Comments by "Peter Lund" (@peterfireflylund) on "Asianometry" channel.

  1.  @ianweniger6620 no, that’s the wrong conclusion to draw.
    63
  2. So you are the Forest Gump of FPGAs, huh?
    38
  3. Because it is cheap enough now. It wasn’t before.
    24
  4. He did. John Scully saved it.
    24
  5. > A guy had something that acted like a transistor back in 1925 but didn't pursue the concept. He patented it. Later, a myth was created that his idea didn't work. This myth was extremely convenient for Bell Labs. It later turned out, in the 90's, that it actually did work. Moreover, Bell Labs had actually built working devices based on his patents back in the 40's before "inventing" the transistor. Somehow, this hasn't worked its way into the textbooks yet. The same guy also invented the electrolytic capacitor.
    24
  6.  @TheLegitAlpha at least they can still use paper tape and punchcards.
    21
  7.  @Raivo_K the Catholic Church has done amazingly well for almost 2000 years.
    19
  8.  @akaroth7542  no, that’s not what happened at all. I think you are somebody who can’t program (not even in BASIC) and who has paranoid tendencies.
    19
  9.  @bearcb  Julius Lilienfeld.
    17
  10. The Integrated Woz Machine came later. It was a single-chip (= cheaper) version of his original disk controller which mainly consisted of an EPROM + some off-the-shelf TTL chips (+ some help from the main CPU). The IWM was key to making later Apple II and III machines cheaper to produce.
    17
  11. There's another reason or two, Harvey, that I didn't go into earlier because I was on my iPad. 1) we have small gas peaker plants now. 2) we have decent megabatteries now and they are only going to get bigger, better, and cheaper. Thank you, consumer electronics! Thank you, electric cars! Megabatteries can replace (and are already replacing) some of the peaker plants. In the long run, we might be able to get rid of all of them. 3) we have better and cheaper power electronics now. We have transistors that work with much higher voltages and much higher currents. They are really useful for EVs but they are also really useful for turning photovoltaic power into DC for battery charging or into 110 (120)/220 (240) volt AC for the grid -- or for going from grid power to DC battery charging power. The losses are much smaller now and they are much more reliable. They are also really useful for transforming electricity up/down, even without using AC. That means we can start using HVDC cables instead of HVAC -- the losses in HVDC are smaller, especially if they are in the ground or under water. 4) we have small, cheap computers everywhere + we have internet everywhere. That makes it easy to keep track of more details of the grid in real-time + it makes it realistic to turn off certain loads in a smart way when the total load on the grid is too high. Perhaps there are big freezers that can be turned off for a while with no harm, for example.
    16
  12.  @wizardman1313  buy a cheap kit and dive in. Don’t stand on the sidelines. Don’t either. Just dive in.
    15
  13. Erasmus.
    15
  14.  @hieronymusbutts7349 no, that’s just modern American revisionism.
    15
  15. Low labour costs, high pollution, theft of technology, limitless state loans... I wonder if China has even made any actual money on solar cells yet...
    14
  16. The Elbrus wasn’t all that great in reality, because you only got the immense performance if your problem had just the right structure and you wrote your program in just the right way. Everybody knows how to make ALUs, the hard part is to make entire computers that lets the programmer use the ALUs as much as possible… the Elbrus was very bad at that.
    12
  17.  @chriswestwood3289  Hard work, yes. "Own development", not so much... Plus lots of pollution in China due to lax regulations.
    11
  18.  @ThePirateParrot  let’s be clear: you don’t understand economics.
    10
  19. It was originally a Danish company. They hired Kahn to run their US branch.
    10
  20. It wasn’t good at strategy either.
    8
  21. So you are saying it’s a bad idea? You are correct!
    6
  22.  @b.6603 it never worked in Brazil and Argentina and it worked very poorly in China before Deng. Russia is also learning that it can’t do nearly as much as it thought it could.
