I think it must be something missing in the article / summary - I don't understand how running a chip slower than design is going to protect against SEUs.SEUs don't have anything to do with the clock rate, but only about the energy levels required to flip a bit.
TFA does mention something about a "software approach" but I'm not certain what that would be, unless it is something like running every computation twice to see if you get the same result both times - which is either going to mean you need twice the equipment or run at half speed.
But... now that I think about it, that might still be a win, because hardened chips run a fair bit slower than half speed at a cost a fair bit more than twice as much...
This depends highly on the complexity and type of part. A vehicle body panel - additive manufacturing will never beat the cycle time of a press. A complex turbine blade - yeah, 3D printing is going to be competitive.
I think some of the skepticism of "3D printing" is that much of the stuff we see in the press is that it's going to replace all manufacturing, which just isn't the case.
Don't forget the other tradeoffs associated with "easy to make custom parts" - this means the engineering rigor associated with each custom part is going to be more expensive (per part) than for mass-produced items. Either that or we'll just get a lot of custom or low-volume parts with really low quality: quality isn't just from the manufacturing technique -
if you print a poorly designed part perfectly, you still have a low-quality part!
No one in OSS has taken the initiative to do anything with it.
The MathWorks is notorious for litigation. There's a reason why they are expensive and there are no (or is it merely 'few'?) alternatives.
I actually like it more for collaborative development because it means you have to break apart your system into libraries.
You've never had two people try to change the same library at the same time? That's what I meant by merging. I guess you could argue then that the library is too big - but if you get any smaller, they aren't libraries but single files for a single block. (Which actually might not be a bad thing, mind you!)
I'm the opposite. I can knock out a complex controller much, much faster in Simulink....
If you said Stateflow I might agree with you - that is a great tool, and I can indeed write state machines much faster using it than by hand.
Changes are even faster. Need to test it on an 8-bit controller, switch everything to fixed point. Need to validate fixed point vs floating point? A few clicks.
That is indeed a technical feat that is nice. But it is kind of terrifying in other ways. Especially since you mentioned DO-178C earlier...I can't imagine the impact analyses required to make such a fundamental architectural change on a DO-178 project (unless you're talking about some kind of library code re-use?).
Rule of Generation
Developers should avoid writing code by hand and instead write abstract high-level programs that generate code. This rule aims to reduce human errors and save time.
In my experience Simulink and its toolchain is painful. It also doesn't help that it's proprietary, expensive, and doesn't lend itself nicely to collaborative development (ever try to to merge two branches with changes to the same.mdl file? Oh, and if you use the new.snx files, you can't even merge them, because they are binary...). Developing complex code in Simulink is painful - you think spaces vs tabs is a pointless debate? You should hear discussions about how to route your signals in Simulink or how you should group your subsystems.
I think autocoding is indeed the future, but I long for something better than Simulink.
While Simulink may reduce some kinds of errors, I have never once had it save time. At least in the code I've written, I can *always* write it faster in C than in Simulink. Now, reviewing Simulink I will say can indeed be faster than reviewing C code, but writing it? No way.
I know this is a really late reply... but there is a false assumption in your statement that inability to produce enough to pay for resources is due to an individual's activities and not systematic.
While I agree that a simple random assignment of property rights is not efficient in some senses of efficiency, it does eliminate systemic wealth concentration effects (basically it trades off simplistic profit efficiency for equal opportunity). Consider if you have a fertile piece of land (say, 100 acres: small enough to have the same climate, soil type, exposure to hazards, etc.) and you simply divide it up randomly among a population. That is actually a good example of "equal opportunity" - each person could, with the same effort, get the same produce out of the land.
But if you allow a person who works hard to buy all the land from one who doesn't work hard - then the one who doesn't work will then have less opportunity - they are subsequently at a disadvantage to "reacquire" the productive resource: they would have to produce so much that they make it worth the (productive) landowner's while to sell the property back. So it's not even "equal ownership for equal work" at that point - the disadvantaged actually has to work harder (or get luckier) to get back to an equal ownership situation. It's not even necessarily out of malice on the property owner's part - it's just rational choice: if I bought a thing for $X, I would want more than $X to sell it.
That's the essence of "systemic" bias against the poor - and sometimes it's not even due to the fact that a person sold their productive resources, but was simply born into an area where all the productive resources were already owned.
This. I recently switched jobs and went from a Mac to Windows10. Why is that if I fat-finger my password on Win10 it takes 15 seconds for it to let me re-type it? (I know the reasons, but they are...questionable.)
In general, what irks me most typically aren't glitches, but annoying design decisions. These are the main ones for me:
Hardware: coil whine.
Office 2016: all that slow typing animation. And the windows even drag slowly. There is a registry hack (seriously?) to fix the typing one, but no solution for the window drag one.
