Before Intel introduced the 4004, a secret processor was already controlling the wings of America’s most iconic fighter jet.
In 1970, Ray Holt and a small team at Garrett AiResearch completed the MP944 chip set for the F-14 Tomcat’s Central Air Data Computer. Built with parallel processing, pipelining and real-time self-testing, this forgotten machine was years ahead of its time—but its design remained classified for 28 years.
This is the hidden story of the engineers who built a revolutionary flight computer, the Navy project that kept it secret and the debate over whether the MP944 deserves to be called the first microprocessor.
Was the MP944 truly first, or does Intel’s 4004 still deserve the crown? Share your opinion in the comments.
Subscribe for more forgotten stories about the parts, inventions and engineering decisions that changed American history.
#F14Tomcat #MP944 #ComputerHistory #AviationHistory #MilitaryTechnology #documentary #documentaries #history #viral #viralvideo #video #siliconvalley
source
The MP944 deserves far more recognition! 🚀 The Intel 4004 helped popularize the microprocessor, but the MP944 was tackling real-time flight control challenges years earlier. Does being first mean inventing the technology first, or bringing it to the world first? That’s what makes this debate so fascinating!
AI generated Script horrible description of everything, BS video clips, BS Explanation; BS Video 👎
Thank you for explaining the MP944. Yes it should be first, The f14 is one of my favorite fighter jets.
The WOs on the A-6 said that the gears in the aiming system responded well to a swift kick
The first DRAM shows up with Intel's 1103 in 1971. Prior to that memory was either 'core' or 'shift registers'. Core is slow and temperature sensitive. Shift registers access a particular memory location once in some number of cycles – the number of cycles needed to shift the value through all the registers. The Datapoint 2200 Version 1 used 512 bit shift registers, so a memory access could take as many as 512 shift cycles to access a particular address.
If the MP944 used solid state random access memory, it would have been just a few bytes, or a few 20 bit words – nothing measured even in kilos.
Having a multiplier chip, and separately, a divider chip is certainly impressive for the time. These are computationally expensive operations, particularly as word lengths exceed 8 bits. The Intel 8080 needed a separate math chip to offload multiply and divide functionality (Intel 8087). However, the 8080 could do 32-bit multiplication in code, if need be.
This is kind of where the pedal hits the metal – could the MP944 do multiply, divide, and square root operations in code using nothing more than add, subtract, compare, and conditional flow-of-control instructions? If that answer is no, then the MP944 is really more of an ASIC (Application Specific Integrated Circuit) then it is a general purpose CPU. It is still an impressive accomplishment, but it fails the general purpose CPU test.
The A7d used a IBM computer with a core memory and was the first fighter with a full combat HUD. This system predates the F-14 and the last time a gear driven system was used was WWII 1950s junk. I worked on wild weasel systems used on F-4 phantoms and these were full custom gate array chips that were compute systems way before the F-14. Some of the things you mention in this episode are not true. The F-111 used an advanced IBM 360 core that was way older that the F-14…..so yes the F-14 was more advanced as a beneficiary of those earlier aircraft not solely based on the MP944 chip……
18:42 NOOOOOOOOOOOOOOOOOOOOOO! Bent pins!
There was also a fluidic (hydraulic) backup computer. Reduced functionality but enough to fly on once the nukes fried the electronics.
Well, the P944 was not a microprocessor in the sense that we mean it today, it was a chip complex, it took several chips together to form what we call a processor today.
The 4004 was not a lone ranger either completely, but the 4000 series processor complex required 4 separate chips to really work:
the Intel 4001, a 256-byte 4-bit ROM;
the Intel 4002, DRAM with four 20-nibble registers (total size 40 bytes);
the Intel 4003, an I/O chip comprising a 10-bit static shift register with serial and parallel outputs; and
the Intel 4004 CPU.
The 4004 WAS the CPU in this setup, the others were required to make it function but were not part of the processor logic strictly speaking. When we take a look at the Mp944 there is a quite different story – a RADICALLY different story in that 4 chips with 1 being ROM, 1 RAM, 1 I/O and 1 CPU still makes sense in the context of how a PC is organized today, we have storage taking the place of the ROM, we have RAM, we have a 1 or 2 chip I/O (called chipset colloquially) and a CPU.
