I'm incredibly pleased to announce that the microcode for the Intel 80386 has been decoded.
-
Here's a little bit of banana for scale to appreciate how tiny the features we're working with are.
I have this acrylic keychain that has an actual 386 die in it.
It's Today's Choice, you see. (The rear side has a 486 die, with "Tomorrows Vision" labelled above it, something that I will never get tired of reading. Oh my god, I'm so old.)
@gloriouscow
Hot damn I love that thing, I own lapel pins of Intel and IBM dies but wasn't aware of that one. Totally jealous! -
@ask Further complicating matters is that once you've decapped a chip like this unless you have some sort of professional-grade clean-room filtration setup you've pretty much destroyed it because of all the microscopic schmutz floating in your average air that will get in and start bridging microscopic traces everywhere.
@gloriouscow @ask Not true if there is a remotely decent passivation layer over the top insulating it.
Of course, I have no idea if these older parts actually have exposed metal anywhere... but on anything like 350nm and newer there's gonna be a micron or two of nitride/oxide on top insulating the die surface and protecting it from physical damage. The top metal traces aren't directly exposed.
-
@gloriouscow@oldbytes.space how long until the 486? that's the one i am most interested in
(only half joking...)@linear
A 486DX2 in FPGA form would be awesome. I cannot start to fathom the amount of work needed to bring that to life
@gloriouscow -
There is a way to extract the contents of an implant ROM. The doping that creates the gates means that you can etch the silicon in a way that the doped areas will stand out.
The acids involved in this process are some of the nastiest chemicals on the planet. Stuff like hydrofluoric acid.
Oh, you spilled it on yourself? no big deal. It's just going to dissolve your bones.
@gloriouscow I have wanted to develop a Dash etch process in my lab for a while and might be willing to make an attempt at some point, but am way too busy to try it in the near term.
I have a proper wet bench with (dilute) HF and nitric acid already, I don't think I have acetic in stock but that's easy for me to procure.
But I have little to no experience working on chips that ancient so it may take me a while to even get decap etc working with good clean results. 350nm is about as old as I go these days.
-
There is a way to extract the contents of an implant ROM. The doping that creates the gates means that you can etch the silicon in a way that the doped areas will stand out.
The acids involved in this process are some of the nastiest chemicals on the planet. Stuff like hydrofluoric acid.
Oh, you spilled it on yourself? no big deal. It's just going to dissolve your bones.
@gloriouscow if you have a rough traveler I can do it for you
-
I'm incredibly pleased to announce that the microcode for the Intel 80386 has been decoded.
It was an group effort by a bunch of talented people to extract and correct the physical bits, but the major work of decoding them was done by reenigne - you may know him from such incredible PC demos as 8088 MPH and Area 5150, as well as being the person who decoded the 8088 microcode previously.
Please, check out his writeup.
#retrocomputing #vintagecomputing #microcode #reverseengineering
@gloriouscow@oldbytes.space That's really impressive -
@gloriouscow if you have a rough traveler I can do it for you
@gloriouscow if you have a clean die that would be preferable to having to decap one though
-
@gloriouscow @datenwolf The pitch between vertical lines on that image is about 5.7 micron.
This places this straight into OCM (optical coherence microscopy) territory. I'm a little bit concerned that speckle could cause issues. But just this Thursday I was in a seminar, in which a couple of ideas for speckle reduction were presented. Worth a shot.
-
This places this straight into OCM (optical coherence microscopy) territory. I'm a little bit concerned that speckle could cause issues. But just this Thursday I was in a seminar, in which a couple of ideas for speckle reduction were presented. Worth a shot.
@datenwolf @gloriouscow I'd be very interested to hear about any results you get!
-
@datenwolf @gloriouscow I'd be very interested to hear about any results you get!
I'll need a decapped 80286 first. I think I might have a few AMD made 286 somewhere in the basement, but I have zero experience in decapping.
-
One of my early experiments in OpenCV produced an unintentional piece of Microcode Art I'm still fond of.
This was a result of attempting auto-segmentation using incrementing hue on the various segments. Needless to say, a lovely disaster.
@gloriouscow oh really neat
-
I'll need a decapped 80286 first. I think I might have a few AMD made 286 somewhere in the basement, but I have zero experience in decapping.
@datenwolf @gloriouscow I think I can provide a few. At least I have the one that was imaged, and I could also try to get you one more with the die stripped down to substrate.
-
@datenwolf @gloriouscow I think I can provide a few. At least I have the one that was imaged, and I could also try to get you one more with the die stripped down to substrate.
If I can help coordinate this in any way left me know. I'd be happy to reimburse you for shipping things - although through some other means besides Ted Pella
-
@datenwolf @gloriouscow I think I can provide a few. At least I have the one that was imaged, and I could also try to get you one more with the die stripped down to substrate.
You can find my OrcID in my profile bio. And this is our research institute's webpage
-
It doesn't stop there. The incredibly talented nand2mario has taken reenigne's work and created a microcode-level Verilog implementation of the 80386. And yes, it runs DOOM. There's even a MiSTer core in the works.
https://nand2mario.github.io/posts/2026/z386/
https://github.com/nand2mario/z386
https://github.com/nand2mario/z386_MiSTer@gloriouscow I’m so excited about a real 386 MiSTer core!
-
@gloriouscow @ask Not true if there is a remotely decent passivation layer over the top insulating it.
Of course, I have no idea if these older parts actually have exposed metal anywhere... but on anything like 350nm and newer there's gonna be a micron or two of nitride/oxide on top insulating the die surface and protecting it from physical damage. The top metal traces aren't directly exposed.
true, the amount of delayering you need to get images of things isn't necessarily required if you were just going to try to probe things.
i still think that fact that everything is incredibly tiny is the real problem.
-
true, the amount of delayering you need to get images of things isn't necessarily required if you were just going to try to probe things.
i still think that fact that everything is incredibly tiny is the real problem.
@gloriouscow @ask well what i meant is more, dust on the top of a decapped chip with intact bond wires and passivation isn't really going to hurt it. I've had chips sitting on a shelf in a cabinet for years open cavity that still run fine.
-
@gloriouscow @ask well what i meant is more, dust on the top of a decapped chip with intact bond wires and passivation isn't really going to hurt it. I've had chips sitting on a shelf in a cabinet for years open cavity that still run fine.
how does heat dissipation work when a chip is decapped? I would have assumed a CPU would immediately overheat
-
how does heat dissipation work when a chip is decapped? I would have assumed a CPU would immediately overheat
@gloriouscow @ask Normally I'm working on embedded parts that don't generate enough heat for this to be a problem
-
@gloriouscow if you have a clean die that would be preferable to having to decap one though
@gloriouscow https://siliconpr0n.org/wiki/doku.php?id=delayer:dash
Ok yeah if it doesn't require any heating I can run this on a delayered die next time I'm working at the acid bench in the cleaneoom
