Comments by "Siana Gearz" (@SianaGearz) on "Asianometry" channel.

  1. Material recovery seems like something that can be effectively pushed back for a while. If the battery is decommissioned it can merely be disassembled into cells, and then those cells tested, matched and reused. Combined with a reduced charge termination voltage, and thus effectively reduced capacity, as well as reduced charge current and reduced discharge current, so reduced thermal stress, they can last another 2 decades after they have had a decade of relatively heavy use. A perfect use for that would be combined with intermittent renewable energy sources (solar, wind) to buffer them and reduce reliance on coal and fossil. This solves two pressing needs simultaneously very cheaply. The moderate capacity of recycling that shall be reached soon or might be reached now can be sufficient to catch just the more heavily decayed cells that cannot be reused. Ah yes, you mention that pretty much.
    52
  2. Ironic that Toyota ended up among the least affected manufacturers. Presumably they took supply chain security more seriously.
    42
  3. Botched the wording a little bit, not to speak of the kind of vocabulary not used by lead representatives of a polite, child friendly company. Shigeru Miyamoto, quoted regarding N64 delay: "A delayed game is eventually good; but a rushed game is forever bad." Interestingly reviews don't even play into it, as they knew they had young loyal fans who would be buying the product without consulting any reviews. Disappointing them would burn this relationship forever, which is much worse than bad reviews. Remember, people didn't have Internet access back then. Magazines were around, but their reach was in tens of thousands copies while product reach was in millions for the same given region.
    35
  4.  @gorillaau  that is potentially a myth, at least Teapo has indicated that Japanese companies blew the story out of proportion, possibly to push up their own profile and generate more sales. It's possible that Taiwanese companies ran into the same issues formulating and characterising capacitors as Japanese did in the early 90s, just later. Indeed there were some truly funny batches from unknown brands (not major Taiwanese brands) which measured "ultra low ESR" right off production line but just wouldn't last but they would have been gone by 2002. The continuing capacitor plague story is probably down to not observing the maximum ripple current and derating by electronics manufacturers, and then shifting the blame.
    33
  5. You man Tseng Labs?
    31
  6.  @scarletmoon777  besides commercial purposes there are also strategic ones. A country without its own armaments is necessarily under someone's control.
    26
  7. We need more outdoor-capable electronics. There was a company that made drop in replacement laptop/netbook displays which are (nearly) full colour indoors but turn (nearly) monochrome high contrast under strong sunlight. I'd like more of that sort of stuff.
    24
  8.  @gorillaau  the 80% water electrolyte with a little boric acid in the funny capacitors with no stabilising compounds is documented, it should have been obvious to the high ranking chemist that is known to have quit Rubycon and moved to Taiwan that it would just eat up aluminium in no time, by all reason someone else who doesn't have the faintest idea or was just up for a quick scam is responsible for that garbage.
    23
  9.  @964cuplove  it's millimetres. The silicon crystal is manufactured either 20cm thicc or 30cm thiccccc (and very tall, those are impressive things in their own right, kill you easily) and then sliced, that's the wafer size. Problem is any equipment works with one size or the other, you can't like peacemeal upgrade the factory.
    23
  10. The SAM is of course an inherited Atmel chip. Indeed all Atmel, Microchip, NXP, Nuvoton, STM low-end micros are sold out a year in advance and have been for a year and something, everything under $3, most things under $7 as well. And it's been a lot tougher to work around than shortages of higher end chips, which are still kinda available. Also special function analogue/linear semiconductors of particular kinds are hit hard. Chinese domestic chips have better availability. Maybe more 45-28nm utilisation? I do think maxing 28nm and node shrinking designs needs to be an intermediate goal, i feel, not adding another 90nm fab. But i think more 28nm fabs won't go to waste.
    22
  11.  @alecpharris1455  Why "Western nations"? Cooperation with the Soviet Union would seem like the only sensible approach.
    20
  12. It's also the only computer intended to be sat on. Very warm and comfy.
    20
  13. I'm just amazed by the breadth of topics you're able to cover!
    20
  14. With most of the economy consisting of state-employed workers, SED didn't have much difficulty manning such a counter-demonstration, under implied threat of withdrawing the wage bonus or disciplinary action that can lead to termination.
    18
  15. You can still make a ladder out of industries that are moving out. Like you have a manufacturing business, gradually it accumulates know-how and engineering acumen, and eventually, you can partition and move out most of the manufacturing and retain the more space efficient and more lucrative engineering and some critical and high tech manufacturing bits. But for that the industry must have to have been there to begin with and allowed to develop in this direction, partially naturally, partially maybe with incentives. It doesn't help if you just tear the industry down and then expect higher value jobs to just appear and stay there, there's no roots they can grow from!
    18
  16.  @Ritternkreis  Yes he does it alone. But he also has an educational and work background in semiconductor manufacturing, which is how he's able to do it.
    16
  17.  @truthsocialmedia  I think you might be confusing something. V2200 is entirely Tseng-free. And Tseng never finished its 3d chip, at least not before going belly up. But it was indeed fairly useless.
    16
  18.  @PhilippBlum  I think the OP assumes he has a team due to density and quality of output; while it's really one dude who has a decade worth of a backlog of topics he was thinking about and writing things down.
    15
  19. Yeah my genuine Z80 never sat or worked right in a Soviet socket. But the whole computer had issues, the semiconductor quality was low.
    13
  20. I knew exactly what the video was going to be about the moment i saw it in my feed. Unless somoene else didn't, or particularly nearly skipped the video due to thumbnail/title, i don't feel a clarification or improvement is necessary.
    12
  21.  @pqsk  It was probably his ONE municipality. I can think of other municipalities in Germany which went somewhat digital early (there are privacy/law related issues why a lot of communication can't be purely electronic specifically) and even maintained their own Linux distro for a while.
    12
  22.  @ChrisJackson-js8rd  Hey i don't feel this video is a qustion of blame, but learning the history and the situation.
    11
  23.  @xrysf03  Near the end of WW2 Carl Zeiss which was located in Jena was split. The company core actually fled to Oberkochen in West Germany; but the leftover equipment that they failed to evacuate continued to run as VEB Carl Zeiss in Jena. Here he's talking of the East German company, later known as Jenoptik after German reunification, since continuing to call it Carl Zeiss would violate actual Carl Zeiss name rights. The key long term partner of ASML has been the real Carl Zeiss, the one in West Germany. Eventually, Carl Zeiss bought back Jenoptik and reunited the company.
    11
  24. I prefer reporting em. But it never seems to do much.
    10
  25. Lenin is usually quoted saying "meals" rather than "bowls of rice". However there is no evidence he actually said that. It's a quote one may encounter in English speaking circles, but i had never heard of it in a Russian language context, although i have experienced the last days of Soviet Union from within. It also rings untrue, since Soviet Union used famines to enslave and control people, didn't care about a little mutiny, chaos did not totally ensue. In all odds, you lot invented the quote.
    9
  26. What goes up must come down Yet my feet don't touch the ground See the world spinning upside down A mighty crash without a sound I can feel your every rage Step aside, I'll turn the page Breaking through your crazy maze Like a laser beam, my eyes on you
    9
  27.  @Conservator.  per transistor cost bottoms out at 28nm, so maybe there's a way to regain some of the cost advantage vs. 130nm? It's just that there wasn't enough pressure to build more, not enough to offset the risks. You do have a problem that IO area probably won't shrinkify much at all though.
    9
  28.  @acedeuce  Rumour has it, has a year long queue of videos which are available to paying patrons immediately while they get unprivated to the general public on a schedule.
    9
  29. China is so rapidly growing that their industrial output gets easily absorbed domestically.
    8
  30. As it happens, i have some relevant experience. 10 years ago, i was a developer on a commercial 3D engine - at the time, it was slowly failing as a game engine, we were shipping some WiiWare and XBLA shovelware, but for all intents and purposes, Unity was eating our lunch. Before Unity came about, we had the most popular indie and small-scale game engine in the region, with some AAA-ish successes. With just a couple exceptions. One, is that we have carved out a niche in virtual studio, that is all these news and weather and all kinds of TV broadcasts which were filmed against a bluescreen, so they had camera tracking, realtime compositing, and the studio would be assembled and rendered in a game engine. I didn't think this trend ever looked good or was good and as it happens, now it's dead, the virtual studios have been gone for 5 years. The second was the CG movie previsualisation. The problem is, when you're filming a movie against blue/greenscreen, where do you put the camera, what will actually be in the shot once the post production team is done with it? Previs doesn't have to look entirely photorealistic, but it has to be all the right things in the right place and it has to respond to lighting, and it solves that problem. On-set previs with camera tracking will allow the DP and director to see on a screen an adequate approximation of what they're aiming towards, on set. So that was our second hope to survive, and we were well equipped, given existing camera tracking partnership and TV foothold and the technical traits that we offered. Unfortunately, a lot of other companies also recognised the need, it was actually a massive push at the time, Crytek was actually the only big game engine players represented in this area directly, and they had a lot going for them, like A LOT, others had some sort of ad hoc tech specialised for the purpose, bound more closely to industry animation software, or were built also on top of existing game engines. What's happening now is that Unreal is eating everyone's lunch, it's digging hard under Unity on all fronts. Indie devs are also gradually less interested in Unity, and well you have mentioned The Mandalorian yourself. What Unity needs is a foothold and some industry expertise to enter this industry. If they can get at least one major studio not going Unreal, they have at least some hope in this industry. There appear to have been a number of what i might describe as panic acquisitions by Unity just to make sure Epic/Unreal doesn't get there first. Also it is a little funny to hear a company where core engineering on their namesake product is from Denmark and is assembled from European devs first and foremost to be described as "American" but that's the world we live in, i guess.
    7
  31.  @kasuha  arguably there's actually more 33khz crystals around than any other. Many devices are outfitted with two crystals, one in the 8-20MHz range for their active mode and a 33KHz one to enable ultra low power interrupt and RTC functionality while the rest of the device is powered down.
    7
  32. ​ @gaboronem8615 that person is trolling you.
    7
  33. I don't see TSMC as a narrowly profitable company. It's very healthy by being ahead. Besides it's an ecosystem. TSMC and UMC work in conjunction with many local semiconductor design houses; which in turn work in conjunction with computer and electrical engineering companies that make products, designed in Taiwan, manufactured there or abroad. The web of tightly knit industry connections allows all these industries to operate faster and more efficiently and be close enough to leaders to be competitive.
    7
  34. Russian companies are de facto disconnected from international customers because payment methods no longer work internationally smoothly. Companies that had any international customer base are going to move to Georgia, Emirates, Egypt, Turkey, Israel and whatnot.
    7
  35. Well there's also Dutch and Flemish languages which are... identical save for pronunciation. Nobody should really care whether you should consider them languages or dalects of each other. Technically a dialect is a way of speech that isn't formalised and evolves wildly, while a language is one that has an official governing body. So indeed British English and American English should be considered languages, not dialects. In turn Bavarian (Boarisch) is not understood by most Germans, but is a dialect, since there isn't somebody formalising it and teaching it at schools in that form. Declaring that something is a dialect can have a bit of imperialistic stink to it especially if the carriers of that way of speech happen to disagree.
