Hacker News new | past | comments | ask | show | jobs | submit login
Xerox Alto Restoration Part 5 [video] (youtube.com)
90 points by kens on Sept 5, 2016 | hide | past | favorite | 13 comments

Logic analyzers connected to pins on a CPU. You don't see that much any more. The connections they're looking at are inside the chip today. Of course, now we often have JTAG access to the innards.

Let me emphasize that there's no CPU chip in the Alto - the CPU is made up of dozens of TTL chips on three different boards. We're tracing the individual microcode instructions that the Alto is running. As Animats points out, this would be deep inside a modern chip.

There's a brilliant talk, given in 2009 at HOPE09, NYC titled "Indistinguishable From Magic: Manufacturing Modern Computer Chips".

At roughly 36 minutes, there's a cross-section of a (in 2009) modern chip. The topmost layer is the shiny rainbowy topmost metal-layer you see on typical (from the top) die photographs. The very, very small comb-looking bottommost-layer is the individual transistors, and structures there are of the size that give a "xxx nm" process its name.


I very much recommend everyone to watch this video, to get a sense for the complexity that goes into producing a modern CPU.

To come back to the thread I'm answering here: The logic analyzer on the Alto (probing microcode) connected to interconnects of individual gates probably would be connected to signals somewhere in the lower middle of the stack in a modern CPU, I guess.

What I'm personally afraid of is that, due to the bad power supply which was repaired in restoration part 1, the wrong voltage was applied to one of the power pins of several boards full of DRAM chips, and this may have fried the entire lot of them.

If the Alto2 has 128KiW of dram, and the drams used are old tms4108 chips(I'm not actually sure about this, they may be even older), and there are 18 or 19 bits of ram per word (16 bits plus 3 parity/ECC? Again, not sure.), then there could be as many as 304 chips to replace (or at the very least, to pull and test each one and replace the bad ones).

Fortunately, if they used triple-voltage (5v, 12v, -5v) 4108 or 4116 drams, you can replace those chips with the 5v-only 4164 chip by bending 2 pins upward on the 4164 and soldering a wire to another pin. This isn't "period accurate" but should work to replace a few bad chips until working ICs can be found.

See http://picmania.garcia-cuervo.net/recursos/redpicdatasheets/...


We tested and repaired the power supplies before powering up the circuit boards, to make sure we didn't fry anything.

Each of the four boards has 80 memory chips. These are 4116 DRAM chips, each holding 16 kilobits. Each board holds 128K 10-bit-wide "chunks". Data is stored as 32 bits + 8 bits (actually 7) of Hamming error correction and parity.

The issue is not when you powered up the machine and the ram boards, its more whoever powered them up before you got the machine.

That power supply may have failed decades ago. Was the machine known working when it was last powered up?

Is there any way you could build a rig to test each 4116 chip, or a rig you could plug each board into which would supply proper voltages (in the correct bring-up and bring-down order, because 4116 drams are weird like that) for the dram, and do a pattern test on the entire board all at once?

Personally I don't like 4116 drams at all because they're very unreliable. (4164s are far superior in this regard.) See the PARC notes on bitsavers re: the notetaker project, the notetaker test systems were frequently out of commission due to failed 4116 dram chips, or running with one or more chips dead and the ECC constantly correcting the errors.


Where's the text writeup? The previous 4 parts had one.

As with the previous parts, the text writeup will be a few days after the video. I'm not as fast at writing as Marc is with video editing :-)

The text writeup on the Alto restoration is now at https://news.ycombinator.com/item?id=12437018

Very cool.

[The http://youtu.be submissions are autokilled. Remember to submit the http://youtube.com variant next time.]

Was the shortener used for spam submissions at some point?

We've updated the URL on this one, too.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact