The Intel 4004 Microprocessor Turns 44

mcpublic writes: Today is the 44th anniversary of the Intel 4004, the pioneering 4-bit microprocessor that powered the first electronic taxi meters. According to the unaffiliated (and newly renamed) Intel 4004 45th Anniversary Project web site, they have just re-created the complete set of VLSI mask artwork for the 4004 using scalable vector graphics, and updated their Busicom 141-PF calculator replica aimed at collectors and hobbyists. Included is some interesting historical perspective: Back in the early 1970s, there was no electrical CAD software, design-rule checkers were people, and VLSI lithographic masks were hand-crafted on giant light tables by unsung "rubylith cutters."

just imagine....

[ganjadude]

a beowulf cluster of 4004s!

Re:

[niftymitch]

a beowulf cluster of 4004s!

Or a cluster of Motorola MC14500 based machines.
http://tinymicros.com/mediawik... [tinymicros.com]
http://www.google.co.uk/patent... [google.co.uk]

One could do worse than doodle through a thick engineering pad
and think about how to build a machine around either of these
classics.

Re:

[jfdavis668]

But does it run Linux?

Re:

[ganjadude]

no sorry, os/2 only

Re:

[jfdavis668]

Good thing I keep a copy.

Re:

[Bengie]

Doesn't matter, it'll be back next year http://tech.slashdot.org/story... [slashdot.org]

no electrical CAD software

[Anonymous Coward]

Weird, tell it to IBM that was routing ICs and backplanes in the 1960s by computer.

Re:

[Anonymous Coward]

Are you sure ?

In the 60's, ICs had a few tens of transistors, for op-amps or apollo's famous 3 inputs NOR gate.
Memory were based or ferrite cores.

Maybe they had CAD for PCB routing, but ICs ?

Re:

[viperidaenz]

in the 60's chips topped out at 500 transistors.
LSI technology didn't come along until 1971.
What commercial IC's are you referring to?

Re:

[AK Marc]

Yet you don't name any CAD software from the '60s. That you haven't plainly given an answer, but asserted others are wrong, one can only conclude you don't know what you are talking about.

Re:

[AK Marc]

So a long list of ACs say there was, but no names, no confirmation. Just assertions I'm wrong. What are you, my wife?

Re:

[Anonymous Coward]

1967

https://www.youtube.com/watch?... [youtube.com]

look at some of the stuff hanging on the walls. The problem is that the text references I have about what IBM was doing in the 1960s is on dead trees only.

Re:

[50000BTU_barbecue]

Intel was a tiny company back then, they probably had to do it by hand. Even in the 1980s companies like Commodore had lots of manual input to chip design even though they had CAD.

Re:

[kenai_alpenglow]

Try this link: http://www.4004.com/2009/i400x... [4004.com] ( Lajos Kintli's Intel 4004 netlist/layout analyzer and animated logic simulator with complete MCS-4 masks and schematics (Zip, v2.0: updated 11/15/09)

Re:

[viperidaenz]

Blocked by proxy:
Malware Found: McAfeeGW: BehavesLike.Win32.Dreform.zz - http://www.4004.com/2009/i400x... [4004.com]

Re:

[segin]

So submit a false positive.

Re:

[KGIII]

Wait...

You're using McAfee?

Re:

[viperidaenz]

Not personally, no.

Re:

[jeremyp]

Avast finds nothing wrong. I'm still not unzipping it though.

Re:

[viperidaenz]

It was the first microprocessor you could buy from someone and program yourself.

Prior to the 4004, you made your own chip, with your own instruction set and your own assembler, with your own chip fab.

Re: 4 bit computing at the time of 128 bit computi

[Bing Tsher E]

More likely you would put together a diode matrix with the bootstrap code to start the paper tape reader.

I was writing code for 4 bit microcontrollers less than 20 years ago.

Re:

[unixisc]

Is that 128 bit a VLIW like computer? Used mainly for parallelizing, and knowing that there won't be too much of unpredictive instructions down the pipe to schedule? You could have 32 cores in a single CPU that would run things at the lower frequencies needed.

As for the 4004, an interesting way to mark the occasion might have been to design one on an FGA or even on an Arduino.

Re:

[AHuxley]

If you had the cash as a nation huge systems that could work as early digital systems could be bought into from the private sector.
Electro-optical digital imaging to look down from space, what the public is been told about what could be done in 1962 with the IBM 7950 Harvest https://en.wikipedia.org/wiki/... [wikipedia.org]
1969 with the 'COINS' (Community On-line Intelligence System)
Some of the changes to the commodity microprocessors could finally be seen in the public with ideas like the 1980's BBN Butterfly https:/ [wikipedia.org]

Commercially significant but 2nd fiddle to TTL

[kyubre]

I was one of those kids who built up simple bread board computers using stock standard TTL parts. I learned more about digital machinery in reading about and figuring out how processors work by trying to create my own bits of programmable/sequence-able logic using the astonishingly complete range of commodity TTL parts that where cheaply available in the late 1970s and 1980s.