    6
  36.  @megalonoobiacinc4863  Pollution? No. I would say pollution is something that you release into the environment uncontained, and i don't anticipate substantial release like that in the future, and to a small extent, radioactive material is just naturally present in the environment, so as long as you stay well under this order of magnitude, it's fine - and also to be considered that burning coal releases radioactive pollution as well. Contained waste products that need long term management is what you might be thinking of. Ultimately the consensus has emerged that you can just dig a deep enough hole in some vaguely stable rock and bury the previously stabilised stuff in there, and then forget about it and it's absolutely fine long term without reservations, at least at foreseeable future scale of waste production (you possibly can't just scale it up endlessly); but then there's NIMBYs. Other solutions include further use of what is currently waste but there are political roadblocks as well. Ultimately current form of nuclear power is a temporary measure, to hold us over for the maybe next sub 50 years until we figure something out. As far as temporary measures go, i'd say it's pretty decent as long as it isn't the only hydrocarbon independence measure taken but part of a broader strategy. It's also a temporary measure because we're going to run out of uranium, so our potentially limited ability to safely entomb waste is well balanced by the raw resource being limited in the first place. So waste isn't really a problem.
    6
  37. Yeah the vaguely plausible looking conversation thread kind.
    6
  38. U880 is a decent chip, this exact die was used under different part numbers across all of Eastern Europe and Soviet Unionn.
    6
  39. Two different Zeisses in a way. The Zeiss spoken of here is now known as Jenoptik. The real Zeiss ran off to the West, and the East German Zeiss was whatever remained.
    5
  40. Strong disagreement on all points. German engineering businesses have their hands in a LOT of pies, thousands of them. It has semiconductor business and others. Infineon is one of the larger semiconductor companies and is German. Germany also has one of world leadership positions in chemistry (BASF etc), medical technology, robotics and factory and engineering equipment, and companies like Bosch are involved in more engineering and manufacturing ventures than you can count, always behind the scenes, critical supplier. SSDs will almost fully supplant HDDs when the price is right for almost all uses, and the price of SSDs has basically no lower bound in sight. HDD capacity per square inch has been barely growing in the last 10 years, and it is an extremely complex assembly which can no longer be reduced in cost. SSDs are suitable for an approximately 5-7-year online storage use, just like HDDs. So when the device is normally connected to power, it will retain the data, and can survive power removal for a moderate length of time. What SSDs are not suitable for is offline storage. So if you disconnect an SSD from power, after a year or two of being unpowered, it will rapidly start losing data, while an HDD loses data when online, but remains perfectly stable offline. An SSD while connected to power periodically "refreshes" so reads and rewrites data stored on it, to nudge the gate voltages into alignment again and reduce data loss, while when it's not powered, the drift occurs uncontrolled. The refresh itself consumes a small percentage of SSD's overall limited write endurance. In this context, "online" means whether the device is powered and running and the data is accessible within fractions of a second after being requested. Currently, the vast overwhelming majority of data storage is HDD-based, online, redundant. To the extent that SSDs exhibit a data loss even online, it's only a matter of redundancy and cost. So when SSDs reach half the price/GB of HDDs, you can simply use twice as many SSDs to store data more redundantly and you will also use less power than with HDDs, and the break even point will come much earlier than that. Considering SSD power efficiency and power cost, it's likely when price/GB reaches near parity, HDDs are dead for all uses except offline backup. In offline storage, HDD will face more competition from tape storage. Currently, HDD only wins in offline storage because low price of HDD is facilitated by the large mass market that needs HDDs and uses them for online storage and general use, which will break away. When HDD price rises, there will be a re-evaluation that favours tape again. There is magnetic tape (LTO) and potentially optical laser tape. I think Seagate is gearing up for LTO expansion. The economic cross over point and thus extremely drastic demand decline of HDD will occur within 5 years.
    5
  41. If you'd like to elaborate in detail, it would be much appreciated.
    5
  42. You prototype it in simulatable or synthesizable form, in simulation or on an FPGA. It can be even shipped into products as an FPGA. But i bet you will come up with something that either isn't 'new', you just missed part of the history lesson; or it will turn out not to be efficiently implementable in silicon, or will turn out inefficient from practical concerns, like lack of great tooling ecosystem. I welcome you to prove me wrong. Also saying "more efficient" incurs "... for WHAT exactly". There's a reason a microcontroller and a server CPU are very dissimilar, in spite of potentially sharing parts of an instruction set. Are you designing an instruction set or a microarchitecture? Two pairs of shoes.
    5
  43. They existed but they weren't available and didn't satisfy potential demand.
    5
  44. But how do you solve the problem that there isn't enough annotated real world data? AI can be bootstrapped on synthetic data, hopefully from various sources (including say 3 or 4 videogames), and then can help convert raw real world data into annotated data under human supervision, which can then be poured back into the training mix. The people would manually exclude any data that was annotated wrong. Let's say you have plenty of annotated real world data (which we currently don't) and use only that for training, then you have another bias problem: it shows all the good driving and little bad and unexpected participants. But we NEED to take extreme circumstances into account, unusual situations, and weigh them heavily vs. the usual, because the cost of running over a human that you can't see yet or driving off a cliff is extremely high. So you have introduced a new bias problem, that your AI doesn't understand the concept of caution.
    5
  45. I don't know shit about Mexico but i'll tell you what about bits and bobs of Soviet Union. Like that whole Russia, Ukraine, Kazakhstan, and the rest of them? Well maybe not right now, but if you look back 10, 15, 20, 25 years, nothing happened there either. So first off why you'd wanna. Well there's the people, who are highly educated, lots of engineers, scientists, but also skilled former factory workers, from when these countries fully supplied themselves but industrial automation and efficiency was low. With a little money they will create, automate, optimise, and they're naturally resourceful, they're used to things not being quite by the book, things being impossible to acquire, but they invent and make do. There's also space and natural resources so the cost factors are very much on your side. But you can't, it would be the worst idea ever. Reason number one is organised crime. "Racket" they call it. Shakedowns. Everything you build, any machinery you acquire, it's not safe, you are probably going to lose it. And of course the organised crime has all the ties to the police and government, you as a budding industrialist do not. They shield each other. The bureaucrats have so gotten used to being fed, they have created a system where they can do anything and you can do nothing, they will create endless impediments, and you will be out of an unpredictable amount of money trying to bribe them, and bribes aren't optional, they are a hard necessity of doing business. You won't even know how and where and why everything is crumbling around you. Even if you were able to navigate the whole thing, like you make your connections to organised crime and you actually get protection and help navigating the bribes, you are likely still to get caught in the crossfire between different crime syndicates. And the hidden tax is overall incalculable, and expands to encompass anything you might earn. So yeah i don't know shit about Mexico, but if they've got some of those issues on their hands, you can't build there.
    5
  46.  Zaydan Naufal  Yeah i have Vizplex based e-readers, but they aren't very suitable for general tasks, actually i have one SONY with a touchscreen and it's an annoying piece of garbage, and i had one by Chinese company Hanlin customised by a Ukrainian partner company, LBook, it was actually top notch. I used to do quite a bit on my Casio PocketViewer and Psion5 in the late 90s, early 2000s. Those had monochrome screens, worked nicely enough in sunlight. I'm pretty sure i had some pretty hefty books on my PocketViewer as well. Oh wait i should just drop some batteries in there, i bet it still has all the data from back then... as opposed to Palm Pilot and Handspring Visor... although Visor Edge just feels SO NICE, i actually have one of those as well, so nice. I actually still play the original Gameboy Advance outdoors :D the only game console that works in the sun better than indoors.
    4
  47. It's funny how just adding "Dodge" to something immediately makes it less noble...
    4
  48. x86 is a prefixed instruction set, so it has potentially millions of valid instructions, most of which are redundant, and more byte sequences that do SOMETHING in spite of not being reasonably expected to be an instruction; see "Sandsifter - breaking the x86 ISA". It is probably more or less tested for a few thousand instructions that compilers and assemblers will actually emit; but for one, nontrivial bugs do pop up from time to time, where an instruction performs correctly under some circumstances and incorrectly under others; and for other, data can become executable.
    4
  49. I wager a guess that be Watanabe Yasushi, Tohoku University.
    4
  50. No.
    4
  51. After people lined up to buy Windows 95, they lined up again several hours later to buy more RAM. But usually found themselves returning empty handed.
    4
  52. Timeline. Apple 2 is one of the first personal computers. Eastern bloc ZX Spectrum clones weren't really a thing until 1988 or so, way after Bulgaria chose to enter personal computers. But of course others made their entrance with teletype based CP/M business machines instead, based around 8080 or z80 clones, way before ZX Spectrum clones.
    4
  53. Do keep in mind that this vid exceeds the target word count in spite of already being very dense and to the point.
    4
  54. Not sure what E is doing there, but it does support the other wafer gegen motion.
    4
  55. Corollary: TSMC needs to bring Taiwan food over to US because Taiwan is extremely tasty.
    4
  56.  @ffls775  There's a large semiconductor industry, yes, automotive, industrial, power semiconductors, sensor devices, specialised stuff. But maybe a little too focused on legacy products and a little aging if my impression is correct? A lot of these are long running products though that don't change all that much decade by decade, new engineering insight has been drying up because the product is well polished, and the well-worn processes are better than fresh ones for a number of things. Legacy customers too, the likes of Toyota and Aisin don't necessarily like change. But just go ahead look through product catalogue of Toshiba semiconductor, Rohm and Renesas, these should be the largest ones. A number of smaller companies are now in Taiwanese hand but should still be manufacturing in Japan i guess.
    4
  57. You know i was thinking of doing evaporative watercooling for my computer back in the day, so Pentium 4 era. I figured i would end up spending less than half a litre of water a day to run it. But i didn't want the loop to get salty, so i didn't do it.
    4
  58. I won't tell you where i'm from but i don't answer where i'm from either. Because it never ends well, people end up trying to westsplain me that i'm from a different nation than i really am. I might volunteer this bit of information on my own, but if somoene is asking, they probably don't have my best interests at heart. If it's their first or only question, they are definitely up to no good.
    3
  59.  @raygumm  All hail the almighty algorithm!
    3
  60. I think once you reach adulthood it's generally stable and doesn't change much, and it has age farsightedness to fight against. Honestly i prefer myopia to farsightedness. Good glasses can massively take the strain off you, have them adjust yours to screen use by adding a tiny bit of compensation.
    3
  61.  @Núi_Lang_Thang  No. Why. It's not a solution to anything in particular, i don't have ambitions of solving world myopia or anything like that - though i'd love to be able to do more work and stuff outdoors in summer, so i don't have to skip all the sunlight. I also don't suggest removing anything. You can have a more powerful laptop/pc/etc but having an extra netbook you can use in sunlight would be awesome, even if you aren't gonna watch movies on it - you can still do a lot of text work. If you have a better solution, please share. I am lead to believe that full colour reflective or incident-light display is technically impossible, physically impossible.
