Does tools for automation analysis image decapsulated IC for extract netlist exist ?

  • \$\begingroup\$ I'm sorry, but this question is very hard to understand. I don't know what you're asking, but I suspect the answer is no. \$\endgroup\$ – Hearth Nov 1 '18 at 18:42
  • 1
    \$\begingroup\$ I think the answer is: no, it is not going to be a fully automated process, I expect that a lot of manual labor is still needed. As technologies shrink this process will get more complicated and difficult as well. Also see: perso.univ-st-etienne.fr/bl16388h/salware/Bibliography_Salware/… \$\endgroup\$ – Bimpelrekkie Nov 1 '18 at 18:50
  • \$\begingroup\$ This guy comes to mind. He decoded the information from a ROM using delidding, photograph and own-written automation software: youtube.com/watch?v=ex_dEEh4dl0 \$\endgroup\$ – winny Nov 2 '18 at 8:39

I'm not aware of any publicly available tools of this sort.

Projects like Visual 6502 and Project 54/74 have relied almost exclusively on manual annotation of images. The older parts that they're analyzing frequently make use of unusual circuits and silicon structures. These features are frequently unique to the part, and are difficult for even a human analyst to resolve even to a transistor-level netlist. Any sort of automated analysis would be futile.

(One example of an unusual design of this sort is the multi-level ROM in the Intel 8087 FPU. It's not even clear how this design would be represented in a netlist, as it is effectively an analog circuit!)

Newer parts are more likely to use a standard cell library which could, in theory, be recognized by an automated tool -- but those parts are also more likely to be fabricated on a smaller lithography process, making their features difficult or impossible to resolve with optical imaging.

There have certainly been some attempts to construct private and/or task-specific tools to aid reverse engineering efforts, but those tools are typically much more limited than what you're probably imagining. You certainly can't throw a couple of die photos into them and expect synthesizable HDL to come out the other end.


Not the answer you're looking for? Browse other questions tagged or ask your own question.