Windows10: still can't handle connecting and disconnecting additional monitors. Windows stuck off-screen or sized incorrectly, etc. And this is on a laptop that is used in two locations, so is always swapping between the *same* two external monitors.
Windows10: still treats the same piece of hardware (e.g., mouse) as a different device depending on the particular USB port to which it's connected. Same mouse, different port, must install new drivers, resets mouse speed settings, etc.
My BluRay player: When it connects to the net, it displays a nuisance modal dialog box saying "network connected" (only show me one if there is an error).
More accurately: people want a one-time payment to purchase a quality product. The reason they demand a supply of security and bug fixes is because the initial purchased product is defective - it has bugs and security flaws (often latent ones which are not obvious at time of purchase).
That's definitely a solution, but I would replace the highest-bidder-wins auction with a certifiably random selection. That way you don't have a systemic bias against the poor.
If your vehicle's firmware needs updates - let alone frequent ones - you've already lost. If your car allows remote access to anything - you're already teetering on the edge (I can't stand remote start for instance, ad key fobs are nice but they are a risk.)
This culture of "eh, just release it - we can issue a patch later" is toxic.
You want to know the historic term for having to repair something on a car after it was sold: recall. Just replace the word "software patch" with "recall" in your mind from now on...
And I've been writing embedded software for almost 20 years. This stuff makes me cry.
Gasoline is arguably more important and more of a daily necessity than health care. We need gasoline reform! None of this daily price change, guessing if you have enough change in the seat cushions of your car to fill your tank or not.
Down with the days of filling your tank, only to have the price drop $0.10 a gallon the next day, robbing you of the $1.50 you would have saved had you known the price would drop and you filled up in the morning after instead of the evening before!
We need price stability for gasoline - something like "no more than a 1% change per week, no more than a 1% change per 25 miles. That is - limits on how fast the price can change with respect to time and space.
Related: Since the price of a barrel of oil is only about $45-$50 right now - why does a quart of (non-synthetic) motor oil cost $7!? That's an equivalent price of close to $1,200 a barrel!
I've always thought that tech FRAND agreements are anything but. This is mostly because they are worded to be some kind of percentage of product price or something otherwise odd like that.
FRAND should be this and only this: Licenses are $X per unit on volumes up to A, $X-b up to B, etc. That is - a test for FRAND should be if the price is a fixed nominal value (perhaps with volume scaling), not a fixed percentage.
You don't have to burn fuel to have a jet - you just need something to spin a compressor. Gas turbines just happen to be an easy way to do this with impressive power to weight and power to volume density. I know there is lots of research into electric compressors, but I didn't realize they were getting that competitive; I've been out of that industry for a while.
I wonder what the amortized emissions of the concrete used in the production of those hydro plants is - concrete production is a significant CO2 source.
I say we forget trying to sequester carbon. Instead, we should create artificial hurricanes. This will have two effects: more transfer of heat from the surface to the upper atmosphere where it can be radiated to space, and since hurricanes are massive heat engines, we can just use them as power plants.
'Safe' vs 'unsafe' I think is the wrong dichotomy.
Consider that the world actually needs 'unsafe' tools - for instance: knives, saws, fire. But what you want in tools like this is they always act in a nice deterministic way so, if you treat them with respect, you can obtain useful work from them.
Code is essentially the same thing. What is necessary is the proper amount of respect for the damage your tools can inflict, not suggesting everyone to use safety scissors (except where it's appropriate of course).
I think the bigger issue is that some people can sacrifice basically their entire standard of living and it still won't guarantee a particular outcome. And this isn't even on the backs of the "wealthy fossil fuel barons" - this is just the average everyday person.
So it boils down to "absolutely everyone must do something" and the backlash of "who are you to tell me what I should do."
So then you get into "since everyone isn't doing it, I'm not going to sacrifice..." and nothing happens. Basically you have to accept forced behaviors or penalties - which a substantial portion of the population (not just in the US, by the way) takes issue with.
There are essentially only two types of economic activity: creation of wealth and trade. Creation of wealth is actual manufacturing or agriculture that generates consumables or what I would call "productive infrastructure" like machines or homes or storage facilities. Trade - while it may indeed generate value for both parties involved in the trade, cannot generate wealth.
In the general economy is a combination of activities where some people / organizations do actually invest in new production to create wealth so in aggregate it is true that the economy is not "zero sum". But at some point things bifurcate: a portion of the population is no longer able (for some reason or another) to generate new wealth, but instead can only trade.
For the portion of the population that can only trade, it is indeed a zero-sum game, except for the instances where the actual wealth producers inject new wealth into the trade markets.
