[HN Gopher] Two dies in one package: Teardown of a vintage ROM w...
___________________________________________________________________
Two dies in one package: Teardown of a vintage ROM with double the
storage
Author : parsecs
Score : 47 points
Date : 2021-03-22 17:34 UTC (5 hours ago)
(HTM) web link (www.righto.com)
(TXT) w3m dump (www.righto.com)
| kens wrote:
| Author here if anyone has questions about vintage IBM integrated
| circuits :-)
| marcodiego wrote:
| Does the metal can packaging make the chip less vulnerable to
| static or fields? I mean, does it work somewhat like a Faraday
| cage?
| spaceribs wrote:
| Really cool! I assume it would be possible to take the pictures
| you have, turn them into 1s and 0s via some sort of image
| processing, then pop out actual code?
| kens wrote:
| Yes, it would be straightforward to extract the data from
| these ROM images. (See the "Pulling Bits from ROM Silicon Die
| Images" article currently on the HN front page.)
|
| I extracted some of the data from the ROM, but it wasn't
| anything obvious. I didn't spend a lot of time examining the
| data since it's probably something like microcode for an
| undocumented IBM control processor. The extracted data did
| let me verify that the 9 bits were 8 + parity.
|
| To extract the bits, I made a simple JavaScript program where
| I click on the transistors. For a small amount of data, this
| is faster than performing image processing and setting up
| automated analysis software.
|
| One problem is that once you get the bits, there are a lot of
| uncertainties. Is a transistor a 0 or 1? What order are the
| bits in? Are the addresses left to right or right to left? So
| you end up with a bunch of cases you need to consider, rather
| than a nice hex file.
___________________________________________________________________
(page generated 2021-03-22 23:01 UTC)