The MP944 or F14 CADC as it is correctly called, was comprised of no less than 28 chips, not all of them part of "the processor bits" in that 19 of those chips were ROMs, 3 of them were RAM(called RAS, Random Access Storage in the description of the system) – but then we get to the juicy bits, which comprise of 3x Steering logic unit, SLU, one each for the 3 Computation Units that were:
Parallel multiplier unit PMU
Parallel divider unit PDU
Special logic function SLF
So to kinda simplify it a bit, the 3x SLU + PMU + PDU + SLF is a rough equivalent of the Intel 4004 – not exactly but we cannot fully compare them exactly, this is as close as we can get it.
SO – all of this has a point which is what I started this comment with, the reason people talk about the 4004 as the first microprocessor is because it arguably was – in a way that the MP944 has nothing to do with. The "micro" part of the "microprocessor" is meant to be a reference to it being a processor integrated into 1 thing. 1 microchip. Even if the MP944 has been celebrated on TV, radio and in the newspapers when it was completed, it still would never have been a "microprocessor" unless someone at Garrett and presumably AMI has managed to integrate all 3 SLU's + PMU, PDU and SLF into ONE chip – then the 4004 would STILL have been the first microprocessor.
What made the 4004 special was NOT that it did computations, but that it did what it did in ONE chip. NOT that it did what it did. There were many systems before the MP944 that did computations using silicon chips, it's just a question of HOW MANY chips were required to do the computations. The 4004 is a slightly arbitrary pick perhaps but it IS a single unit that DID multiple types of computation and logic on one single chip.
They could have used an ANALOG computer,
they do division, square root, …
Trouble is precision. Analog isn't 16 bit. You don't even need trans-linear circuits.
Met Mr. Ray in Mississippi, he was my daughter's robotics coach and teacher. Consider him a good friend.
The 4004 had an adder, but no multiply or divide circuits. If it had them then they would have also been on separate chips.
5:10 Look at all those Gooses!
The airframe on the lower right is a bit curious; otherwise this was almost a pretty good ad hoc AI generated photo.
Nice video 👍
More of a asic rather than general purpose CPU.
So this was an air data computer.
The sense that this is a microprocessor is the same sense where anythng that computes is a computer. Though this was a breakthrough for creating a digital flight computer where these have been principally mechanical computers, the argument of whether this is the first microprocessor is sort of akin to whether the eniac is the first computer. The eniac was predated by several mechanical programmable digital computers and several electronic hybrids that were purpose built for things like code-breaking. So the eniac is there for defined as the first fully electronic and general purpose programmable computer. Similar is the comparison of the MP 944 with the 4004… The MP 944 is a purpose-built computer, All digital. And a marvel of a flight computer. Here's a description of how it worked:
The Garrett AiResearch MP944 chipset—used in the F-14 Central Air Data Computer (CADC)—did not feature a traditional general-purpose CPU instruction set like commercial processors of the era. Instead, it used a custom hardwired microcode and parallel pipeline architecture designed specifically for real-time aerodynamic equations.
Again this is akin to the purpose-built digital computation devices of world war II. If this device had been unclassified it would have been unsuitable for any other uses than a flight data computer.
This contrast with the Intel 4004 which had a general purpose instruction set and could be used for many different applications.
It is akin to the eniac, which differentiated itself by being fully digital fully electronic and fully programmable.
So the MP 944 and the Intel 4004 just aren't even in the same class.
Both are multi-chip implementations but that isn't really the point.. the Intel 4004 was the first commercial device that everyone could imagine using in different control applications.
The MP 944 even if had been made public would not have served that purpose and wouldn't have been interesting in computer history really.
The sad part is is that the aeronautics industry continued to use electromechanical flight data computers for decades when it could have shifted to microprocessor based flight data computers.
Side note, the A-5 Vigilante actually had the first Fly-By-Wire system, a Nav/Attack system with heads up display, radar, radar equipped inertial navigation, and a closed-circuit television camera in the nose. All of it was run by an early digital computer called the Versatile Digital Analyzer. The A-5's first flight was in 1958! Not much out there on this computer unfortunately.