but it’s a theory, first by Konrad Zuse in 1945 later also by Nobel prize awarded Gerard t’Hooft. And that is more retro than the Antikythera Mechanism

Today I’ve seen “Through the Wormhole with Morgan Freeman”, episode “What is Nothing?”. Very interesting, mainly about black holes. Not sure what the exact name of this theory was called in the series, maybe the Black hole information paradox or maybe the Holographic principle. Stephen Hawking was wrong first. Black holes change when swallowing a planet, what consists of bits.

For the second largest computers (pictured)

Another question is, what is the smallest computer? But that also depends what you call a computer.

1 Like

Ha ha - doesn’t it always! That’s the problem with looking for superlatives…

Yes, but there isn’t definitely any larger one.
But you can’t say the largest of the world.

Maybe well worth mentioning: It’s because of an extra / spare AN/FSQ-7 that ARPA IPTO (Information Processing Techniques Office) was founded, with Licklider in the lead, who eventually started the Intergalactic Network program. So we’re reading this – at least to some degree – because of the AN/FSQ-7. Technically, many of the networking features eventually developed relied on the technology of the MIT computers that followed Whirlwind, and the DEC machines, which came after this. (Even the code for the IMPs was developed and cross-compiled on a PDP-1, which has a consistent line of heritage back to Whirlwind, and later interconnecting network stacks ran on PDP-11s, hooking up the larger mainframes to the ARPANET via ELF.)

(Also, you got two Whirlwind II in one with the AN/FSQ-7 – quite a bargain at $238 million, and it even came with a 3MW PSU, not too shabby. :slight_smile: As compared to the universe, this is quite an occasion.)

A slight variant… Computers vs. Clusters ?

Today, you can put a few 1000 boxes in racks in a data centre (or many DCs), connect them together with something fast and you have something physically large, but is it “a computer” or “a cluster of computers” ?

For something to ponder over, here is a photo of something I helped build in the early 90’s:


So that’s a custom frame with boxes containing boards. There is little indication of scale there, but the big “L” shaped unit is 2m high. The black boxes are RAID units which were latterly incorporated into the frame.



That still leaves the question, is the the EARTH a computer?
Only the mice know !
Are we talking size of boxes or size of memory for a single core cpu
for the largest computer? Bigger is some sort of data center,
or cloud setup.
The universe may be digital, but not a computer.
A computer requires memory, the unverse has none.
In fiction I favor the Multivac, a self-adjusting and self-correcting computer
from The Last Question.

1 Like

This isn’t a retrocomputing answer, but I did wonder recently if the Starlink constellation might be considered as a large spherical shell of a computer. If it is, then it’s bigger (volumetrically) than the Internet, which might also be considered as a world-spanning computer. And the Internet is respectably old now, if not retro.

But then yesterday I had another thought: perhaps the GPS constellation and its ground stations could be considered a computer - in which case it’s a great deal bigger. (Certainly digital messages go up to GPS satellites and others come back down. But I’m not sure if that’s going to be enough for me to call it a computer.)

GPS is not only respectably old, it was respectably large rather early on. But I’m still not sure if it’s retro!

And then if we think of sending digital messages, having some remote computation go on, and getting digital results, perhaps the communications to the Voyager and Pioneer satellites gives us a very large computer, as measured in linear span rather than in volume.

From the strongly recommended Lisping at JPL and relating to 1999, and the Deep Space 1 probe:

Debugging a program running on a $100M piece of hardware that is 100 million miles away is an interesting experience. Having a read-eval-print loop running on the spacecraft proved invaluable in finding and fixing the problem. The story of the Remote Agent bug is an interesting one in and of itself

1 Like

@NoLand, Aren’t you thinking of the [AN/FSQ-32](

You are right about the AN/FSQ-32, but not about me thinking of it. I got seriously confused…

@NoLand, I’m confused about your confusion. Can you please explain about the Q7 and Q32 and IPTO and what you were getting at?

Well, at the time of writing, I actually thought that the Q32 hadn’t been ready then to have produced a spare unit of no apparent use. On the other hand, the Q7 had quite extensive telecommunication capabilities (the experience with which eventually gave IBM a head-start into the logistics business, like airline reservations, the TOS train management, etc) — and there had been lots of it. It seemed to fit and I didn’t give it a second thought. I just recalled that IPTO was founded on that spare unit (probably, the cold war thermonuclear bomb shock that inspired ARPA provided another connection to the Q7). As a bonus, this produced the connection between the physical large machine that was the Q7 and the even larger virtual machine that is the Internet, which seemed to be worth mentioning.
(Well, I’m not an historian… and, apparently, however disappointing this may be, sometimes I do get things wrong, based on a bright conjecture.)

Thanks to @pmcjones for putting this right!

As ARPA intended when they provided the Q32 to SDC in Santa Monica, California, several research projects were carried out with the machine, including a timesharing system [1], [2], early networking, and the LISP2 project [3], [4]. I was able to track down and scan many of the documents and listings from the LISP2 project. [5]

[1] Jules I. Schwartz, Edward G. Coffman, and Clark Weissman. 1964. A general-purpose time-sharing system. In Proceedings of the April 21-23, 1964, spring joint computer conference (AFIPS '64 (Spring) ). Association for Computing Machinery, New York, NY, USA, 397–411.

[2] Jules I. Schwartz and Clark Weissman. 1967. The SDC Time-Sharing System revisited. In Proceedings of the 1967 22nd national conference (ACM '67 ). Association for Computing Machinery, New York, NY, USA, 263–271.

[3] Paul W. Abrahams, Jeffrey A. Barnett, Erwin Book, Donna Firth, Stanley L. Kameny, Clark Weissman, Lowell Hawkinson, Michael I. Levin, and Robert A. Saunders. 1966. The LISP 2 programming language and system. In Proceedings of the November 7-10, 1966, fall joint computer conference (AFIPS '66 (Fall) ). Association for Computing Machinery, New York, NY, USA, 661–676.

[4] P. McJones, “The LISP 2 Project,” in IEEE Annals of the History of Computing , vol. 39, no. 4, pp. 85-92, October-December 2017, doi: 10.1109/MAHC.2018.1221040.

[5] P. McJones, History of LISP project, LISP2 section, Software Preservation Group.


Here’s a paper that discusses the networking research involving the Q32:

David Hemmendinger
Two Early Interactive Computer Network Experiments
IEEE Annals of the History of Computing
July-Sept. 2016, pp. 12-24, vol. 38


Thank you, @NoLand and @pmcjones for bringing more light on this. I’m intrigued LISP 2 source listings are available!

OK. When the earth is not really the largest computer, what about NVIDIAs Earth-2 Supercomputer simulating the complete earth.
I’ve seen a TV documentary about digital twins. It’s not complete, yet, but there are other digital twins of other planets.

By coincidence, I found the world’s smallest computer as of 2015 Michigan Micro Mote M3

For another sense of smallest computer, we could consider the early Kenbak-1 (no microprocessor!) or the later Cambridge MK14 - both with 256 bytes of RAM. Both with I/O too, in a minimal sense, but not self-contained for power.

I was seriously tempted by the MK14 at the time!


I always figured the 50s era ATT switchboard network was technically a single large computer due to ingesting data, and modifying that inload to produce outload.

Isn’t this the technology that Shannon used for his Theseus mouse-in-a-maze?


We previously unearthed this link on Shannon’s machine: