16 comments

  • aappleby3 hours ago
    I reverse engineered the Game Boy from pictures of the die - <a href="https:&#x2F;&#x2F;github.com&#x2F;aappleby&#x2F;metroboy" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;aappleby&#x2F;metroboy</a> - so yes, it&#x27;s totally doable. It is also incredibly tedious and frustrating.<p>Not sure if that gives me an advantage in this challenge, but I have too many things to do already. :D
  • zoenolan17 hours ago
    I thought the article was going to be about how people scan chips<p>Rapid Chip Reverse Engineering Using Laser, Focused ion beams, and Scanning electron microscope <a href="https:&#x2F;&#x2F;academic.oup.com&#x2F;mam&#x2F;article&#x2F;30&#x2F;Supplement_1&#x2F;ozae044.314&#x2F;7720268" rel="nofollow">https:&#x2F;&#x2F;academic.oup.com&#x2F;mam&#x2F;article&#x2F;30&#x2F;Supplement_1&#x2F;ozae044...</a><p>FIBs are also used to test modifications before doing a respin. I&#x27;m still in awe that matter can be manipulated so precisely
  • zie1ony18 hours ago
    At my uni, 15 years ago, one postdoc reverse engineered NVIDIA chip and wrote more performant compiler. He did that by connecting oscyloscops to all chip&#x27;s outputs and started with applying random current on inputs. Using ML and his genius he rediscoverd all opcodes including a few hidden ones. Eventually he got hired by some company that was doing a lot of GPU on supercomputers.
    • Aurornis15 hours ago
      &gt; He did that by connecting oscyloscops to all chip&#x27;s outputs and started with applying random current on inputs<p>This is absolutely not how reverse engineering a digital logic ASIC works.<p>Either the story got embellished through retellings, or this person was a fantasist.<p>There are people who hack on GPUs but it’s done at the software level.<p>I did get a kick out of imagining a scene where someone is trying to connect an oscilloscope to a circuit board to reverse engineer the CPU opcodes. That’s like the CSI: Miami version of what this would look like.
      • MomsAVoxell8 hours ago
        Many laymen confuse logic analyzers with oscilloscopes. Don&#x27;t take it personally - the person you responded to is clearly not a native English speaker.<p>You absolutely can reverse chips with logic analysis. It is big business in some parts of the world.
        • i_am_a_peasant4 hours ago
          with very, very expensive logic analyzers... But yeah, nothing weird here. Plus OP was retelling a story of someone else doing it, while probably not being a specialist in this field. So I wouldn&#x27;t take the &quot;random inputs&quot; part literally.
      • RicoElectrico7 hours ago
        Thank you for taking your time to squash another potential &quot;urban legend&quot; in the making. We&#x27;ve got enough of them already in tech.
    • inigyou18 hours ago
      You can&#x27;t do that by applying random inputs to any single-chip GPU - it has far too much state. I can see that perhaps it worked on some of the early multi-chip cards - where one chip was a texture sampler, and so on.<p>You&#x27;ll have more luck reverse engineering the software driver first. They&#x27;re not hidden, you can just open the driver files in Ghidra, the almost-universal tool for open-sourcing proprietary code. Hidden opcodes can be discovered first by just trying all the opcodes you couldn&#x27;t discover any other way. You only need to go to the physical level if they&#x27;re <i>really</i> hidden.
      • sigbottle17 hours ago
        To be fair, Ghidra was released in 2019 and in general knowledge was still hard to find even back in 2010 I feel (well, compared to 2026 in the age of AI)
        • inigyou17 hours ago
          Before that there was, and still is, IDA Pro. Works largely the same but costs a lot, on the order of $1000&#x2F;seat&#x2F;year. Useless for hobbyists unless pirated, but reasonable if it&#x27;s your job. Probably had academic discounts.
          • rzzzt15 hours ago
            Also a few free-but-cut-down versions like 5 (which still recognizes MS-DOS executables). I&#x27;m not going to recommend downloading it from any of the locations mentioned here but it&#x27;s out there: <a href="https:&#x2F;&#x2F;reverseengineering.stackexchange.com&#x2F;questions&#x2F;19179&#x2F;older-versions-of-ida-free" rel="nofollow">https:&#x2F;&#x2F;reverseengineering.stackexchange.com&#x2F;questions&#x2F;19179...</a>
      • JSR_FDED14 hours ago
        Small pedantic nit: Open-sourcing is not the same as reverse-engineering.
