7 ms·

When I read things like this I start to think about the following question: Given no technology at all (i.e. after an apocalypse!), how would one go about creat

When I read things like this I start to think about the following question: Given no technology at all (i.e. after an apocalypse!), how would one go about creating a computer (as we know them). What sort of knowledge would you need? The science of these things is basically multi decades of trial and error in various disciplines. Imagine a manual that explained how to create everything from scratch!

speedplane 8y ago
Most electrical and computer engineering undergrad degrees (not CS) do give you this information. You first learn how to create a transistor, then basic AND/OR logic gates with individual transistors, then adders and subtractors, and then eventually a pipelined processor with look ahead execution.
- fyi1183 8y ago
  The fundamentals of a transistor, sure, but where does computer engineering include the chemistry of the actual lithography processes?
  - jakeogh 8y ago
    I suspect knowing the fundamentals gets you pretty far, just knowing that UV and masks are a way to do it is enough for people to re-create the tech given the resources. Maybe e-beam would be a easier start? IDK. Not related but fun given the q: https://news.stanford.edu/news/2015/june/computer-water-drops-060815.html https://news.stanford.edu/news/2015/june/computer-water-drop...
    - InitialLastName 8y ago
      It's all well and good to know the fundamentals of photolithography, but in a societal boot-strapping scenario it's not enough to get you out of the technological subbasement. For example, I don't know where or how to derive the necessary chemistry. Or how to construct a building and fans to achieve a clean enough room to do this in. Nor do I know how to build a UV lamp. Hell, I don't know enough about metalworking to reliably manufacture enough wire to build a basic generator to power all this.
    - baybal2 8y ago
      >I suspect knowing the fundamentals gets you pretty far, Not much further than eighties, and you are not getting anything done with seventies era ICs. I'd say. After seventies, most of academic research gets almost wholly separate from the fab floor, and of course fab owners and equipment makers keep to their valuable secrets. At TSMC, the EUV workgroup, for example, is a very select cohort of people with 10+ years at the company, with posted security bonds, screened for loyalty, and signed an omerta style NDA.
      - MrEldritch 8y ago
        ...considering just how much got done with, say, the 6502, saying you're not getting anything done with 70s-era ICs is patently ludicrous. 70s-era microprocessors brought us the home computer revolution.
  - speedplane 8y ago
    There are electives under materials science departments that most engineering students can take (I took a 2 credit intro course). If you want that to be your focus, then you become a materials science major.
  - analognoise 8y ago
    Yes, it's an undergrad elective to build things like this.
bleke 8y ago
I'm interesting in this too and wondering why nobody did huge task and compiling everything into one book, starting from refining materials to making 8051/Z80 era pc
- Retric 8y ago
  Just going from nothing aka sticks, mud, and ore to copper is probably a book on it's own.
- bradyd 8y ago
  There is a book about building a toaster from scratch like this called "The Toaster Project" [1]. The conclusion is that it's extremely difficult and nearly impossible for even a simple device like a toaster. A computer would be significantly more difficult. [1] http://www.thetoasterproject.org/ http://www.thetoasterproject.org/
phendrenad2 8y ago
Building from zero is a fun thought experiment, but it’s a fair assumption that the rebuilding will take place after some catastrophe befalls our world, so perhaps a more useful compendium of knowledge would explain the artifacts that surviving generations will find buried in rubble. Copper wire? CPUs? They’ll have plenty. But will they know how to use them, once the internet ceases to exist?
darkmighty 8y ago
I have a position that the one of the most important technologies needed for interstellar travel (and basically colonizing the entire galaxy) is simply a compact semiconductor fab at a pretty good node -- and it's a technology that almost no one is giving any thought or attention to. Why do I say so? Because imo the most feasible form of interstellar travel is extremely slow (think a few times Voyager probe speed). Most of the technology (energy, propulsion, etc) to achieve this already exists (unlike proposals like Project Starshot), and you can take huge amounts of mass with you. The problem is, things break down on your way. Going slow permits bringing large radiation shields (made simply of water), but the time between Star systems would still be of the order 10,000-100,000 years. Not much needs to be done on the way -- hence most circuits and mechanisms can be kept inactive -- but still you're going to have circuits and all sorts of equipment break down. There's a minimum energy expenditure needed for course correction and an associated simple but constant information processing required. So I believe you essentially need to have a way to keep fabricating chips on your way (unless you could store some reserves in a very well shielded compartment?), probably every few hundreds of years. Some elements may need recycling too. Some mechanical components e.g. propulsion systems might need periodic reconstruction/recycling also, but the technology for compact mechanical manufacturing is pretty well established -- a good machine shop with a few CNC mills, maybe some 3D printers and the like can be made to manufacture a complete propulsion system, while also having total self-fabrication capability. There's also necessity of chips for when arriving at the destination -- again simply bringing the chips with you probably isn't viable, since it would be a while once you could establish a large modern semiconductor manufacturing plant. Again a compact fab is needed (the Starship plant itself could be used in the beginning). Once you have those kind of capable Starships (which require some difficult but not unthinkable technology), colonizing the galaxy isn't really out of question, provided there are enough Earth-like planets (or other kinds of technologically-feasible planets) out there. You would colonize a planet and use it to build a new Starship once it were well established. Probably needing a few millenia to take over the planet and mine the needed minerals and nuclear fission fuel -- not really a big deal on our colonization time scale of 100,000s of years. I'm also assuming this is some kind of Generation Ship, because otherwise compact self-fabrication of state of the art semiconductors may be too difficult (one of the amazing things about organic life is its amazing compact self replication). The psychological realities associated with generation ships aren't too rosy, but I think the end goal makes it acceptable (and some engineering can make them not too infernal for their occupants). Plus I find the idea of humans themselves colonizing the Stars more pleasing, but it may well turn out that AI probes will be the ones accomplishing it (if compact state of the art semiconductor fabrication becomes feasible).

speedplane 8y ago

Most electrical and computer engineering undergrad degrees (not CS) do give you this information. You first learn how to create a transistor, then basic AND/OR logic gates with individual transistors, then adders and subtractors, and then eventually a pipelined processor with look ahead execution.

fyi1183 8y ago
The fundamentals of a transistor, sure, but where does computer engineering include the chemistry of the actual lithography processes?
- jakeogh 8y ago
  I suspect knowing the fundamentals gets you pretty far, just knowing that UV and masks are a way to do it is enough for people to re-create the tech given the resources. Maybe e-beam would be a easier start? IDK. Not related but fun given the q: https://news.stanford.edu/news/2015/june/computer-water-drops-060815.html https://news.stanford.edu/news/2015/june/computer-water-drop...
  - InitialLastName 8y ago
    It's all well and good to know the fundamentals of photolithography, but in a societal boot-strapping scenario it's not enough to get you out of the technological subbasement. For example, I don't know where or how to derive the necessary chemistry. Or how to construct a building and fans to achieve a clean enough room to do this in. Nor do I know how to build a UV lamp. Hell, I don't know enough about metalworking to reliably manufacture enough wire to build a basic generator to power all this.
  - baybal2 8y ago
    >I suspect knowing the fundamentals gets you pretty far, Not much further than eighties, and you are not getting anything done with seventies era ICs. I'd say. After seventies, most of academic research gets almost wholly separate from the fab floor, and of course fab owners and equipment makers keep to their valuable secrets. At TSMC, the EUV workgroup, for example, is a very select cohort of people with 10+ years at the company, with posted security bonds, screened for loyalty, and signed an omerta style NDA.
    - MrEldritch 8y ago
      ...considering just how much got done with, say, the 6502, saying you're not getting anything done with 70s-era ICs is patently ludicrous. 70s-era microprocessors brought us the home computer revolution.
- speedplane 8y ago
  There are electives under materials science departments that most engineering students can take (I took a 2 credit intro course). If you want that to be your focus, then you become a materials science major.
- analognoise 8y ago
  Yes, it's an undergrad elective to build things like this.

bleke 8y ago

I'm interesting in this too and wondering why nobody did huge task and compiling everything into one book, starting from refining materials to making 8051/Z80 era pc

Retric 8y ago
Just going from nothing aka sticks, mud, and ore to copper is probably a book on it's own.
bradyd 8y ago
There is a book about building a toaster from scratch like this called "The Toaster Project" [1]. The conclusion is that it's extremely difficult and nearly impossible for even a simple device like a toaster. A computer would be significantly more difficult. [1] http://www.thetoasterproject.org/ http://www.thetoasterproject.org/

phendrenad2 8y ago

Building from zero is a fun thought experiment, but it’s a fair assumption that the rebuilding will take place after some catastrophe befalls our world, so perhaps a more useful compendium of knowledge would explain the artifacts that surviving generations will find buried in rubble. Copper wire? CPUs? They’ll have plenty. But will they know how to use them, once the internet ceases to exist?

darkmighty 8y ago

I have a position that the one of the most important technologies needed for interstellar travel (and basically colonizing the entire galaxy) is simply a compact semiconductor fab at a pretty good node -- and it's a technology that almost no one is giving any thought or attention to. Why do I say so? Because imo the most feasible form of interstellar travel is extremely slow (think a few times Voyager probe speed). Most of the technology (energy, propulsion, etc) to achieve this already exists (unlike proposals like Project Starshot), and you can take huge amounts of mass with you. The problem is, things break down on your way. Going slow permits bringing large radiation shields (made simply of water), but the time between Star systems would still be of the order 10,000-100,000 years. Not much needs to be done on the way -- hence most circuits and mechanisms can be kept inactive -- but still you're going to have circuits and all sorts of equipment break down. There's a minimum energy expenditure needed for course correction and an associated simple but constant information processing required. So I believe you essentially need to have a way to keep fabricating chips on your way (unless you could store some reserves in a very well shielded compartment?), probably every few hundreds of years. Some elements may need recycling too. Some mechanical components e.g. propulsion systems might need periodic reconstruction/recycling also, but the technology for compact mechanical manufacturing is pretty well established -- a good machine shop with a few CNC mills, maybe some 3D printers and the like can be made to manufacture a complete propulsion system, while also having total self-fabrication capability. There's also necessity of chips for when arriving at the destination -- again simply bringing the chips with you probably isn't viable, since it would be a while once you could establish a large modern semiconductor manufacturing plant. Again a compact fab is needed (the Starship plant itself could be used in the beginning). Once you have those kind of capable Starships (which require some difficult but not unthinkable technology), colonizing the galaxy isn't really out of question, provided there are enough Earth-like planets (or other kinds of technologically-feasible planets) out there. You would colonize a planet and use it to build a new Starship once it were well established. Probably needing a few millenia to take over the planet and mine the needed minerals and nuclear fission fuel -- not really a big deal on our colonization time scale of 100,000s of years. I'm also assuming this is some kind of Generation Ship, because otherwise compact self-fabrication of state of the art semiconductors may be too difficult (one of the amazing things about organic life is its amazing compact self replication). The psychological realities associated with generation ships aren't too rosy, but I think the end goal makes it acceptable (and some engineering can make them not too infernal for their occupants). Plus I find the idea of humans themselves colonizing the Stars more pleasing, but it may well turn out that AI probes will be the ones accomplishing it (if compact state of the art semiconductor fabrication becomes feasible).