Maybe we should start pulling chips of trashed electronics. There must be hundreds of thousands decently fast chips going into our trash.
How many dead xboxes and PlayStations have chips in them that are still good?
I'm thinking of a reverse pick and place machine that pulls chips of boards (boards that can not be used to fix other dead devices) and is able to test each valuable component for later reuse.
> Maybe we should start pulling chips of trashed electronics. There must be hundreds of thousands decently fast chips going into our trash.
This is already been happening for decades on industrial scale in countries doing eWaste recycling. Just a decade ago, you could've still find people selling refurbished chips in Shenzhen on every corner. Now it kinds of became a much more low profile enterprise.
The US defence department, colloquially known as the pentagon, ran a wargame scenario a few years ago to see what will happen if US will be hit with a worst case scenario, years long semiconductor unavailability due to sabotage of domestic fabs, massive industry wide cyberattack, or something happening to Taiwan.
One of comissioned think tanks recommended that exact methods of military going door-to-door to gather old gaming console, smartphones, and PCs for conversion to use in munitions, and military equipment.
My first reaction to this plan is that it’s for the birds. The government would be better off just building a fab themselves. The raw materials aren’t that hard to get hold of; integrating consumer electronics into military systems would likely take just as long, if not longer (with all sorts of other negative effects on quality and security). I guess the analogy is with scrap metal drives in WWII, but this situation is nothing like that at all. It’s a fantasy scenario that makes no sense.
We defend Taiwan with everything we’ve got. There is no plan B, and it’s not just about semiconductors.
The government would be better off just building a fab themselves.
But you can't just 'build a fab' can you? Thats the point. You need a shipment of ASML's magic EUV machines and a bunch of their engineers and a bunch more technicians and managers with the right kind of skills for running a large cutting edge fab.
You only need those for the latest and greatest. A lot of options open up if the priority is any decently fast processor over the latest. Multi-CPU systems fell out of favor in the consumer space because single CPUs got so fast with these processes, but multiple CPUs with an older process is fine in this scenario.
Military-grade equipment and ammunition don't require cutting-edge manufacturing nodes. There are fabs for older nodes aplenty, and new ones don't require cutting-edge equipment from ASML. Also, since older fabs are usually kept around, there is going to be demand for older tech and spare parts.
To sustain AI and other applications that require cutting-edge tech, the best strategy would be to either stockpile chips (not really feasible since the tech is deprecating) or to build spare capacity, along with the necessary supply chain.
> a bunch of their engineers and a bunch more technicians and managers with the right kind of skills
In a WWII scenario, how long does it take to rapidly retrain a bunch of people with these skills. We have no shortage bunch of brilliant, quantitatively minded people trading derivatives and writing ML models to recommend TV shows. In a total war scenario, I assume most would be drafted for the war effort. Semiconductor engineering is complex, but probably not so complex that your average Math Olympiad couldn't pick it up in a month or two.
I did some materials engineering coursework in college, and I think your comment significantly underestimates the technical complexity of <100nm process node manufacturing.
We had many brilliant minds working on these problems for decades, but it took time, skill, effort, and money over these decades to achieve each milestone of development. Now each of these firms (and each of their suppliers) has trade secrets that go deep, and barring some unlikely breakthrough of brilliance, these secrets can't be independently discovered by the brightest minds in any less than the time it already took to discover and develop them in the first place.
There's also the problem of supply chains -- modern fab tech requires a ton of downstream tech: the fab machines, the parts in them, and the places those parts were developed and manufactured... and the machines that were required there as well (repeat...). Also the raw materials, especially rare earth minerals.
> Semiconductor engineering is complex, but probably not so complex that your average Math Olympiad couldn't pick it up in a month or two.
We have a chip shortage going on for months now, if all it took was two months and a bunch of smart people, those billions of dollars in chip orders would've made it happen. That tells me it has to be a bit more complex than you think.
Has any major organization started or entered the semiconductor fab space in the past year? From speaking with industry analysts, my understanding is we could spin up additional capacity. The issue is that most everyone fears the shortage will be over after the Covid supply chain disruptions, and therefore any new investments will not recoup their necessarily long payback periods.
This is a very different calculus than a WW2 type scenario. Fully mobilized, the government would almost certainly pay up for microchip capacity today, regardless of the long-term payback residual.