        • inigyou4 hours ago
          We are taking back control of our computers by force.
    • tverbeure17 hours ago
      There is absolutely no way that happened. 15 years ago, we&#x27;re talking Fermi class GPUs and chips with hundreds of millions of bits of on-chip state and much more if you include the DRAM.<p>You can&#x27;t tease out the right information by applying random inputs. Which input would you even use? The PCIe interface? You&#x27;d first &quot;randomly&quot; need to get past its complex training sequences...<p>Your postdoc probably wrote micro-benchmarks of some sort. That is a common technique.
      • Taniwha9 hours ago
        You might is they included DFT (design for test, stuff to make sure when you make a chip all of it actually works) - a scan chain thru all the internal flops will provide you with info about the internal flops and how they are connected - reverse engineering this into a model of reality would certainly be extremely hard, but maybe not impossible
        • tverbeure9 hours ago
          While that&#x27;s theoretically possible, it&#x27;s even harder than trying to do it over, say, PCIe, because the latter at least still has a higher meaning to it while scan chain FFs are just grouped based on spatial proximity.<p>And of course that&#x27;s assuming that a) the JTAG port is accessible on these boards and b) the raw scan chain access before scan chain compression hasn&#x27;t been fused off or isn&#x27;t locked behind some authentication protocol.<p>So, no. That didn&#x27;t happen either. :-)
      • kjs317 hours ago
        And you don&#x27;t use an o-scope in anycase, since you&#x27;d need...what...a thousand of them to watch all the signals. You&#x27;d use a logic analyzer. I think I read somewhere that those older nvidia chips had something like 2000 BGA balls, and Tektronix <i>does</i> make an LA that can scale to 2000-something channels (TLA7000), for a modest US$500k or so. Then you gotta figure how to mount the thing to attach the probes.<p>So...agreed...far more likely there was a software solution of some kind if this happened.
    • pixelatedindex17 hours ago
      “oscyloscops” is a way better spelling I gotta say.
  • kayson16 hours ago
    Getting a logic-gate-level netlist from a GDS is trivial with industry standard chip design tools. Circuit designers do this every day. The hardest part will be reverse engineering the functionality.<p>I wish I had more time and I&#x27;d throw Calibre at it.
    • roadbuster13 hours ago
      Indeed, this is a problem for an energetic student who has free access to every tool in the Cadence, Synopsys, and Mentor portfolios: take the GDS-II layout, run it through extraction, generate a netlist, convert to higher-level blocks (&quot;netlist-to-gate recovery&quot;), translate to RTL HDL, then attempt to work out &quot;what it does.&quot;<p>This is just a job posting in disguise for Jane Street: solve the puzzle, get a call from one of their recruiters.
    • Taniwha15 hours ago
      It&#x27;s even easier in this case because they&#x27;ve included the original verilog source ....
      • mercurywells15 hours ago
        No, that&#x27;s for the warmup. The real puzzle only has a .gds file available.
  • Dwedit14 hours ago
    I&#x27;ve looked at Visual 6502 and it&#x27;s way beyond me. I&#x27;ve even looked at the scans where it shows how they severed the connection to disable 6502 decimal mode on the NES.
    • monocasa14 hours ago
      To be fair, the 6502 is very dense, and very hand optimized.<p>Modern systems tend to use &quot;standard cells&quot; for logic which is a lot more digestible.
  • supernetworks_9 hours ago
    I wrote a guide for newbies to learn<p><a href="https:&#x2F;&#x2F;siliconzoo.org&#x2F;tutorial.html" rel="nofollow">https:&#x2F;&#x2F;siliconzoo.org&#x2F;tutorial.html</a>
  • BobbyTables212 hours ago
    Hasn’t Ken Shirriff been doing this for quite a while? (:-&gt;
    • kens11 hours ago
      I&#x27;m too busy trying to reverse-engineer the Intel 8087 floating-point chip, although it would be interesting to look at this puzzle chip.
  • marethyu15 hours ago
    What prerequisites I need to solve this puzzle? I don&#x27;t have much knowledge in electrical engineering.