    3
  62. Just don't put the MEMS oscillator in helium.
    3
  63.  @falsch4761  A feel-good button. I hope YouTube implements it someday for real. How hard can it be? I see the same texts over and over again!
    3
  64.  @empatikokumalar8202  Have you invented a transistor with sub 0.1ps tr? If not, why even mention teraherz?
    3
  65. I'm lead to believe that a number of MEMS devices are indeed MCMs - multi chip modules, such as 9-axis IMUs. But i'm not super certain. And yeah putting an existing MEMS onto an MCM leadframe makes sense i suppose. But also it may be unnecessary. Have you seen WLCSP packages? Like damn.
    3
  66. That has negligible effect on legacy chips, where the real shortage is. It will lessen the pressure on the cutting edge chips though, but it will not decrease actual manufacturing volume of these chips, just more of these products will be available to regular consumers. To that end, it will not affect the shortage as an industry wide problem affecting all dependent industries at all, or rather it will, but only after those lucrative markets are saturated, so years later!
    3
  67.  @tomlxyz  There was a shortage incoming. There hasn't been enough capacity provisioning to meet demand medium term. Even without the bursts in demand due to work-from-home and crypto boom(s), the capacity would run out shortly anyway, maybe not in late 2020, but in just a couple years. Every year, people need slightly more flash storage, more cloud servers, more this and more that, now fridges have computers too for some unfathomable reason. Overprovisioned capacity is, to an extent, a good thing. Sure it just costs money when it's not needed, but not having capacity when it's urgently needed can be even more devastating. Whole industries depend on this capacity being present. Besides, the big growth cycle right now isn't actually that big relative to the existing size of the industry.
    3
  68. Oof that topic. Unfortunately the research is reeeeally shaky, and the public understanding of it is even worse. Odds are, if you've heard of microplastics, the only thing you've heard is that you eat 5g worth of plastic every week in form of microplastics. It seemed like it required a little explanation to me, so i followed where it came from. The figure actually comes from a private report done to the WWF Singapore by scientists of University of Newcastle AU. Then, WWF produced pamphlets with the wording that people "could be ingesting approximately 5 grams of plastic every week on average". Subsequently the scientists, Kala Senathirajah et al produced their own paper with the underlying methods and conclusions. The paper has two halves basically, first they estimate the number of particles an average human ingests, and the sources and pathways of it, it's meta-study or a survey of existing research and it's actually pretty good, it comes up with a figure of 2000 particles a week with a confidence window of an order of magnitude. Continued in the next comment because i'm running out of comment length.
    3
  69. Continued... The second half of the paper estimates the weight then, by fitting different solid shapes of up to 1mm in size (wat) to the number of particles, and... well it's really a load of handwaving at that point. The study authors themselves give their confidence range of between approximately 0.1g/week and 5.6g/week, with their best-effort estimate being 0.7g/week. And given their methods, i think they are undercounting their confidence interval, i think it's much worse than that, with several orders of magnitude of actual confidence. Really with existing data, no good statement could be made on how much plastic in terms of weight one ingests. That WWF chose to run with a 5g figure should tell you something. They presumably requested this estimate and then they took something near to the upper bound and ran with it, it's purely a scare tactic to drum up more funding. Which i'm a little ambivalent on, given yeah more funding is needed; but i wouldn't entrust WWF with it. It erodes the public trust into science and eco activism both, since it's fundamentally a dishonest tactic; someone at WWF will pat themselves on the back for a successful campaign, for winning a battle, while edging closer to lose the war. I also think lack of good weight data is not entirely a coincidence, due to limited relevance of bulk weight of small particulate, as environmental erosion and chemical interaction are surface based. For weight, what you want to know is how much (macro)plastic you put into the ocean, and that's hardly a secret. Microscopic surface estimates are difficult, but particle count might just be a good enough proxy. Just not for weight. Like you request useless data, you get useless conclusions.
    3
  70. But the video hasn't "aired" yet, thus i had no opportunity to enjoy it.
    3
  71.  @johndoh5182  50TB? Probably on 12 platters. Not cost efficient. Dying industry's last spasms. I deem it inevitable that price/gb will reach parity in about 5 years between SSD and HDD. And then SSDs will win based on power efficiency. The prices of HDDs stay constant or increase; while capacity per platter went from 1tb in 2011 to just north of 2tb in 2021, this is GLACIAL compared to prior and predicted SSD progress. None of the major industries actually care that much for HDD in principle, they only win on cost per gb for now. For now. They'll jump ship when the price is right. And once they jump ship, HDD prices go up, as economies of scale collapse, and i predict LTO tape comeback.
    3
  72. What advantages does density bring beyond purely more semiconductors per unit of area? Gate capacity (power usage), propagation delay? Will some of these start scaling worse with density or have they started already? What about leakage current? Because i have a feeling that a lot of the time, the device size is hardly the goal, as the packaging takes up too much room anyway for it to matter, and we're well past bottom price per FET; and for large and powerful chips, power density is becoming critical anyway, like if a chip that takes like half a kilowatt, ridiculous, how do you even cool that without having it turn into a Pringle?
    3
  73. Wow you have been at it for long. My first HDD was 500MB. But my computer before that used primarily 5.25" floppies, and before that audio cassette tape :D
    3
  74. NVidia's weight in HPC has become so enormous, this little niche of semiconductor design doesn't really move the needle much one way or another. AMD could get their shit together but doesn't; and of course there's Intel with Xeon Phi but they think that whoever hasn't run away to NVidia cannot move, that they have a captive customer base. I don't know that this sort of complacency is a good way to go about things, it usually doesn't end well. If anything this here goes to show why NV wins, because they keep pushing the field forward, they are not complacent not even for a second. Perhaps they deserve this monopoly, who knows, even though we all know it's not good.
    3
  75. Well Russians shouldn't have a say in which political/economic (or indeed defence etc) alliances nearby nations join. Not to speak of the fact that Finland and Baltics are EU and adjoining Russia and Poland is adjoining Kaliningrad, while Moldova does not share a border with Russia. So like that ship has long sailed; and if you want an alliance, you can't just do it by force, you have to offer something better than another alliance in return. If Russia doesn't have much to offer besides blackmail and passing stinky gas well it only has itself to blame for lack of economic political cultural and scientific development.
    3
  76. We need a collab between Asianometry and Huygens Optics.
    3
  77. Yes they are by far the most complex subassembly of a computer, that you can get them for like $60 is kinda outright astounding.
    3
  78. Yeah my early-2000s fridge is semiconductor-free as well.
    3
  79. There's an exploding need in total battery capacity needed to buffer renewables and so on. Not every use has such high requirements as an EV, and every alternative made from cheap and abundant materials can help decrease price of Li-Ion batteries, since there will be less competition for the limited and expensive resource. Something that i expect to become fully electric in foreseeable future is municipal public transport. The bus traverses 150-700m between stops, and then halts for a minute in an area with dense infrastructure, often with longer 5-10 minute halts every few km, so fundamentally giving the battery a little boost while it's just sitting there waiting for people to embark or for the schedule to catch up makes sense. Since a diesel bus idles a lot, it is a fundamental energy saving. It would also improve the quality of life of local residents, due to noise and vibration, and improve public transport acceptance. You can feel the vibration of the bus diesel through the foundation of nearby buildings! But the volumetric energy density is of little regard to such a vehicle, especially if it can be boosted often.
    3
  80. Parhelia, aka too little, too late, too expensive.
    3
  81.  @alexforget  Not much difference really, i would say...
    2
  82. Times Square ads don't have much reach, they're merely a status symbol.
    2
  83. There's a little problem with that - there's only so much you can shrinkify a chip with mere several tens thousands of gates before it gets dominated by unshrinkable padspace and IO. But i do think some movement needs to be happening in that direction. Maybe dropping more peripherals into the chip, like at least digital IO rerouting, IO statemachines like on RP2040 and new atmegas, small FPGAs and whatnot, just to eliminate a bunch of little junk chips on the board. As long as these are purely digital and use the same pin drive circuits as already present, these should shrinkify right along.
    2
  84.  @effexon  well recovery is going to happen, if it's economically viable at all, you can't even stop it. But you can direct it towards a sensible and responsible use. The problem with counterfeit ICs isn't limited to salvaged ones. The bulk of manipulated components are just a different product with a similar pinout that have been made to look like a product that's worth a lot more. Those either don't work at all or often don't hit even close to spec.
    2
  85. It does not create perpetual motion. It transfers the liquid exactly once on its own for a given amount of material, and then part of the liquid is effectively lost as well, stuck in the fleece and difficult to retrieve.
    2
  86. I do think the world, especially US&Europe has done a MASSIVE mistake by not treating the disease in the first few weeks as if it was exceptionally dangerous and super deadly, just to err on the side of caution, a very temporary true zero case goal. Maybe it could have been stopped dead in its tracks and we would have avoided the whole everything that followed. This is something i advocated back in Feb-March 2020 but i'm a nobody, i decide nothing, and indeed i have no relevant professional background or advanced knowledge, but then, it's not like politicians put epidemiologists in the driving seat on this either; i saw no opportunity that it could have worked several months later as the situation developed. In turn China having to deal with the issue now, is from it being reimported from the rest of the world. Developing world got it from the first world, and there it kept mutating and couldn't be stopped. As to longer term policies, i don't have a strong definitive opinion. It pains me to see people die, and it pains me that some people are so dumb as to bring themselves and people they encounter in easily avoidable increased danger; but the even semi lockdown state with limited freedom is also insanely depressing and has consequences as well, as it goes on for months and years.
    2
  87. I don't know for certain, but as far as i can see here from Europe, diminishingly little or not at all? Chinese semiconductors and rest of the world are like a semi permeable space, like, Chinese stuff has some availability (some product lines are canned) but all the US, Euro, JP, even Taiwan companies have pretty much ZERO availability of trailing edge chips. Chinese electronics companies use Chinese stuff and classic global stuff, after all they learn from the rest, non-Chinese companies use non-Chinese stuff almost exclusively, with very little spilling in, say via Diodes Inc acquisitions and this and that, but really nothing all too important there. These days people around the globe are starting to familiarise themselves with Chinese domestic chips in part even just to have an alternative available, and Chinese companies are starting to write English datasheets more than they used to.
    2
  88. @roro Ya doing business with Chinese suppliers is often a good time, like they'll often find a way to accommodate you, they're quick on their feet, they spot an opportunity and they go for it; while EU companies usually try to find an excuse to just not work with a smaller customer or unusual orders, they find themselves a comfy spot somewhere on the periphery of a handful of large clients and just stick to it. Naturally things sometimes don't work out perfectly but they're generally willing to cut through it and work it out eventually. Good luck to your electronics business, happy you lot have been able to keep it afloat. Hope the politics doesn't get in your way too badly, i know there are... tensions.