If there was evidence of deep-pocketed firms trying and failing to get into semiconductor fab in 2020, I'd change my opinion. Like if Google or Amazon were trying to open their own fab plants, but were failing, then I might suspect that a WW2 fully mobilized war effort would run into difficulties. But as far as I can tell, the issue is that nobody wants to jump into the fab business, regardless of the temporarily higher profits over the next 2-4 quarters.
A 10nm/7nm fab costs $10B and takes years to build, assuming you've got a process which works - developing that from scratch will take you a decade before you can even design the fab. You could compare the enterprise to a smaller scale NASA.
It's not a question of knowing how to build industrial capacity, it is actually building that capacity. Before you can build a fab, you need to build all the machines that go into a fab, and before that all the machines for building those machines, and so on and so forth. 9 women can't make a baby in a month, at some point you come across steps that just take time.
If you think there's a scenario that justifies it in 10 years, you start today. In our quest for ever-more-just-in-time-efficiency we've forgotten that it doesn't work for certain things.
Of course this is a far cry from ripping a SoC from a smartphone and putting it into e.g a bomber drone, but I assume there's much more being done - just not published.
One of the reasons cited for using XBox controllers was the availability. If one breaks, you can find them at any port. Very apropos in this context...
I have a vision that on a submarine there is a cable box, not unlike the one that most of us probably have, filled with random cables with connectors we haven't used in years. And somewhere in that tangled mess is a spare XBox controller. I can imagine some poor Seaman trying to untangle a the controller cord so that they can use a periscope. I'm sure the Navy is far more organized, but it's funny to think about.
I absolutely believe it-I use one professionally to maneuver an underwater ROV. They're intuitive, already familiar to a lot of people, and surprisingly reliable.
There’s nothing macho about a conflict of this magnitude, it would be absolutely horrible. But we have to set clear expectations and unambiguous boundaries to avoid a dangerous miscalculation, not pretend we have contingency plans that don’t actually exist.
There is nothing inevitable about a Chinese invasion of Taiwan. My sense is that the Chinese leadership are far too insightful to actually start that war, but you never know.
Seriously. We're talking about horrible things, here.
But if it came to a nuclear exchange, China has ~300, and the US and Russia ~6,000 each. Total yield numbers are harder to come by(?), so those are +/- considering tactical-scale inventory.
Historically, this was because China's nuclear doctrine was to limit their stockpile to a minimal deterrent.
Personally? Sitting about 40km from a priority US target, I'd support the US risking a strike to guarantee Taiwan's freedom.
Either you stand for something, or you don't. And China doesn't get to claim modern territory by dusting off historical documents. Hong Kong was untenable, but Taiwan should be a line. If its citizens don't want to be Chinese, they get to make that choice.
PS: Russia would probably care less about a strike on China, unless China negotiated their support via treaty. That ship sailed in the early 60s.
Respond with overwhelming force if being pushed. If however you are the one starting pushing you are just a bully or in this particular case fuckin' war criminal.
It’s not “dusting off historical documents” it’s official government policy on both sides of the straight that there’s “one China”.
One could argue that the only reason why the ROC hasn’t declared independence is because they’d fear that it would provoke an attack from across the straight, but the reality is that independence is not a cut and dry issue in Taiwan.
Take some time to actually educate yourself on the issue before advocating for nuclear war.
Taiwan has been Dutch, Han, Qing, Japanese, and ROC. Any appeal to history is nine^H^H^H^Hten dashes of propaganda.
> it’s official government policy on both sides of the straight that there’s “one China”
That's an oversimplification. The KMT / Pan-Blue believe there's "one China" in the sense that they believe the PRC is an illegitimate government, currently occupying mainland Chinese territory that belongs to the ROC.
The DPP / Pan-Green believe there are two Chinas, with Taiwan as an independent entity from the mainland PRC. Their current official position is that this state of affairs already exists, and therefore there is no need to make any declarations or changes from the status quo.
So in summary, about one half of Taiwanese political power considers the communist government as illegitimate, and the other half considers Taiwan independent. Which I would guess is fairly different from what the PRC defines "one China" as.
Yes, there are different definitions on both sides of the straight. And while the independence side has gained strength in recent years, the official stance has been that China and Taiwan are a part of one country, however defined, and, to my knowledge, no official government act of the ROC has been promulgated to the contrary.