    • supernetworks_9 hours ago
      It’s hard without some basic knowledge maybe this can help <a href="https:&#x2F;&#x2F;siliconzoo.org&#x2F;tutorial.html" rel="nofollow">https:&#x2F;&#x2F;siliconzoo.org&#x2F;tutorial.html</a>
  • NooneAtAll319 hours ago
    looks like they didn&#x27;t post any blog post about 2nd NN challenge (<a href="https:&#x2F;&#x2F;huggingface.co&#x2F;spaces&#x2F;jane-street&#x2F;droppedaneuralnet" rel="nofollow">https:&#x2F;&#x2F;huggingface.co&#x2F;spaces&#x2F;jane-street&#x2F;droppedaneuralnet</a>)<p>I was waiting for some writeup about permutation decyphering
  • whitten18 hours ago
    Is there something like an Extract-SPICE tool that takes a circuit and gives you back a text rendering of it ?
    • kayson16 hours ago
      Yes. There is exactly that, and we call it an &quot;extraction&quot; tool. It takes a GDS (text representation of shapes in the physical layout), and gives you back a &quot;netlist&quot; (text representation of components and connections in a circuit schematic).<p>Circuit designers use these tools basically daily for two reasons - the first is Layout Versus Schematic. We want to make sure that the physical layout matches the schematic, so the tool turns the layout GDS into a netlist and compares that to the netlist created from the schematic (basically a diff, but more complicated). The second is so we can run simulations that take into account the &quot;parasitic&quot; resistances and capacitances of the wires and metal shapes in the physical layout. It&#x27;s basically the same procedure as LVS with an extra step that analyzes the metal shapes to determine said R&#x27;s and C&#x27;s.
      • Taniwha15 hours ago
        yes, but at the transistor level .... it has no idea at that level how those transistors are arranged into gates directly
        • Taniwha15 hours ago
          I take it back, it&#x27;s a sky130 .gds the gates are explicitly in there
        • kayson13 hours ago
          The tools can definitely recognize static logic gates purely from the topology (nothing special in the GDS needed). I don&#x27;t know that I&#x27;ve tested it for anything more complicated, but as you&#x27;ve pointed out, if the hierarchy is in the GDS it&#x27;s trivial to recover
        • Taniwha15 hours ago
          In fact if you have the sky130 libraries installed (it&#x27;s where I&#x27;m working) you can get a spice netlist out of it in about 10 seconds - really they should have flattened it
    • Joel_Mckay18 hours ago
      Practically No, the stack-up of metal layers often hides the gate structures underneath, and the billions of process cells may not all be the same.<p>Theoretically Yes, as an ion-beam-mill and electron-microscope combination machine can slice up semiconductors layer-by-layer. Given these machines can often also give precise x-ray analysis material data, the exact makeup of the chip can be extracted by competitors given enough time. =3
      • saltcured17 hours ago
        Now you&#x27;re making me imagine some kind of 3D-scanning, confocal x-ray fluorescent spectroscope.<p>Or maybe some kind of hybrid of x-ray microtomography and spectroscopic analysis all in one.<p>But, maybe the energies involved would be about the same destructive power as some microtome slicing technique...
        • inigyou17 hours ago
          We already know that X-rays don&#x27;t destroy chips.
          • saltcured16 hours ago
            I was thinking of the kind of high-energy photon sources that sprawl across a large laboratory campus behind rings of security fences.<p>These things can definitely erode the targets. Tomography experiments have to think carefully to optimize the set exposure angles used during a session, because the target degrades more with each shot.
          • bofadeez16 hours ago
            [flagged]
            • Joel_Mckay16 hours ago
              I’d recommend the Chenming-Hu books on solar cells and semis available as pdfs on his site.<p><a href="https:&#x2F;&#x2F;www.chu.berkeley.edu&#x2F;modern-semiconductor-devices-for-integrated-circuits-chenming-calvin-hu-2010&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.chu.berkeley.edu&#x2F;modern-semiconductor-devices-fo...</a><p>LLMs have already been shown to cause cognitive&#x2F;skill performance losses in some users. =3<p><a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=axOcn--n_lM" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=axOcn--n_lM</a>
              • inigyou15 hours ago
                You&#x27;re replying to a probably now shadowbanned LLM bot that advertises Fable all day
                • bofadeez13 hours ago
                  Are these &quot;shadow bans&quot; in the room with you right now? What a nut.<p>Fable needs no advertising. It was more important than the printing press. Code singularity.<p>But it&#x27;s old news, Fable is no longer my favorite LLM. I mostly use Kimi k3 now. It&#x27;s also more competent than you are in general.<p>The primary point you&#x27;re avoiding is that programming is not a job for professional humans anymore. At least not humans with any dignity or self respect or economic value.<p>I&#x27;m sure you at least agree with this point, like every other reasonable person does now too.