    2
  89.  @mathew00  i have the solution... YOU MUST CONSTRUCT ADDITIONAL PYLONS.
    2
  90. The way it started, i was convinced i wasn't going to like this. But you did turn it around into the areas you're more comfortable talking about.
    2
  91. Soviet Union is 12 times the population of Oz, so why exactly you expect Australia to outinnovate USSR is beyond me. Also if you say Soviets innovated, what DID they actually innovate? Space ventures are literally not a product (at that point at least), it's closer to fundamental research. Soviets had all the space tech eventually and did they install sat TV, did they install satnav? They did not, it was military and research use only.
    2
  92.  @PBnFlash  Ah one of those things that are endlessly marketed but haven't actually happened.
    2
  93.  @TheMichaelJu  I am a computer engineer and worked in automotive and medical but no longer do at the moment, but why would you be making it about me. You're skirting the question entirely. SoCs CPUs GPUs DRAM and whatnot are 45, 28nm or higher end. They don't directly compete for the fab space with 130nm 32-bit micro or a 350nm 8-bit micro. And those things are exactly the things that you can't buy. Every typical NXP LPC, STM32, ATSAM, even Nuvoton in the sub-$5-ish range (depends) are sold out a year in advance. Where are THESE chips if you claim there is no shortage in specifically this area?
    2
  94.  @ScottGrammer  Fun fact, even people with extended range hearing cannot differentiate frequencies above about 16.5-17KHz from each other. They all just sound the same. You can even induce ultrasonic hearing with bone conduction in most people, reaching maybe around 50KHz, and they still won't be able to tell the difference. There is also sort quite little acoustic energy up there.
    2
  95.  @empatikokumalar8202  Well if you don't have a basic building block, a logic gate, that can switch at THz speeds, you aren't going to have anything running at a THz clock speed. Can one make a more efficient processor? Sure. But starting off with ludicrous claims isn't helping. And if you did have a solid foundation for what you're claiming, you wouldn't be having the question you have opened the thread with, the path forward would be crystal clear.
    2
  96. I don't see GPUs being obsolete in near future. What happened with soundcards is that they grew ever more complex, synthesizers, gaming 3D audio, all the stuff. Actual processing power. But then, CPUs outran them and it made more sense to just integrate a simple dumb soundchip on the mainboard. By the way, Intel killed HD Audio interface (i2s multichannel + i2c command) that was used previously, now onboard audio is USB. I actually find that exciting, maybe we'll see Crystal (Cirrus Logic) or Burr-Brown return. Crystal is gearing up for sure. Now, CPU progress is slowing down, but GPUs are still trending up in performance, growing with software (game) requirements, into monsters which far outstrip the power budget of the CPU package. Memory interface of the CPU is unfortunate for them too. For sure i see a possible distant future where everything is an SoC, but not yet, not in the next 5 years. Anyone who says they can see longer than that into the future probably lies. Now main GPU uses besides game are crypto and AI, and these are of course just waiting for ASICs to catch up. But indeed as far as standard office computers, laptops as well, no need for a dedicated GPU, but been that way for decades.
    2
  97. There isn't anyting particular about Toshiba laptop is there? Last i looked they were mostly Compal machines like many others, basically every brand carries somewhat customised macines made by Taiwanese ODMs. Which is a story in and of itself, since back then i expect Compaq was one of the major benefactors of Toshiba business being shunned. And guess what Taiwanese company they spun off.
    2
  98. That be an ASRock A790GX with an AMD AM2 CPU socket, so no APU support. It has a Radeon HD 3300 integrated into the board chipset.
    2
  99. It's not. For the chip shown for example, none of its direct competitors that i can think of have good availability, almost all are gone completely, and a bare handful are just borderline available. Many achievable redesign opportunities have been picked pretty hard. If more designs transition to the available ones, those will be sold out a year in advance as well in short order. Industrywide shortage is an industrywide shortage. Your suggestion would make sense if there was a giant stockpile and overproduction of some functionally comparable chips but there really isn't. We probably need more yield per wafer. Maybe same function chips on smaller and cuter nodes. Not just all the 130nm stuff. But when that happens, can packaging even keep pace? Breaking out of an industrywide shortage is HARD.
    2
  100. I can't find your vid on Mead-Conway VLSI revolution, can you help me out?
    2
  101. There wasn't enough capacity to satisfy internal demand. There were some sales within the Eastern bloc but they didn't matter; also fundamentally ICs were manufactured both with the domestic 2.5mm pitch lead frame and the export 2.54mm lead frame, with the latter being used in East Germany and elsewhere.
    2
  102. It's easy to quote Feynman because he's so feynmous.
    2
  103. Russia must this, Russia must that. Russia must get their feet and paws out of Ukraine, or it's gong to hurt more!
    2
  104. That's just the (magnetic) core memory and core rope ROM. Integrated circuits once they started establishing themselves quickly killed them off.
    2
  105.  @jasbrar9873  everydhram I'm shuffling (the lmfao song)
    2
  106. He had many videos on the subject. Netherlands company making lithography machines for every leading semiconductor manufacturer.
    2
  107.  @jan.tichavsky  rp2040 has external firmware and no provision for ROM encryption, which makes it nominally unsuitable for most commercial products. And open source projects often get fouled by its glitchy ADC, which is not entirely surprising of a company with no such design experience, they'll get it right eventually.
    2
  108. Bulgaria is also just about the only country far and wide where there are almost as many women working in tech as men, and more in science. I wonder how that came about?
    2
  109. High expectations? How much vertical domestic high tech do you expect of a city state of 5 million? I think you're doing OK.
    2
  110.  @rafaellisboa8493  Freedom? Not in Russia.
    2
  111. DRAMa!
    2
  112. Because there's a requirement for 180-350nm chips as well as analogue chips and power semiconductor devices designed for an old process. Old production line needs old spec wafers. These seemingly obsolete semiconductors have been experiencing shortages, because as it turns out, they're EVERYWHERE literally everywhere, even the fanciest devices have some supporting circuitry which was designed against existing classic semiconductor devices, and of course since there was no shortage before, there was no pressure to produce more modern replacements of these semiconductors. People keep needing more electronics stuff, so the pressure is now on.
    2
  113. Team?
    2
  114. Oh hey only half a year, not too terrible. Most small microcontrollers are sold out more than a year out!
    2
  115. You probably already have dozens or hundreds of MEMS IC. The most common is a MEMS clock oscillator, instead of large cans it's now tiny barely die sized chips that have saved an insane amount of space, and space is everything in so many use cases. Then a handset typically also has those but also a MEMS microphone or two and of course inertial and magnetic sensors. A number of life saving medical devices couldn't have existed without getting the electronics down to size or the use of MEMS sensors. And bendable parts (flextures) have already changed the world centuries ago, they enabled book binding. So just not really new.
    2
  116. Cryptography has a demand for sources of randomness and fundamentally watches can ve used, just like someone is using a video feed of a bunch of lava lamps. But it's not super convenient, when you can trivially make an electronic noise circuit, you just have to amplify the shot noise from a resistor. Modern processors often include such a circuit for true random seed data. Unfortunately the better the watch is, the worse it is as a source of randomness, and a lot of decorrelation is needed for average watches already.
    2
  117. Very "fake it till you make it". But the problem is in the interim, you product STILL needs to deliver not only a promise but also an immediate value proposition or unique feature. Some reason to own.
    2
  118. The Zeiss he speaks of here is not he same company, as real Zeiss, the one that's still Zeiss, fled to West Germany. With German reunification, you couldn't have real Zeiss and East Germany's fake Zeiss share the name, so the latter was renamed Jenoptik.
    2
  119. SanDisk is Toshiba. When you need good flash semiconductors, they were an unavoidable supplier for a long time, now spun off as Kioxia, and also they have a long running partnership with Western Digital. Another big Toshiba field used to be power semiconductors. But also in some crucial ways it's not the same company that did this, hasn't been for a while.
    2
  120. They can be i2s and have an internal DAC, and they can use acoustic lowpass as a reconstruction filter, so they can be fundamentally be executed effectively fully digitally, without ever coming up with an explicit voltage or current representing the amplitude of the signal, but driving the MEMS assembly with a delta-sigma signal directly. I haven't looked at MEMS speakers, but i have seen MEMS microphones with i2s digital interface and analogue both, though i don't know that i've seen a manufacturer spill the beans on how they do it specifically.
    2
  121. The progress is growing slower harder and more expensive, but we appear to be far enough from the hard limits to be busy for a couple decades yet. Asymptotically approaching?
    2
  122. Clock oscillator, microphone, accelerometer, magnetic field sensor, pressure sensor. A lot of semi mechanical devices can be made smaller, higher performance and more efficient, produced as a MEMS semiconductor.
    2
  123.  @Rubicola174  I actually wonder why there hasn't been an Asianometry about Kuai-Kuai.
    2
  124. Can you give a better set of keywords or a full title?
    2
  125. It's a question of time and cost. You buy a tool, you jump easily 5 years ahead in the process, and if you think about the other tools involved, it can compound to more years. Tools are relatively inexpensive like that since you need a handful to make for a lot of progress. You can build a tool in months but to refine it and to iron out the niggles can take an arbitrary amount of time, the time and effort that then takes away from achieving the end goal. It's the work that someone else has done and that desn't need to be repeated.
    2
  126.  @LMB222  It's not that ESP's chips aren't used at scale, a lot of iot power products like colour changing lightbulbs and remote switchable power sockets used them at scale; then again these are products with... shall we say no particular expectation of reliability, often just borderline compliance to basic safety norms, and many of those have even since switched to even cheaper and wonkier chips, they plain don't care. You simply can't sustain a semiconductor company like that on hobbyists, but you can sustain it on kilotons of future e-waste.
    2
  127. Huh he sounds like a typical Canadian to me.
    2
  128. It's a horrible waste of money. We have seen that in Russia. Massive amounts of money and no output. Plus the expetise of the institutes maintained this way is... debatable, since their hands-on experience is with long obsolete technology and doesn't include overcoming scale and quality obstacles. The premise that you can have a handful of experts is flawed, since production of semiconductors requires combined expertise of thousands of people, whether these people actually do something or only sit on their hands because they don't have equipment funds, a customer, etc. Every person is capable of developing and internalising experise in only a very small part of the process. Then you have brain drain, you teach up people as best you can, but after a while sitting on their hands, people get frustrated and up and go somewhere where they can develop their skill more and they never come back. Hibernated insular industries are extremely wasteful. You either need to leverage someone else's hard earned expertise and economy of scale, or you develop your capability to competitiveness, either way you need to slot into the international ecosystem.
    2
  129.  @thealienrobotanthropologist  Ironic: you're commenting this on a channel which is mostly about how we made small grains of sand think with electricity. We made THINKING SAND. And to think it wasn't considered promising because what could it possibly do that vacuum tubes couldn't? And besides the properties of this new material were kind of garbage and very unstable.