So yeah, a Gates or Buffet or whatever doesn't necessarily prevent new wealth from being created - but as wealth concentrates, there isn't necessarily any impetus for the wealthy to generate more wealth (in fact, diminishing marginal returns say at some point it won't). We are just assuming that the wealthy will continue to want more - but at some point, they are effectively self-contained and would no longer have any need for trade. At that point - what happens? This is the source of fear (rational or not) from globalization and automation - when the owners of productive capital no longer have a need or desire to trade with non-owners.
To summarize - and I do agree it is theoretical/hypothetical: super-wealthy entities harm non-wealthy when, instead of using that wealth to increase production (through productivity increases or just new capacity at the same productivity levels), that wealth is only used to acquire ownership of a larger portion of the total available wealth (e.g., rent-seeking).
A VAT is actually counterproductive - you don't want to tax adding value!
What we really just need is an ownership tax - perhaps something like a property tax. The simplest form is this: your income tax rate is proportional to your ownership percentile.
This means if you own a lot, but have zero income, you get low tax - so you can keep your wealth. If you own nothing, and suddenly get income - you get to keep most of your income.
If you have massive wealth and massive income, you get taxed massively.
This solves most of the adverse incentive problems in any other form of taxation, but it has to be accepted that its unabashed purpose is to dampen wealth concentration effects.
If someone (or small enough group) has so much of the wealth that it's no longer possible for anyone else to obtain means of production, that's the problem.
Put another way - poverty is a result of being unable to produce something you need yourself or being unable to trade what you can produce for something you need. The reasons for that can be many, but the result is the same: if you can't produce what you need or get it by trade (or by gift), you're poor.
If all the wealth - that is, means of production and tradable goods - is owned by a small group, and you can't even provide a service for which those owners are interested in trading, and the government doesn't "tax" that wealth to redistribute it to you, you're SOL.
Funny thing with this rationale is that it fails to quantify the benefit of not having a monthly payment obligation.
Also - if you have $800 in your pocket now and you want to take out a 0%/24-month loan and "invest" the cash instead of buying outright, that means you have to have an investment where you can withdraw $33 a month to make the payment on the 0% loan. You can't just pay that $800, say, into your mortgage because that money is not really liquid and you can't use it to make the payments on that 0% loan.
I'm pessimistic and think, sadly, that you have too much faith in the patent office.
The correct response to this doesn't even have anything to do with "an abstract idea" - it should be "You have a device that is designed to take pictures and can identify objects in those photos, and that device has location information and storage. Using that device to keep track of the locations of objects is obvious - not patentable.
The only thing patentable would be if they had some novel way of storing location information perhaps, or some novel method of object recognition. The "idea" "keep track of objects' locations with a hololens" should not be patentable at all.
I suddenly just realized what you're saying I think - you're saying that a portion of people who currently could pay $300 a month in rent but don't because it's not stable and they might lose their homes, would be willing to pay that with a UBI because it is stable. And, knowing that there are now people with a stable income, builders would build properties at that lower price point (where they currently don't). Because this is new construction in a new segment of the market it would not therefore necessarily have an impact on price levels in other markets.
And I would agree with that assessment - when you have new production in a market that doesn't compete with other markets, then you simply have economic growth with minimal impact on price levels.
Slashdot Mirror
Twice lets you know if something isn't working, which sometimes is enough. You only need triple redundancy when you have to know the correct answer.
I think it must be something missing in the article / summary - I don't understand how running a chip slower than design is going to protect against SEUs.SEUs don't have anything to do with the clock rate, but only about the energy levels required to flip a bit.
TFA does mention something about a "software approach" but I'm not certain what that would be, unless it is something like running every computation twice to see if you get the same result both times - which is either going to mean you need twice the equipment or run at half speed.
But... now that I think about it, that might still be a win, because hardened chips run a fair bit slower than half speed at a cost a fair bit more than twice as much...
This depends highly on the complexity and type of part. A vehicle body panel - additive manufacturing will never beat the cycle time of a press. A complex turbine blade - yeah, 3D printing is going to be competitive.
I think some of the skepticism of "3D printing" is that much of the stuff we see in the press is that it's going to replace all manufacturing, which just isn't the case.
Don't forget the other tradeoffs associated with "easy to make custom parts" - this means the engineering rigor associated with each custom part is going to be more expensive (per part) than for mass-produced items. Either that or we'll just get a lot of custom or low-volume parts with really low quality: quality isn't just from the manufacturing technique - if you print a poorly designed part perfectly, you still have a low-quality part!
The MathWorks is notorious for litigation. There's a reason why they are expensive and there are no (or is it merely 'few'?) alternatives.
You've never had two people try to change the same library at the same time? That's what I meant by merging. I guess you could argue then that the library is too big - but if you get any smaller, they aren't libraries but single files for a single block. (Which actually might not be a bad thing, mind you!)