              • perching_aix15 hours ago
                &gt; LLMs have already been shown to cause cognitive&#x2F;skill performance losses in some users. =3<p>Sorry to interrupt your smugness, but have you actually read that study? It&#x27;s basically tautological: people remained mostly unfamiliar with code they did not write, and remained mostly unfamiliar with a library they did not use. That&#x27;s kind of the whole point to begin with! <a href="https:&#x2F;&#x2F;homepages.inf.ed.ac.uk&#x2F;rni&#x2F;papers&#x2F;realprg.html" rel="nofollow">https:&#x2F;&#x2F;homepages.inf.ed.ac.uk&#x2F;rni&#x2F;papers&#x2F;realprg.html</a><p>Unless you want to argue that anyone who cannot code is brain damaged, I&#x27;d say it&#x27;s a bit difficult to suggest that this would amount to any kind of clinically relevant cognitive impairment, much to the contrary of the masses desperately hoping and asserting so. Losing a skill you don&#x27;t use &#x2F; want &#x2F; need is <i>very</i> normal cognitive function.
                • Joel_Mckay7 hours ago
                  Quod erat demonstrandum =3
                  • perching_aix5 hours ago
                    I genuinely wish you were at least half as smart as you think you are, cause then you wouldn&#x27;t be acting obnoxiously like this, willfully betraying your own values.<p>It&#x27;d be generally awesome if the <i>last bastions of humanity</i> actually demonstrated any of the values they&#x27;re championing so hard at all.
                    • Joel_Mckay12 minutes ago
                      I would never kink shame a persons cognitive dildo, or irrational off-topic cult dogma.<p>Have a wonderful day =3
  • mentat17 hours ago
    30 minutes with &#x2F;goal for the solution from Sol w&#x2F; high.
  • q3k18 hours ago
    In a simplified scenario (not too far from this)? Yeah, we&#x27;ve done that in CTFs almost a decade ago.<p><a href="https:&#x2F;&#x2F;blog.dragonsector.pl&#x2F;2017&#x2F;10&#x2F;?m=1" rel="nofollow">https:&#x2F;&#x2F;blog.dragonsector.pl&#x2F;2017&#x2F;10&#x2F;?m=1</a>
    • inigyou18 hours ago
      From where do I know the name Dragon Sector and q3k? You aren&#x27;t the ones who hacked the train DRM, are you? Or maybe active in the demo scene? Or maybe I&#x27;m just confusing you with TRSi?
      • q3k18 hours ago
        Maybe. :)
  • IshKebab19 hours ago
    That sounds like a fun challenge. Feels a lot more tractable than the neural net one.
  • arjie11 hours ago
    Man, these guys always have insane puzzles. What the heck.
  • inigyou18 hours ago
    [flagged]
  • ck218 hours ago
    people who can do this stuff are super-smartypants<p>but reminds me how we&#x27;re going to find out on an industrial level when the Saudis give China some nvidia chips they were grifted<p>they&#x27;ve cloned lots of chips before but nothing that advanced
    • Taniwha15 hours ago
      I&#x27;m a chip designer, back in the early 90s I visited a small silicon house who were doing some work for us, their boardroom table had been taken over by a giant sized photomicrograph of IBM&#x27;s VGA chip and they had a bunch of summer hire grad students tracing wiring and matching standard cells (easier back then because we only worked in 2 layers of metal).<p>They weren&#x27;t trying to steal IBM&#x27;s design, they already had one of their own, what they wanted to know was &quot;is there any as yet unannounced functionality in the IBM VGA chip?&quot; they didn&#x27;t want to be caught out when some unknown registers popped up. Of course it turned out there was as yet unannounced functionality - but not by IBM&#x27;s design, more because of the orthogonality of the design - MODE-X was discovered and became the basis for DOOM&#x27;s speed, anyone who didn&#x27;t support it lost out
      • jsLavaGoat14 hours ago
        MODE-X launched 1,000 ships too with the demoscene.
    • inigyou17 hours ago
      You can do this. If you commit the whole next month to it you&#x27;ll make quite some progress. But you won&#x27;t.
    • JSR_FDED14 hours ago
      They did give the chips to China, but they didn’t work. Turns out they’d first taken them to the basement where someone cut them into pieces before sending them to China.
    • inigyou18 hours ago
      China has no shortage of Nvidia chips. It costs nothing (relatively) for someone to just buy a 5090 off the shelf and send it there.