    2
  130. How does China manage? Ok you have 3 or so major industrial areas connected by sea, nice, but the inner land is increasingly industrialised, it runs on rail, it seems to be doing fine. Well rail is more expensive; but it's quick and reliable, there's value to be extracted from that, lots of. I also don't think 12 times is even the right order of magnitude. Like yeah if you add up all modes, it might just add up, especially if heavily weighted towards road transport. But rail should be only 2-3 times more exppensive than sea, i would think.
    1
  131. Were the lines shut down? Recently - i don't think so. Most of the 200mm shutdown happened decade+ ago. The industry was moving to 300mm, so it wasn't a temporary problem, it was lack of future viability seemingly. Except then it turned out everything needs legacy semiconductor devices huh, and the total consumption of electronics keeps growing. Hindsight 20-20. By the time it became evident that there may be a problem on our hands several years ago, there wasn't much to be done about it other than just to try to move things to fresher nodes. Try. And i mean temporary problems which last for 5+ years aren't so temporary if there's no money to be made in the interim, you run dry, you run bankrupt. Where do you get money from, what kind of an investor pitch is it when you say "help us maintain ancient obsolete stuff, trust me bro it'll matter someday again, just a rough patch bro, we know it's looking like what we're making is useless in 5 years but we just need to sit it out bro". Line on a chart going down tends to be a death sentence.
    1
  132.  @UpLateGeek  I am PAINFULLY aware of 2 year lead times, you don't have to tell me. I don't think anything substantial was shut down during pandemic, or i haven't heard of that. I know car manufacturers have dropped some orders, and then scrambled to get them back, but it wouldn't cause fabs and suppliers to get closed. They couldn't get new orders in because the capacities were already booked by some other semiconductors. Fundamentally, it's unnecessary that anything got shut down to get us where we got. Ingredient 1, is that electronics consumption keeps growing year to year by maybe a couple percent. 2, everything needs legacy components, jellybeans, classics, which are manufactured using old tech and which are not economical or impossible to move to new tech. 3, so as total electronics demand and demand for modern electronics is increasing, it drags alongside with it a demand in legacy components upwards. 4, in 2020, worldwide electronics sales shot up by 14%, which is WILD, pretty much never happened before; and the demand is remaining high even now. Given the huge jump in demand, some end device manufacturers cancelling their chip orders doesn't matter and wouldn't cause production stops, as the total demand increase is more than high enough to absorb any capacities that were in use previously. I have heard a conjecture somewhere that the problem with legacy component production capacity was a long time coming simply because demand keeps growing, but the legacy capacity is all we've had for a decade; but it wasn't expected to actually hit until 5+ years later, and more gradually, and was pushed to now and urgent by a sudden jump in demand. Do keep in mind that we have some overcorrection on our hands, that electronics manufacturers which thought that they didn't need to keep extra stock of jellybeans and were buying them on a weekly basis as needed (Just in Time manufacturing), now suddenly do, and many have decided to keep a year of advance stock that they're currently filling up; so the situation will take a long while to relax a little.
    1
  133. Electrical Engineering has a number of arcane units, and for better or worse, they're going to stay like that. Some are explicitly confusing, like 1 Mil = 1/1000th of an Inch. And you have a wild mixture of component footprints some of which are designed in millimetres and other in Mils, so good luck choosing a layout grid that you don't hate. We don't love it but it is what it is.
    1
  134.  @punditgi  yeah it's down to parts and when you design something in Europe, you can't just be designing in European parts you know, you're going to have a mixture of American, French-Italian, Japanese, Taiwanese and Chinese semiconductors, and in turn they use standards borrowed from different company standards from somewhere else in the world entirely with a different design culture, the footprints are pretty universal across the world and they are universally a mess across the world. Just some company did something and then the rest did something footprint-compatible and it became a standard eventually. Soviet Union tried to put things on a millimetre grid but that was even more infuriating, since you might have functionally compatible and similar looking chips with both 2.54mm spacing (international) and 2.5mm spacing (domestic) and domestic chips were occasionally produced in export leadframe so you never know what you get and the subtle and prone to bad contact misalignment on a 40 pin socket oh my God, and I can't even bear the thought of a 64-pin chip going through that sort of torture... Oh yeah that kinda still a thing in the rest of the world, with some connectors having same diameter pins but pitch of 2.5 or 2.54mm, so best not confuse the footprints or think of a last minute substitution... But at least that's not chip sockets. Newer footprints tend to be based on metric but we couldn't get rid of occasional Mil footprints in 35 years... So yeah that's how it stays. You can tell I built my first computer in Soviet Union - by soldering it together chip by chip. And I had a curse of finding only an imported Z80. But only domestic DRAM. It was the worst combination.
    1
  135.  @autohmae  It will change once price/GB of SSD reaches parity. Thing is HDD capacity per square inch of platter space has been barely growing recently, not much investment in the tech. The price can no longer be reduced either. SSD price/GB will continue free falling for the foreseeable future. The parity point is likely within 5 years. One thing SSD really dislike is being disconnected from power for weeks/years on end. But most storage is daily used devices, NAS, servers, they are all fundamentally suitable for SSDs, just waiting for prices to nudge their way down. Server use is also sensitive to power consumption, and an SSD is so much more power efficient. A disk in a server runs for typically about 5-7 years before replacement and this timespan is easily covered by SSD just like HDD.
    1
  136.  @autohmae  When powered, SSD have excellent long term endurance, not really substantially better or worse than HDD, but they do need to be able to auto-refresh the data once in a while, which consumes just a little percentage of their RW cycles, thus they need to stay powered on regularly. The reason HDD are used for cold storage is because they survive that nicely and are cheap. And they are cheap because there's a huge mass market that needs them for their sheer capacity per $ as online, powered up storage. Once that market collapses due to price parity, HDD should go up in price because economy of scale collapses; probably will cause tape to become more prevalent for cold storage. LTO or maybe laser optical. Seagate is one of the forces behind LTO, WD and Toshiba are major SSD/Flash players. In 10 years, capacity per platter went from exactly 1GB to 2.2GB with heat or microwave assist. How much in the next 5-10 years? Maybe 3GB? Maybe 4GB? At increasing expense in hardware, and no way to keep pace with SSD tech that is in engineering stages and where expense is trending down with no roadblock in sight.
    1
  137. But oversupply is a problem for the semiconductor industry, it's a blessing for the electronics industry. As it stands now, the shortage is short-term a benefit to semiconductor industry, long-term hazardous to semiconductor industry players who drag their feet to the benefit of those who are quicker, and extremely painful to borderline murderous to electronics industry.
    1
  138. I, for one, love semiconductors and won't tire of listening about them.
    1
  139. I commend his commitment to cute.
    1
  140. We've been at this stage for a long time. None of the advanced chips of today are possible without advanced computational optics.
    1
  141. Started in mid 90s, by that time the tech gap was no longer bridgeable.
    1
  142. What lame story? Demand is suddenly up 20%+ worldwide and the industry hasn't been prepared with scaling. Suddenly so many people needed a new TV, maybe a game console or computer, more gadgets, more LED lighting. The demand sure keeps rising year after year normally by a couple percent and a semiconductor shortage was about to happen anyway pretty much but it was spurred along and amplified by recent world events. Semiconductor shortages have regularly happened in the past as a cycle of overproduction and overconsumption is built into capitalism, you can see it in anything, but they were never this painful, these were minor adjustments rather than industry stopping events.
    1
  143.  @cheongwenpa  a lot of 200mm wafer tech was decommissioned over the last 10 years. And then it turns out people still need now-old stuff and not less but more of it, because it's been the lifeblood of countless hardware designs. But the leading edge having better availability is fully expected anyway. Because you the end customer are paying like $1000 per sq. inch of semiconductor as retail product, so there's some room to adjust, and do keep in mind CPU and GPU prices went up over the last 5 years. Try getting the same price out of less space efficient 130nm microcontrollers and even less space efficient power semiconductors.
    1
  144. Ah, easy way to ensure that the son will be bullied in school. I helped out at times in an intrnal publishing division of a scientific research institution. I remember Xerox Ventura; and then QuarkXpress.
    1
  145. On the one hand, you say you can't judge complexity by lines of code. Presumably because line boundaries can be arbitrary. And yet i say you can judge complexity by lines of code. Because it's natural for a line to encompass one complete simple thought, akin to a sentence. You would n't sepa rate things into lines like this because it hampers reading, you wish more information density. And in turn, allow me not to demonstrate, you wouldn't want too much in a single line. Also, whose face have we revealed this time?
    1
  146. Tthis is among the topics which i have been curious about for quite a while!
    1
  147. Well where are the chips then, that have 50, 55 week lead times, according to you?
    1
  148.  @symbolsandsystems  the chips in question are hardly suitable for a secret purpose. I'll grant you though that their underlying manufacturing capacities are suitable, but also, a lot of people would have to know and notice, and no whistleblowers seriously? The way I see it the shortage was a long time coming and it's a part of natural boom, overproduction, recession & scarcity cycle that forms the fabric of capitalism. You have to think of the 20s where an unrestrained boom caused overproduction caused a whole economic crash called The Great Depression. The current scarcity was a long time coming and was pulled forward by a couple years and amplified by recent events, when the global demand in household electronics jumped up a sudden and unprecedented 20%. But that's just my opinion, I don't have actually hard evidence to say that this is how it is. And I wager a guess, you don't have hard evidence for your position either.
    1
  149. I think successes are only a part of the story of how accomplishments came to be; the often forgotten part is the failures. The scientific and business endeavours that attempted working in the same goal field effectively pruned the field of approaches until the most viable ones remained, where a prediction wasn't possible, so even when they failed, they contributed to the end goal.
    1
  150. Its larger than Singapore and it wouldn't have to compete against China if they had just some industries which were one or two steps ahead.
    1
  151. Or indeed fortunate.
    1
  152.  @richardlumleysmith2843  mhm in part tech gets abandoned. But say power semiconductors, mixed mode devices, are such a gargantuan part of semiconductor market, perhaps the majority, and they are produced on decades old tech.
    1
  153.  @baronvonslambert  Of course a degree of incompetence is responsible for famine. And yet there's another layer - the state CHOOSES who suffers from it and how. And also global food shortage is not really a thing, there's always food SOMEWHERE, and we can transport it, rail and sea trade was already quite capable of dealing with this. Also... corruption and laziness? Corruption yes, its corrosive effects go deeper than you can even imagine, but trying to explain anything, anything at all with laziness is just... lazy.
    1
  154. There is actually a zero-tuberculosis policy in a lot of the world, especially used to be commonplace before modern medication came about. You would be contact traced and there would be a trace and quarantine for everyone that you got in touch with. There are a lot of communicable diseases where zero case policy is actually implemented without it really affecting the life of society as such, since it's literally zero to single cases and the measures can be designed accordingly. Zero HIV might not have been politically desired. Also if you infect someone on purpose or by avoiding the necessary measures while you know you have it, you can expect several years of jail time.