As I stated in my original reply, independence isn’t a cut and dry issue. To assert that these claims are “dusty” historical claims in light of what you’ve written here is disingenuous.
Edit: I should add that there is only one government in Taiwan, the ROC. The stances of the individual political parties cannot be taken as synonymous with the stance of the ROC itself. To do so would be as ludicrous to say that there’s no right to an abortion in the United States, because 1 of the two major political parties opposes it. The most that can be said is that both abortion in the US and independence in Taiwan are controversial and delicate political issues within each respective jurisdiction.
Because that's literally the land the hypothetical war would have been started to capture? Because the casus belli to their public has been that it's (PRC) Chinese land by historical and cultural right? Because the entire domestic and international point of provoking a Strait war would be to demonstrate China's strength?
It's a pretty bad look if your "winning" looks like "killed millions of people and rendered the land in question uninhabitable for a decade+." (To say nothing of the potential non-Taiwan collateral damage)
The stockpile is a ridiculous measurement. How many would it take to have mutually assured destruction? A dozen? One of the most worthless dick measuring contests in history.
A 100kt nuclear warhead airburst will cause moderate damage to an area of 33.5km^2. Beijing has an area of 16800km^2, 501 times larger. China has 160 cities with over 1 million inhabitants. Even if they were all used against major population centers (which they wouldn't be) and all of them made it to their targets (which they wouldn't), thousands of nuclear warheads would still not be enough to destroy a large nation.
So far, we had only _two_ instances of nuclear weapons fired in anger. Post-WW3 (if civilization survives) would be a world where nuclear strikes between countries are a reality and recent history. Not sure whether such a world is worth living in either.
I can kinda see the value of getting a million PS3 cpus for in your new drone swarm or industry controllers. But how is gathering generic decades old hardware going to help? For Computation as a service its not that valuable or easy to install. And for reuse in 'new' devices it's even more questionable. All this stuff barely just works in this specific configuration.
I'm only doing a few things on machine with around 32GHz worth of processing power (# of processors * GHz) that I couldn't do on my 80486 running 33MHz with a few MB of RAM. Emacs starts up a little bit faster. Pages in Chrome are more interactive than in Lynx, Mosaic, or Netscape. My word processor fits more text.
On the whole, that 1000x increase in performance mostly bought the ability to have most of the code that runs be interpreted or JITed rather than statically compiled, to play videos, to do some heavy numerics and graphics (including things like family photos and videos), and so on.
It's hard to think of anything I really /need/ in a life-or-death sense or an economic-survival sense that couldn't be done on older hardware with appropriate software. There's a ton of things I want, that help, or where migrations would be massive projects.
If computers were to entirely disappear, and I couldn't automate things, communicate digitally (emails, messaging), word process, write code, I think that'd be a major systemic-collapse-level implosion of society. A lot of people would simply die.
On the other hand, if computers were to regress to 33MHz-level performance, Youtube isn't sticking around, but we'd likely adapt as a society with some structural change, but without such a collapse.
Older levels of performance wouldn't cripple humanity. But putting together something that actually works out of a big parts bin is a crapshoot, and supporting a deployment of machines which are each completely unique is an IT nightmare.
I think the best chances to get something useful would be to collect x86 consoles. x86 means you've got a well supported target, and consoles means you've got limited hardware variations.
Military could probably pressure Microsoft to give them some way to turn an original xbox into a PC; and if you give people $50 for an original xbox most will be happy.
There are a number of commonly-used microcontrollers. Z80 comes to mind. So does Atmel. So do a few common ARM variants. If you can cover those, you're not in bad shape.
As an end user it would be manageable but dont forget that the whole world runs on servers and many Industries would collapse if computing power is gone
I feel like BBSes did okay for most purposes over 300 baud modems. I mean, I'm not getting a picture of what I'm buying on Amazon, unless I want to wait 5 minutes for it to download, but I can still make the purchase.
On the server side, you'll be running C code rather than a nice high-level language, but it will get the job done. You might not have all the ML which lets me get the most relevant product recommended, but....
Emails definitely won't be HTML, but they'll come through.
because for a lot of things you don't need that much cpu power.
Creating something that can read several sensors (switches and buttons are also sensors), and putting aggregate status updates on a serial line, or displaying status lights, triggering relay or something similar can be done, on decades old gaming consoles, and is probably one of the most common use of processors in the world today.