    6
  23. The German z is pronounced /ts/ and not /z/ (voiced s).
    5
  24. They go back to the 1700’s.
    5
  25. Lump sum of labour fallacy.
    5
  26. ​ @kngharv  Yes. But when you zoom in, some of those incremental improvement are not that simple. There are lots of total shifts of technology in delimited areas. The improved power electronics did not come about because power MOSFETs became better (which they did) but because of a different type of "transistor", the IGBT. And that's just one example. If you zoom out, it looks like incremental improvement but if you zoom in, you can see it is a new technology which happens to share most of its "ecosystem" with preexisting semiconductors and which is compatible with preexisting electronics. Here's some background reading... https://en.wikipedia.org/wiki/Insulated-gate_bipolar_transistor (just look at the table if you are in a hurry) Our grids are AC for one reason only: it makes it a lot easier and cheaper to transform the electricity down from high voltages (and low currents) to lowish voltages and high currents close to the consumers. The power lines are really just very, very long resistors and the loss in a resistor is proportional to the current. The power (energy per unit of time) equals the voltage times the current, so if we can step the voltage up and the current down, we get lower losses. Transformers are good at this job but they require alternating current. Other methods have been used, for example electric motors that drive dynamos, but they are not as efficient. With modern power electronics, we finally have something that can do it better than transformers and without the alternating current. HVDC is useful for transmitting electricity with smaller losses = it's practical go much longer distances + underwater in more cases than before. The DC part also means that the connected subgrids don't have to be synchronized, which they do for AC. That's a huge deal. It makes it much less risky to connect up gigantic powernets because the risk of cascade failures is much lower. And we want our nets to become bigger because wind and sun are intermittent power sources but hopefully the wind will be blowing somewhere and the sky will be clear *somewhere*. It also makes it easier to use pumped storage for "peak shaving" -- pumped storage usually requires mountains and low population densities but the usefulness requires industry/cities. Better electronics (computers) and communication (internet) also makes it easier to keep synchronized grids running. https://en.wikipedia.org/wiki/High-voltage_direct_current We have a big connected grid in Europe and we are trying to make it bigger (but less synchronized): https://en.wikipedia.org/wiki/Synchronous_grid_of_Continental_Europe https://en.wikipedia.org/wiki/European_super_grid https://en.wikipedia.org/wiki/List_of_HVDC_projects#Europe Pumped storage: big lakes (possibly artificial) at high elevations + a big height drop + turbines that can turn into pumps. https://en.wikipedia.org/wiki/Pumped-storage_hydroelectricity It's currently the best/cheapest energy storage method but it is not feasible everywhere + we have more or less built what we can in the West and some of them do have a local environmental impact. Other options for energy storage is batteries, of which lithium-ion is clearly the technology to bet on. Lead iron used to be cheaper (and might still be cheaper for a couple of years more) but lead is really unpleasant whereas lithium-ion batteries are safe enough to put in landfills (but can also be recycled). Lithium-ion have had a superior energy density for a long time but the prices have been too high, particularly when taking the low cycle count into account. Every single charge-discharge cycle wears a little on the battery and lithium-ion used to have a limit of a few hundred cycles. This has been improved immensely in recent years, mainly through the use of better additives. The technological enabler for this was a better measurement device that could measure microscopic wear on a battery over a single cycle. It's basically a really, really, really good amp meter. Prior to that, researchers had to perform many(!) cycles in order to compare their additives. This device was developed in Jeff Dahn's lab at Dalhousie University: https://www.dal.ca/diff/dahn/about.html There's a reason why Tesla has been working closely with him for about a decade... Tesla sells huge batteries commercially and expect that part of its business to grow a lot: https://en.wikipedia.org/wiki/Tesla_Megapack Tesla had a "Battery Day" event last year to tell investors about their battery technology, both current and upcoming. The press mostly ignored it but they showed several things that will change the world because they promise to make lithium-ion batteries much cheaper to produce (and also better than they are now). The official stream (140 minutes): https://www.youtube.com/watch?v=l6T9xIeZTds Much shorter version (16 minutes): https://www.youtube.com/watch?v=HK79ioBW8Mg The main competition for lithium-ion batteries for the grid is flow batteries. Unfortunately, they tend to use big tanks of really nasty chemicals (which may have to be heated to work properly). They are also unlikely to be able to compete with future lithium-ion batteries. https://en.wikipedia.org/wiki/Flow_battery