    1
  155. Is it your first time watching Asianometry?
    1
  156.  @FelonyVideos  It's probably my first exposure to µ-humour. It's so small, it's not even funny any longer.
    1
  157. Even if it's tiny, the growing population will still need a growing network. Higher density of the network, more routes that may now be considered redundant but in due time will be necessary to take up the capacity.
    1
  158.  @Cotswolds1913  Honestly the population pyramid looks about the same as typical developed regions and countries, and most of them keep growing, like Germany doesn't look much different either. It's still an attractive region in worldwide comparison to move to, in spite of structural issues! Which is how they keep growing in spite of low fertility. Between 2020 and now, HK grew from 6.7 million to 7.5 million, and every year except 2003 and 2020 (no data for 2021 published yet, so that might be another one), the population has increased by an average of about 0.6%, but it compounds into a substantial total growth. I got the population figures from world bank data. I don't think you can conclusively say that the population is going to stagnate from now on.
    1
  159.  @sciencecw  I am aware of people who moved away for this reason - but still the population only shrunk by like 0.2% in 2020, which is not that much. There's probably plenty of people who don't mind taking the place of those who move out. You're only looking at one side of things. Chinese nationals don't have to be afraid of being less free in HK than they are on the mainland for example.
    1
  160.  @sciencecw  You literally can't, the pyramid in question is too old (2020, unspecified month) to really account much for the current wave. And the pyramid looks the same top-heavy as in every developed nation.
    1
  161.  @sciencecw  But there was literally no population reduction in 2014, nor in 1989, nor in 1990, nor in 96, 97, so... what does it have anything to do with the topic of whether population is growing or not? In fact in 96 the population grew by well over 4%, which is remarkable. Like sure some people go away, others come in to replace them, because the region is relatively prosperous, and these new people have different circumstances and aren't afraid of whatever drove the others out. Right now you have two things overlapping, the epidemic and its effects on the world economy, and political shakeups specific to HK, and i think it'll likely recover and keep growing slooooowly. Every exodus wave stops eventually.
    1
  162.  @autohmae  I might be mistaken, but Japan beat the odds and did almost the impossible spinning up its semicon industry from basically zero somewhat late in the game. Korea, Taiwan and now China have been able to buy into Japanese and other foreign export systems available and grow from there. Ultimately you need a massive heroic level commitment from the government with a lot of money on the table to get things moving. I think English proficiency in general population is neither here nor there, the relatively few key professionals needed to spin up such an industry can acquire technical, non-conversational language with a modicum of effort in comparison with the technical and scientific knowledge that they need to build up alongside it, and it might need more than one language as things stand.
    1
  163. I don't know, are the dies produced domestically or outsourced to Taiwan? But also have you tried to buy an STM chip recently? Impossible. The company is falling hard on its face trying to satisfy demand. You can order chips with lead time of 50 weeks! First come first serve. I'll be moving my projects to CSK clones and RISC V for sure.
    1
  164. A year is too long. By the end of it i barely remember what was at the beginning. I guess i can just rewatch the vids.
    1
  165. These both stories sound like myths. One is that Japanese labour is expensive while materials are cheap; so disassembling keyboards wouldn't have been profitable. Disassembling is actually more effort than assembling. The problem with the other one is that it describes Clean Room re-engineering which is a way around copyright... but Bulgaria did not allow enforcement of international copyright on its territory until 1996, nor did any communist country in the 70s and 80s! So there was no point doing that, you could just steal the BIOS as-is. And if you have business partners that demand you buy their BIOS or they won't ship you something else that you need, then you can't weasel yourself out of that application via Cleanroom.
    1
  166.  @msimon6808  Sometimes they do, sometimes they don't. I don't think low-ESR electrolytic caps are a good idea to begin with, since they become standard-ESR caps with time or even no-caps in my experience, with a few nice exceptions. When you need to handle high ripple currents, you might as well pair a standard-ESR electrolytic with a ceramic of maybe 1uF or 100n, where the ceramic will take up the bulk of ripple current and the lytic will take up the bulk of capacity. When used correctly (which, i know, is maybe a little much to ask for of mainstream manufacturers today), even cheap standard aluminium electrolytics are generally very reliable, which is not said easily about low-ESR types. It's probably mostly a cost saving measure, since i had to specifically buy 10uF 25V ones in 1206 what a coincidence, and they cost... like a substantial amount of money, several cents per cap, while even low-ESR lytics from some off-brand cost nothing. The ceramics have their own pitfalls as well - the capacity is different every time you look at them, and they often fail short, presumably due to defects or external thermal or mechanical shock rather than electrical abuse. Companies aren't really interested in whether something will survive beyond 2-3 years, but they do need to keep warranty period failures fairly low.
    1
  167. Almost all MEMS devices are i2c if they're low data rate digital and i2s if they're audio specific and digital, but analogue devices exist as well. Their packages and pinouts tend to be engineering driven rather than standardised, especially since the industry is pushing for smaller packages, you can have packages the size of the actual die pretty much, but that leaves no room for standardisation. A neat thing about i2c is that you can have a single bus of 2 data wires running throughout your whole device and all the sensors can be on there. Indeed there's an i2c bus running throughout your PC called SMBus which contains a great number of sensors and also your SDRAM SPD chips and it's also exposed onto PCIe connections. However most sensor ICs are not SMBus compliant, they cannot hotplug themselves an address, you can't auto detect them, you just hardwire them an address from the available ones via pin pullups/pulldowns and then do a register read, about a few seconds of looking in the datasheet and hardwired into your firmware according to your hardware, more complexity really isn't generally desired by the hardware designers. What use-case are you thinking of? USB HID is suitable and Bluetooth (LE) GATT is intended for sensor use.
    1
  168.  @vputilov  As long as there are no address collisions, and every address corresponds also to a particular type of device, you can treat i2c devices as polled hotplug. The addresses aren't software determined, they are determined in hardware at board layout stage, or the device can include address set jumpers. You can also get SMBus to i2c chips which handle just the dynamic address assignment (the adapter chip talks to a sensor on a fixed address, and to host it just gets a valid address from the arbiter) and device type query.
    1
  169. The lot of it, an unsawn crystal? Do you think you can afford it, do you think your ceiling can hold it?
    1
  170.  @samgeorge4798  You could ask the office of your municipal chief architect; he will direct you to the paperwork for your house filed with them, and you can get hold of some qualified architect to interpret it for you. Did you mean this by "one way to find out"?
    1
  171. Lithography is my favourite family of topics on this channel; so don't be too discouraged.
    1
  172. JRC4556 is actually my waifu.
    1
  173.  @middle_pickup  Yeah a higher current audio opamp, 70mA sunk or sourced officially, doesn't start misbehaving until you try to run about 110-120mA through it.
    1
  174.  @middle_pickup  yeah I know, I know. But just about every piece of hifi gear if it's a receiver or amp has a 4556 doing the headphone output. Incidentally usually a 4558 or 4580 as well, doing the line driver or input duty, alternatively the 5532 though. But invariably made by JRC, though most or all of these aren't their original designs.
    1
  175.  @alexanderswift6362  Typo: should be Stahlbaron, probably.
    1
  176.  @Asianometry  Hey i enjoy your nerdy goofball personality and disagree with that person. Life's too serious anyway.
    1
  177. If there's no shortage, then where are the chips though? Have you tried buying a reel of them through any reputable distributor? Or even half-way reputable? All these microcontrollers made on 90-350nm nodes are various shades of gone, with typical lead time of a year.
    1
  178.  @trudyandgeorge  actually a fair number of circuits support automatic output muting to gnd; and some others have enough precision to actually drive very low magnitude signals quite precisely, though that's actually trickier. However you also have quantisation noise, from the discrete amplitude values that can be recorded. This noise overlays all sound, and for a CD, it's at a magnitude of about -96 or -98dB, i forget which marks the nominal dynamic range. But we have noise shape dithering, which shifts this quantisation noise into a band where hearing sensitivity is lower, up in the frequency range, and keeping it out of the sensitive midrange. So the perceived dynamic range of CD is actually larger than the nominal figure. Various values have been suggested where one should stop chasing extra equipment fidelity, usually about -99 to maybe -110 dB noise and distortion in midrange. Pain threshold is generally around 120dB but you have to consider that you're not in a noise free environment, hearing threshold is 0dB but a quiet room is around 25dB environment noise. If you wear headphones to block it out, your hear your blood vessels making noise. Music is usually mastered for 80-85dB average listening level, with it being perhaps 10-15dB below peak amplitude, more for classical. Even -80dB noise floor can appear quite inoffensive, usually hidden, if not quite ideal.
    1
  179. They say, hitory knows no subjunctive case; for a reason. The way things played out isn't ideal, but the world is largely alive, and you don't have predictive powers as to what would have happened, you're suggesting a little more than a mere flap of a butterfly. At best a curious thought experiment, a false nostalgia for something that never would be.
    1
  180.  @zodiacfml  I thought the main reason was mainboard chipsets. SiS wasn't doing so hot, VIA was technically a competitor and was drifting away and anyway too big, ULi was absorbed by NVidia so they had to act quick to ensure supply of vital mainboard parts. Having viable graphics division that didn't need to be resurrected helped too. AMD per se before acquisition doesn't care whether they beat Nvidia or not, they just needed to make sure they're not buried by Intel.
    1
  181.  @simplemechanics246  Different video delivery method. Androids only have H264 hardware, so YouTube ships only H264 video to Android and other mobile-like devices. But on PC, they know you have a bigger CPU, so you get VP9, AV1. VP9 is something in particular that is Google property and GPUs have shown some resistance to adopting it, and AV1 also has hardware compatibility issues. So often you get a software-only decoding, CPUs are slow at this and shovelling that much video data to GPU is hard. Then differences in surrounding software, webpage with control overlays vs. an optimised app. Google controls, understands and optimises the whole Android ecosystem, but on PC things are "good enough" and are left at that, so a lot of sloppy craftsmanship. As to other video services, there's again web overhead, DRM overhead much more on an open platform like PC - on Android they can just check that it's locked down. Console APUs are scary at around 300W, you can't simply put that in a PC, you need to design the computer around something as powerful as that. Of course it's a matter of time till this power fits into a 125W envelope, at which point it very much can be shipped in a PC, which is why nobody is in a hurry to design a PC platform which can carry a 300W APU. Would probably take more than 3 years, and not certain either whether it'll happen.
    1
  182. Unfortunately it might not do much unless the rest of the world follows.
    1
  183. China has half way burned through the workforce which would tolerate anything you throw at them about a decade ago, and today young work force is increasingly revolting against even moderate inconveniences, or just lie flat, do bare minimum, and Chinese worker salaries and costs of living have approached European ones. Long term, there will be another low wage country to offshore to, the advantages China maintains in the foreseeable future is infrastructure, supply routes, and DFM engineering, i.e. it isn't all that nominally cheap to get stufff done in China, but it's quick and efficient to get stuff done in China, since they've got all you need. This points to a missed opportunity: had we lot taken care of infrastructure that way, we would be well equipped to maintain local industry. That not to negate anything you said, just an aside observation.