In fact it is easier than trying to use modern cpu, that wasn't designed for that.
Not everything is about computation, in fact in terms of units sold its probably a minority.
"The US defence department, colloquially known as the pentagon, ran a wargame scenario a few years ago to see what will happen if US will be hit with a worst case scenario, years long semiconductor unavailability due to sabotage of domestic fabs, massive industry wide cyberattack, or something happening to Taiwan."
There is also the danger of counterfeit chips. It is already a plague for those of us hoping to get some good deals from Chinese vendors, but the US Navy as well was hit at least once with fakes.
Aren't these BGAs? Safely removing and then also re-balling large non-socketed chips is tricky.
Do they actually have machines for this? I understand that the re-work house we use do it by hand and also spend a chunk of time inspecting the chip once removed since it's very easily damaged. Maybe there are more automated places doing this on a mass scale, we don't do it very often which might explain the general poor results.
Xboxes and PlayStations have cryptographic chains of trust going back to their on die boot ROM. You can't just pull it out and use it on another Xbox or PlayStation, much less use them for generic computation.
I think the Xbox One is the first console to have no unsigned code execution at all during it's lifetime.
You can trivially run whatever code you wanted to if you seized the signing keys; I suspect that in such a "collect all the general-purpose computers for general-purpose computing things" that would be the first order of business.
Not that the government doesn't already have copies, of course.
>How many dead xboxes and PlayStations have chips in them that are still good?
Those things have one big APU (with weak CPU cores), some VRM and memory, GDDR5 + useless DDR3. I'd discount the APU as it'd need someone to make OS support for with the blessing of Sony/Microsoft. That leaves the chips of GDDR5, which can be used by low end GPUs - I don't think they'd ever recoup and investment and likely cheap manual labor of preheat + hot air is the easier option. No idea if anyone would get a license from AMD or NVidia for such GPU.
Edit: the capacitors would have the best value but well... I cant see anyone relying on old caps for anything but repair.
Serious answer: Reusing old components makes no sense as the changes in technology are too big (say from a PS3 which has 45nm chip to a 7nm PS5 cpu). However you can keep using your playstation for longer as the new one will be more expensive or harder to get.
I bought a Tesla T4 on eBay. Many people will buy a set of broken laptops and use the parts to make a fixed one like Like Miami. Louis Rossman constantly has to find donor boards to fix laptops.
A PS3 only has 256mb of ram, so you need to get it out of the case, remove it from the board, check it still works, package it up and sell it.
But I can buy 2GB for 5 GBP. Can you do all the aforementioned things for less than that?
I wonder if we could sort and melt down chips as a source of raw semiconductors (I hear rare earth metal mining/refining is very dirty). Maybe that would be economic?
If latest chips were unavailable (war, civilisation collapse), we'd do just fine with existing chips for decades or maybe even centuries.
With clever software, you can eek much more real world use out of hardware. For example, your PS3 is probably sitting with its CPU idle right now, when it could be doing useful computations for someone else who needs more computing power.
> existing chips for decades or maybe even centuries.
Chips experience solid-state diffusion, so I'd expect the expected survival probability of chips (ignoring any infant mortality effects) to go something like feature_size / sqrt( time ). If this is true, then I'd guess modern process nodes would have very low survival rates out at 100 years. Though, it's been over 20 years since I had any solid state chemistry, and most of my experience with diffusion modeling is in financial models, so take my wild guess with a grain of salt.
But the failure rate in observed time for non-nand components is very very low.
Take a 5 year old phone, and the chances of the main board being bad might be 10%... But the chances of the main CPU being bad is probably under 0.1%. The other 99% of failures are mostly water damage, physical damage, bad soldering, fatigue failure, bad nand, etc.
Maybe they'll all fail at once, but I somehow doubt it. IC's from the 1970's are still going strong, also with a low failure rate.
Yet here we are https://www.youtube.com/watch?v=qNje63vx73s Chinese "Motherboard" made using salvaged VRMs and transistors, salvaged capacitors, salvaged Intel Chipset!, and made to work with retired Intel Xeons and older DDR3 ECC memory.
How many dead xboxes and PlayStations have chips in them that are still good?
I'm thinking of a reverse pick and place machine that pulls chips of boards (boards that can not be used to fix other dead devices) and is able to test each valuable component for later reuse.