The 4004 was an important inspiration, but TTL is what launched our pervasive digital age.

Unlike the 4004, it blows my mind how much of the original TTL part library is STILL available.

https://en.wikipedia.org/wiki/... [wikipedia.org]

Re:

[toonces33]

Same here - I remember all of those TTL things.

A few years after this, we got some *very* early 8086 chips on an educational discount. We breadboarded the thing by attaching power, clock, and wired up the memory lines to make it look like it was just reading NOP instructions from ever address, and then watched the address lines to see the thing count up.

Re:

[__aaclcg7560]

I got into electronics as a teenager and even took a few courses in college, but I never got far with it. Didn't have the information and money to get beyond the basics. Fast forward 30 years, the Internet has plenty of information and as senior I.T. technician I got plenty of money. I'm fiddling around with 555 timer circuits and looking up designs for a TTL computer in. Rather than buying a handful of parts from Radio Shack (back then) I'm ordering lots of 100 from Jameco to build up my parts inventory.

Re:Commercially significant but 2nd fiddle to TTL

[kyubre]

Similar story on this end. I learned all kinds of electronics as teenager and then went off, first working on Air Force Radar for a number of years and then transitioning to software engineering as a civilian. A couple of years ago, I got my highly coveted treasure trove of TTL parts trays from my dad. Started playing around again on the same old breadboards, discovered SparkFun, EBay, and rediscovered Jameco.

Seems nobody personally knows much of anything about the 4004 anymore, but Don Lancaster's TTL cookbook is just as applicable today as it was 30 years ago.

Re:

[GrahamCox]

I was the same but with the 4000 CMOS range. OK, it was slower, but it was far less finicky about fanout, fan-in, power supply voltages and general interfacing. If I needed a fast section of circuitry I might use TTL, but CMOS elsewhere. Just so much less fuss. By the mid-80s it had caught up in speed and eventually surpassed it.

CPU-wise, it was 6502 FTW, though the 8051 wasn't bad as a stop-gap until the 680x0 was cheap enough. 4004? A bit before my time, but also pretty hard to use and do much with comp

Re:

[Anonymous Coward]

oh man, excellent post and link! ahhh, the memories.

yeah I built lots of stuff with TTL. I love the 74181, built a CPU with it, used a 74189 for registers. why does a 4 bit machine need 16 registers?!?!! ridiculous instruction set. but I was a kid.

there was an improved ALU, 74381? found the spec, drooled, never found one retail...

Re:

[RockDoctor]

A few years ago - 5 or 10 by memory - there was a story on Slashdot about someone who built a logical equivalent of an 8086 or 80386 or some such 1980s era processor.

This story [slashdot.org] perhaps? No.

Those mask cutters look like Bond girls

[michaelmalak]

Those mask cutters look like Bond girls

Re:

[Mashiki]

Impossible, everyone knows there are no women in technology related areas. Just look at the amount of whining over it.

Rubylith was state of the art!

[Terje Mathisen]

My father spent months at his home-made light table back around 1965 cutting traces in rubylith film in order to create the offset masks for orienteering maps.

He needed one such mask for each color in the finished map, any mistakes had to be fixed with small amounts of red lacquer which then had to dry completely before it could be recut.

The big advantage for VLSI vs a map was that most lines were straight so you didn't need to trace curved lines like you do for the contours on a map.

Terje

In early 1970s, there was no VLSI, not even LSI;

[bsharma]

Back in the early 1970s, there was no electrical CAD software, design-rule checkers were people, and VLSI lithographic masks were hand-crafted on giant light tables by unsung "rubylith cutters." In early 1970s, there was no VLSI, not even LSI; It was MSI. https://en.wikipedia.org/wiki/... [wikipedia.org]

Re:

[avandesande]

It's always fascinating to track the progression of a 'tools', where creating the tool aids in the development of something more advanced. It's obvious that you cannot design an advanced chip without a computer.

Used to service Data General Nova 4X

[advocate_one]

had a 16 bit 'processor' made by using four of these chips in parallel...

Re:

[rrohbeck]

Those would have been bit slices not complete microprocessors, probably 2901s

VLSI

[sunspot55]

The 4004 is VLSI? I don't think it means what you think it means...