    1
  184. Sounds like passivation faults. You find these in Japanese ICs as well but it's very rare, scant few batches.
    1
  185. Not whatever happens to be available, they're made in-house at a 350nm node, it takes years of effort to transition to a new node and fab so i suspect it was never done. Consider, it is not synthesizable, it's a hand taped out hard core; and it cannot be truly synthesizable due to various analogue functions - ADC, comparator, BOD, oscillator, etc, expected pullup/pulldown current and so on.
    1
  186.  @NoNameAtAll2  It's presumably cheaper, and how much benefit are you really going to get from producing a chip with mere tens thousands of gates on a higher end fab? You can shrink the logic but the IO circuits won't get shrunk. Most products, and there's really a countless number of them, need maybe just one very low end chip and some power semiconductors.
    1
  187. Buran was original development and operated differently from Space Shuttle. But it turns out if you want a reusable versatile vehicle that would land from space at an airport, you don't get much leeway on how it's going to look, because physics is the same everywher.
    1
  188.  @IrishCarney  Fair. Buran was fairly ill conceived and useless, there wasn't really a good place for it in the grid of Soviet space requirements, but it was made because "oh look they have a reusable spacecraft, and we should do something similar", just as you say. I wouldn't say that semiconductor push of East Germany was useless. After all Soviet Union was cut off from Western and Japanese semiconductors and semiconductor manufacturing tech suppliers, and VEB Elektron filled a vital niche for that whole region, it could potentially make a lot of bank under the circumstances. It was wasteful and there was mismanagement at hand. I think crucial thing is that already somewhere between 1986 and 1988 it became possible to import Japanese and European ICs into Soviet Union, and they weren't exactly expensive compared to the cost of semiconductor operation, so East German industry had to either adapt (at least get honest and stop chasing increasingly expensive stuff that they can't actually build economically) or crumble. The fact that Soviet Union was opening up and trying to reduce the amount of mismanagement and waste, and the rest of the world was opening up towards Soviet Union, broke East Germany's business model that they bet big on.
    1
  189.  @steffeneilers8530  wait why. I finished school on a Tozai calculator that cost me 10 DM. It was a Sharp EL-506P clone and it was gorgeous. I wish I could still get one like that in good condition. I mean I can get functionally identical ones for like $4 but aesthetically they are garbage.
    1
  190.  @allangibson2408  Russia manufactured export leadframe chip variants with 2.54mm pitch as well. These were needed in East Germany and Bulgaria.
    1
  191.  @allangibson2408  that they did, but they also exported chips to and imported from Soviet Union to make up for gaps in capacity and allocation. East German manufacturing was better geared towards higher end semiconductor products.
    1
  192. Ahahah, going from 4200ti to FX5200, that must have hurt for real. Good thing 6200 improved on that drastically on the low end, if you got the 128-bit version at least. The whole FX series was kinda not good.
    1
  193.  @nahometesfay1112  The vid might have spent a few months in a stash.
    1
  194. Silver is kind of cheap in small quantities. Why isn't it used more in semiconductor products for iterconnects and such?
    1
  195.  @paulschreiber9384  The Pico boards? Probably. RP2040 chips are available in 500 and 3400 reel quantities if that's how you want to roll. Supplies are a little limited, if you want to buy 5 reels, you might be in a little trouble.
    1
  196. Consider that as you're pushing the process ever further towards higher transistor density, defect density increases. A very large chip today is maybe 3cm across, that's some vaguely 60-ish dies for the 30cm wafer. What happens if you have just a dozen tiny little defects for that wafer? Your yield is suffering, you could be throwing away a fifth of your product. Now imagine if you had some gargantuan level chips that are even larger than that, your count of defect free dies will be decimated somewhat fierce even further! So you can't increase transistor density and also increase die size any further, you get to choose one or the other. You want to make transistors smaller to decrease their gate capacity, so that changing their state takes less energy every time. And the problem is making the wafer larger doesn't make the problem go away, the defect density per unit of area just isn't going to get better. So larger wafers would hypothetically save some cutoff waste on the periphery, but they wouldn't actually help you make large dies.
    1
  197. Mhm why isn't L3 a separate unit or unit group integrated with the memory controller? Also why would MCM be a curse word?
    1
  198. I think the most used ancient chip might actually be a TL431. Obviously i'm not the first to voice such an opinion, but i have seen maybe two 555s in a product so far in my hands, but i have had dozens or low hundreds of TL431s in my hands.
    1
  199.  @dinoscheidt  I'll check it out but i assume he found someone to fix his horrible audio.
    1
  200. Yeah but you spin up a chip at an enormous expense, it's super niche single use case, and then after a short time it's technically obsolete such that the next general purpose chip can outpace it, what's the point? This is why niche tasks even heavyweight ones need to be done in software with off the shelf hardware. After you're able to get everything out of the off the shelf hardware such that it really can't get better and have gained a crucial insight about the problem field, maybe you can design custom. It's funny how much things changed. Back in the 80s throwing together a custom chip was that much more par for the course. Of course it was lower complexity and cheaper and then off the shelf options were so limited in flexibility.
    1
  201.  @Enrico-  so basically Bulova.
    1
  202. A 10kg foup sounds not too difficult to lift, and on something with caster wheels, would be super easy to bring where it belongs. I suspect it's something else. Humans, even when they're bagged up in layers of cleanroom suits, are still filthy creatures, which is a bit of an antithesis to cleanroom, i imagine you'd want as few of them in there as possible. I mean humans in an HDD factory cleanroom might be fine, but semiconductors are a whole different calibre.
    1
  203.  @johndoh5182  Several million? A lot of ICs have something like 100k transistors. A lot have only a couple thousand. A lot of ICs MUST run at 5V. A lot of ICs must run at 20V, pass substantial currents, and only contain less than a couple dozen transistors. All of these add up to a lot of wafer die area consumption, but don't have high purity requirements. Can they even be moved over?
    1
  204. Soviet Union was a threat. Its governance was unpredictable and often times irrational, and it had submarines with missiles and nuclear warheads. But for a company from a smaller country, the profit from trade is all their own, while the resulting threat is distributed across the whole world, it's an externalised cost/risk. Also F U, favourable terms for Ukraine? like Ukraine not existing? There was no "Ukraine crisis" until you and your operatives like Girkin, Borodai and Aksyonov snagged Crimea and part of Donbas. If you wanted to not be perceived as a threat, you lot should have learned where your borders are and stayed within them.
    1
  205.  @Bialy_1  You say "personal size computer" which is true and impressive but a little sneaky since it's a mini- not microcomputer, with the corresponding complexity. Still, it's a feisty thing and should have been manufactured and would have been immensely useful and given a good headstart. I just don't think Soviet Union had the money for that and didn't deem digital technology important enough (strategic mistake), and yeah Western companies wouldn't want to tie themselves down to stuff from behind Iron Curtain due to being afraid Soviets would forbid exports or sanctions would interfere with parts import to Poland, leaving the customers abandoned. Better to have something more expensive and slower than a surprise brick. Nobody gets fired for buying IBM. In a small configuration, its performance should be comparable to 6502 running at 3MHz. If you grow it with expansions to cabinet size, it wins on capability obviously.
    1
  206. You don't believe he's actually a Sika deer huh?
    1
  207. Surprising? 2 years is a short time and the crisis has been a long time coming and wasn't done anything about. The boom and bust cycles are pretty much inherent to capitalism.
    1
  208. This is not true.
    1
  209. But FPGA is also an SRAM device, which makes it somewhat slow, space and power inefficient and cost inefficient compared to dedicated silicon at sufficient volume. Back decades ago, you'd have antifuse FPGAs which were essentially programmable ROM based configurable logic devices, but those seem to have died out, having been in a bit of a no-mans-land between in-system programmable devices and dedicated silicon. Before that, there were ULAs, which were mask ROM configurable logic devices. And then microcontrollers and configurable peripherals such as DMAC and if you look at something like RP2040's PIO peripheral probably eat quite substantially into CPLD/FPGA territory for a lot of tasks these would have been used for prior, as they're much easier to program. You're also correct in recognising a configurability increase in insanely complex circuits, and that it's absolutely vital, like PC CPUs contain fixed ALUs but the actual software facing instruction set is implemented as "microcode", a rough and usually barely functional version of which is on the silicon, but it gets updated from the PC firmware on every boot. Spinning up such a device costs a fortune and they are prone to bugs being discovered in the field which need to be fixed. Also so many hardware designs have leaned into SoC territory as time went, like "oof, this is hard - let's stick a processor core in there", and thus everything even devices which you think of as fixed functionality like SD-Cards and just about every USB peripheral have become firmware driven. When you buy a USB-to-serial dedicated function chip today, you find inside a 8051-compatible core and a ROM. As universal as FPGA are, they have also always been a niche device. The promise of FPGA everywhere somehow is always nearer and always further away and this has been this way for like 20 years.
    1
  210. I wonder how long a lead time for a die shrink is. It seems likely to me that the PS2 logic was designed in parallel for the stable node and the future cutting edge node, but of course the future node couldn't be manufactured... until the future came about. It was likely too expensive to make systems on a node which would be out of date by release - beefier PSU needed, more silicon consumed, reliability compromise due to higher temperature, all that. And it was likely too expensive to wait for new process to make early devkits - the cost of opportunity is the highest possible cost.
    1
  211.  @someonespotatohmm9513  I suppose the effort alone is neither here nor there; it's whether if you make a copy, you can make a better price, which can be extremely difficult at times, and then whether you can actually shift enough of a volume to offset that effort.
    1
  212. Well that's awfully reductionist of you. Thing is scientific community has only one goal that it's laser focused on: INSIGHT. It's very good at it, at generating insight. As soon as they have gained interesting insight, they stop. Universities aren't there to make something better, more useful or more accessible. Naturally innovation piggybacks off universities - but it requires both components, it requires both universities and some sort of a driving force to actually get this stuff to completion and in the hands of the people. This sort of force was almost entirely lacking in Soviet Union. Finer points like how you're supposed to tax the results of public funding, or how they are to be used to the benefit of society that funded them, are naturally debatable and should be debated, but it's not going to be a simple discussion that survives any sort of reductionist approach.
    1
  213. Estimated delivery date. I think it's likely to draw sooner in a future update. But yeah I think some bigger customers are using the opportunity now to build stock of their favourite parts so it can take a while to trickle down to the rest of us.