    4
  27. No, they demanded quite expensive licenses until the patents ran out.
    4
  28. A free South Vietnam could have prospered. Wouldn’t that have been better?
    4
  29. There are huge costs associated with the backup electricity generation when the sun doesn’t shine, both in terms of money and in terms of political instability. It turns out that the best backup in practice is pipelined gas, high is exactly why Russia thought it was safe to 1) start the Ukraine war in 2014 and 2) escalate drastically in February this year. Eventually, batteries should be good enough, plentiful enough, and cheap enough so we can avoid using gas for backup. Here and now, it is much better to just use nuclear power plant.
    4
  30. Unix didn’t originally have virtual memory. When it got it, it was originally a rather dumb form that didn’t use pages. It took a while before the machines people wanted to run Unix on had paging MMUs.
    4
  31.  @kd6788  good.
    3
  32.  @chriswestwood3289  "come back talk"? They have been done. Try harder, 五毛.
    3
  33. Licensed, not copied.
    3
  34. They should have seen the writing on the wall when their low cost system using off the shelf bit slicing ICs was faster than their top models. It was obvious that cheap single-chip (or “few-chip”) CPUs would beat them soon after.
    3
  35. It’s news to me that there is a “massive US electric bus market”!
    3
  36. Ex-man, actually.
    3
  37. It has been “not a word” for many centuries, with dictionary entries and lots of good writers using it and everything.
    3
  38.  @NUCLEARARMAMENT  I don’t think you know much about compilers or CPU architecture. There is no need for special features for “object-oriented” programming (which is not a single thing but many, many different things). All you need is a good branch predictor and a good return address predictor (they have been standard for more than two decades on fast CPUs). No, the problem with the Elbrus is with the architecture itself, not with the compilers. Some parallel architectures are simply harder to wring performance out of than others. A more recent good example is the IBM Cell. “Sufficiently smart compilers” don’t exist.
    3
  39.  @rainbowhyena1354  we can get several thousand cycles with lithium-ion batteries now. No, the real problem is cost and availability. They cost too much and we don’t make enough.
    3
  40.  @AC-jk8wq his way is wrong, though.
    3
  41. CPLDs are useful because they are ready immediately. FPGAs need to clock bits out of a ROM or Flash EEPROM one by one to program them on boot. So, FPGAs are not ever going to take over everything. On the other hand, PALs/PLAs are not as flexible as CPLDs.
    3
  42.  @Diggnuts  the laser light source in those machines is American. So, yes, it is up to the US.
    2
  43. A lot of the IP in those chips is licensed from ARM.
    2
  44. Your memory is not very good.
    2
  45.  @G.S.30 ”it was not Communism. True Communism hasn’t been tried yet.” Moron.
    2
  46. But you also have Bremsstrahlung and Hohlraum radiation :)
    2
  47. @JasonAtlasThey didn’t need any help. They have a special talent.
    2
  48. Compare average national IQs of Taiwan and Brazil. Brazil never had a chance.
    2
  49. Proteases and Maltesers do NOT rhyme.
    2
  50.  @waqaszubair8906  Huawei owns a company that designs SOC chips (“system on a chip”) but the work is mostly about taking designs bought or licensed from others and combining them into a single design. The manufacturing was actually done by others (read: TSMC) on a process that wasn’t cutting edge. I don’t know where they get them manufactured now.
    2