    1
  214. Not very first, sorry.
    1
  215. There existed a number of NKRO (n-key rollover, so ability to register and release an arbitrary combination of keys) keyboards before gaming keyboards came about, and because there is no limit over AT/PS2 keyboard bus to the number of keys held down, they actually were more universal and worked better than modern gaming keyboards. Also it's not quite like every key needs to be input separately, standard keyboards with a not fully disambiguated matrix do guarantee a 2-key rollover and with anti-ghosting logic (which i'd call shadowing logic) are usually made to support 3 key rollover in the usual input situations plus modifier keys. NKRO was valued by fast typists and has always been a feature you'd have to pay extra for though. One example of an old NKRO PC keyboard i can think of is the fittingly named Omnikey. The stock keyboard that came with Commodore Amiga 2000 computer was also NKRO, as were a number of early IBM keyboards and others.
    1
  216. You don't think Soviet Union has top notch engineers and mathematicians who can throw together a program that spits out specialised G-code, or even hand write G-code with logarithmic rulers and punch cards? I saw the tail end of that era first hand there. Things they could develop on paper were well far ahead of things they could actually manufacture. Getting from quantity of 1 to quantity of 100 or more was a problem. I don't have any particular rebuttal to your conjecture that they would have achieved this result with French made machines, i don't know about that. But by all reason you do need more axis for the surface quality, not suitable for simple copy milling.
    1
  217.  @mariusvanc  That was not true when nVidia was formed and rose to power. This is an NVidia story, not a GPU story. Well there was an NVidia in the original Xbox, and it was a decent GPU stuck in a sub-par system, so difficult to say whether that or GeForce 3 was actually better, on paper the Xbox one was better, but it was probably bus-choked. But of course Ti 4200 would come out for PC a few months later, and it was no contest. The second time when NVidia GPU was in a console was the PS3, and it was not good, just not good at all, the only saving grace is that with Cell, it doesn't have to do quite as much work, as a lot of T&L and post-processing could be moved to SPUs. ATI's chip that found its way into Xbox 360 on the other hand was truly the next gen stuff at the time of release, pretty exciting. But then Geforce 8800 came out and wiped the floor.
    1
  218. Your washing machine or your car are rotting children's minds? You better get your mind checked. The shortage affects the hardest chips which are maybe nominally faster than 386, but don't have nearly as much memory, often kilobytes. In high tech electronics, utilisation of 90nm, 130nm chips isn't critical, but it's what a lot of the pretty much invisible stuff runs on.
    1
  219. It makes me sad that we don't have chinatowns in Europe. Well i know there are one or two but in none of the large cities that i tend to be in. Or even not neessarily chinatowns but some sort of cute weird enclaves that straddle the line between a theme park and a lived-in livable space.
    1
  220.  @David-lr2vi  Pressure from the "Green" protest groups and the "Green" party coming to power temporarily, it was their election programme and something they had to push through. Probably not entirely without some external influence. It was decided around 2000. Then in 2010, the shutdown suspension was implemented by the new government, to help achieve overall climate goals. Then in 2011, Fukushima happened, and the delay was cancelled, because of course politicians have to always do something about things that worry people.
    1
  221.  @jannegrey  HPC OGs are Intel with AMD and in the past the likes of Fujitsu with their DSPs, one or another odd entrant. Nvidia is sort of the new guy, sweeping up from under both categories, both microprocessor and DSP based. Though honestly DSP as they used to be were on borrowed time anyway ever since CPUs gained SIMD. And GPGPUs have more SIMD density.
    1
  222. I already appreciate that the audio has improved recently. Up until not too long ago, there was an upper interference band and it was quite grating, but i endured through that. Now sure it's not perfect, but i'm no longer suffering.
    1
  223. "climbing up the value chain" culminates at where you don't have to even manufacture anything per se, you just own stuff in poorer countries, where stuff gets made instead. But it is a long way to get there. "domestic manufacturing" breeds "domestic process engineering and design engineering experience", which eventually allows you to move post-industrial. End goal is not to manufacture, the less you manufacture and the more you control, the better.
    1
  224. sigh why did you have to mention Blockchain, half the comments are gonna be bots again. That being said it's again a solution looking for a problem where a better solution already exists. It's also... crypto, but classic. Web of trust, chain of trust. You receive a certificate with your chips that got signed and augmented all the steps of the provenance. It requires the exact same degree of industry collaboration as a blockchain solution, but is potentially more flexible and won't burn basically any energy since there is little to no guesswork involved - easy to compute, difficult to reverse computations.
    1
  225. There really isn't much of Japanese electronics industry left, not like it used to be. Engineering subcontracted to Taiwan, manufacturing anywhere abroad. I don't know if they can rebuild it like that. Automotive and heavy are holding on. Foremost if they were to become a military power like that, it would be weak to sanctions if they were to choose the wrong enemy, not necessarily threatening. But i do expect that militarization of Japan is potentially starting soon, in order to have defensive capability against China.
    1
  226. We know they're scrambling to upgrade from 130-200nm to 28nm. But there is such a thing as next generationn of weapons to come eventually. 5G peer-to-peer, loitering munitions, smart stuff incorporating AI, lower cost based on phone chipsets just temperature and vibration hardened, 4K cameras. It would be inconceivable that China isn't moving in this direction, it's just so obvious given their tech trajectory. But US and EU have slept right through, they will have nothing to counter that. So yeah maybe it's not fair but it makes sense.
    1
  227. Yeah if you wanted your phone to tell time in the 80s, you could but it wasn't convenient and would cost you like 40 cents for the premium phone call.
    1
  228. Can we read?
    1
  229.  @GuerreroMisterioso95  You think so? I lived in the most calm and prestigious area of one of the most advanced scientific and engineering cities, which has hundreds of thousands of students, and also that order of magnitude scientists and engineers, and every week, there was a deadly stabbing in the neighbouring streets. There was barely any trash removal, it wasn't timely, and there were a lot of rats. Dead cats and dogs could be found everywhere, often burned alive burned by unparented kids roaming in gangs. All the time, the number of memorial plagues in the streets around was growing. I don't know if this is not an apt description of "rot and incredible violence". I don't know how it is now, i imagine it might be better.
    1
  230. I've played plenty of Test Drive Unlimited series and Euro Truck Simulator and I'm actually hopeful. Because the behaviour of traffic is not perfect and somewhat buggy, but so are real human drivers... There's a trap there that the AI learns the bugs of the game AI but still. These tools shown at the end like Synthia actually don't have any of that, they work with manmade scenarios that are known to occur and would need to be handled by driving AI. I think there's a ticket somewhere in combining real world data, data generated by various games each with their own quirks, and data synthesised from manual input.
    1
  231.  @weksauce  Spatial reasoning bound image recognition is a very hard task, which requires by all reason a large neural network. The network needs to generalise from the data set. When the data set is too small for the size of the network, the network generally learns it verbatim instead or focuses on detection of secondary coincidental traits. Problem, we don't have a great understanding and manual access into the function of neural networks. As to ad hoc, hardcoded solutions, they tend to be very difficult to integrate, except specifically using synthetic data. But i would agree a handcrafted solution should still be present as a failsafe running alongside neural network subsystems, specifically because we have such a mediocre grasp on them. But also the two systems should ideally mostly agree rather than fight, as that's a source of danger as well. Synthetic data also helps solve the chicken and egg problem. Like yes ideally you'd have more real world data, and predominantly more diverse real world data with better corner case coverage. But the raw data we can get easily isn't marked up. You can train the markup on synthetic data, and human supervision and correction is orders of magnitude cheaper than manual markup from ground up. Then you can gradually increase the amount of real world data used in training.
    1
  232. You might not want to put them in a car, but you very much might want to put them on a bus.
    1
  233. I wonder, does it matter that the chips that are targeted are huge? They do after all consist of repeat elements, so doesn't it suffice to run computationla lithography on smaller building blocks and just copy it over?
    1
  234. They're very simple products, one step above memory products. Memory is kind of a shit market to get into, you'd be competing against companies that have it down, that have the cost efficiency you cannot reach without decades of experience, and the margins are extremely slim across most of the memory product lifecycle. I think AI chips are a great way to test drive a semiconductor tech that isn't quite ready for other uses, due to how regular they are, you can simply detect and mark faulty cells in software and compile around them, so you can salvage tech that for anything else would have an unusably low yield. And at the same time it's a field that has been growing in importance and software/architectural/research accomplishments by leaps and bounds, extremely promising to get into before it's too late.
    1
  235. I am particularly enamoured by thousands of stock photos of soldering around. Usually with people holding the iron by the OUCH HOT bit instead of the convenient plastic handle. And poking random bits, like reaching with the iron right onto the mainboard of an assembled prebuilt office computer with just the side taken off. And that one company which made an advertisement photo of someone holding their iron by photoshopping it into a hand that was originally holding a wooden pencil. Both the hold posture and the scale was hilarious.
    1
  236.  @danz9268  Knallgas reaction.
    1
  237. What other ways?
    1
  238. Huh it's almost like one better be careful what they wish for.
    1
  239. Does ASUS still engineer its own products, or are they basically a marketing company, who contracts engineering out to Pegatron and others? What about ASRock, another Pegatron daughter supplying computer components? I don't know about Taiwan being unable to cultivate global brands. Isn't Acer such a brand that just about everybody knows? Then again people buy an Acer because it's practical (inexpensive, accessible), not because they like the brand. They have spun off a brand which has a much better image with Benq, but also much less broad awareness. HTC, Gigabyte and MSI have also been successful global Taiwanese brands, but they are very niche brands, nothing of the likes of LG, Samsung or Xiaomi.
    1
  240.  @Graham_Wideman  Those low pin count atsam chips that the video starts off with as an example, you could get them really cheaply.
    1
  241. Does it mean that onboarding new customers onto the currently leading edge node is going to be a spectacularly complex and prolonged endeavour?
    1
  242.  @asgdhgsfhrfgfd1170  Another day on the Internet, another serious rebuttal to a joke comment.
    1
  243. 10 years behind now is actually remarkable gjven just before 10 years ago, they were 30 years behind. Where in the graph it has that 17 year span painted in, what was just before it? Decades of pure stagnation not adequately described by the straight line painted in there.
    1
  244. You get the gist quickly but when it comes to quirk for quirk compatibility... That's generally quite an undertaking!
    1
  245. Sika deer - easily get accustomed to people and will just come up and flirt, they also absolutely love photobombing. Hold on a sec, maybe he did a vid on these deer, you should check.
    1
  246. Curiously, most BIOSes will happily boot a 4KB-sector HDD up to 8TB in size in MBR mode; but Windows might break on anything that isn't 512b-sectored due to bootloader issues, thus mandating the use of GPT. Linux and BSD had to handle large disks long, long before GPT was invented, thus more flexible support. Not sure what kind of incentive since MBR boot has been kinda dead for a decade?
    1
  247. Hey my C64 can do maybe about 100 000 to 300 000 calculations per second on around 4000 transistors. Ok mine is CMOS variant actually so it should be like double that on the transistor count but also it could be clocked higher but just isn't because the design is a little older, that's how it keeps video in sync.
    1
  248. He covers the whole world actually. All of it. He has an inside perspective on Taiwan.
    1