If you said Stateflow I might agree with you - that is a great tool, and I can indeed write state machines much faster using it than by hand.
That is indeed a technical feat that is nice. But it is kind of terrifying in other ways. Especially since you mentioned DO-178C earlier...I can't imagine the impact analyses required to make such a fundamental architectural change on a DO-178 project (unless you're talking about some kind of library code re-use?).
In my experience Simulink and its toolchain is painful. It also doesn't help that it's proprietary, expensive, and doesn't lend itself nicely to collaborative development (ever try to to merge two branches with changes to the same .mdl file? Oh, and if you use the new .snx files, you can't even merge them, because they are binary...). Developing complex code in Simulink is painful - you think spaces vs tabs is a pointless debate? You should hear discussions about how to route your signals in Simulink or how you should group your subsystems.
I think autocoding is indeed the future, but I long for something better than Simulink.
While Simulink may reduce some kinds of errors, I have never once had it save time. At least in the code I've written, I can *always* write it faster in C than in Simulink. Now, reviewing Simulink I will say can indeed be faster than reviewing C code, but writing it? No way.
I know this is a really late reply... but there is a false assumption in your statement that inability to produce enough to pay for resources is due to an individual's activities and not systematic.
While I agree that a simple random assignment of property rights is not efficient in some senses of efficiency, it does eliminate systemic wealth concentration effects (basically it trades off simplistic profit efficiency for equal opportunity). Consider if you have a fertile piece of land (say, 100 acres: small enough to have the same climate, soil type, exposure to hazards, etc.) and you simply divide it up randomly among a population. That is actually a good example of "equal opportunity" - each person could, with the same effort, get the same produce out of the land.
But if you allow a person who works hard to buy all the land from one who doesn't work hard - then the one who doesn't work will then have less opportunity - they are subsequently at a disadvantage to "reacquire" the productive resource: they would have to produce so much that they make it worth the (productive) landowner's while to sell the property back. So it's not even "equal ownership for equal work" at that point - the disadvantaged actually has to work harder (or get luckier) to get back to an equal ownership situation. It's not even necessarily out of malice on the property owner's part - it's just rational choice: if I bought a thing for $X, I would want more than $X to sell it.
That's the essence of "systemic" bias against the poor - and sometimes it's not even due to the fact that a person sold their productive resources, but was simply born into an area where all the productive resources were already owned.
This. I recently switched jobs and went from a Mac to Windows10. Why is that if I fat-finger my password on Win10 it takes 15 seconds for it to let me re-type it? (I know the reasons, but they are...questionable.)
In general, what irks me most typically aren't glitches, but annoying design decisions. These are the main ones for me:
More accurately: people want a one-time payment to purchase a quality product. The reason they demand a supply of security and bug fixes is because the initial purchased product is defective - it has bugs and security flaws (often latent ones which are not obvious at time of purchase).
That's definitely a solution, but I would replace the highest-bidder-wins auction with a certifiably random selection. That way you don't have a systemic bias against the poor.
If your vehicle's firmware needs updates - let alone frequent ones - you've already lost. If your car allows remote access to anything - you're already teetering on the edge (I can't stand remote start for instance, ad key fobs are nice but they are a risk.)
This culture of "eh, just release it - we can issue a patch later" is toxic.
You want to know the historic term for having to repair something on a car after it was sold: recall. Just replace the word "software patch" with "recall" in your mind from now on...
And I've been writing embedded software for almost 20 years. This stuff makes me cry.
This post is only half tongue-in-cheek.
Gasoline is arguably more important and more of a daily necessity than health care. We need gasoline reform! None of this daily price change, guessing if you have enough change in the seat cushions of your car to fill your tank or not.
Down with the days of filling your tank, only to have the price drop $0.10 a gallon the next day, robbing you of the $1.50 you would have saved had you known the price would drop and you filled up in the morning after instead of the evening before!
We need price stability for gasoline - something like "no more than a 1% change per week, no more than a 1% change per 25 miles. That is - limits on how fast the price can change with respect to time and space.
Related: Since the price of a barrel of oil is only about $45-$50 right now - why does a quart of (non-synthetic) motor oil cost $7!? That's an equivalent price of close to $1,200 a barrel!
I've always thought that tech FRAND agreements are anything but. This is mostly because they are worded to be some kind of percentage of product price or something otherwise odd like that.
FRAND should be this and only this: Licenses are $X per unit on volumes up to A, $X-b up to B, etc. That is - a test for FRAND should be if the price is a fixed nominal value (perhaps with volume scaling), not a fixed percentage.
You don't have to burn fuel to have a jet - you just need something to spin a compressor. Gas turbines just happen to be an easy way to do this with impressive power to weight and power to volume density. I know there is lots of research into electric compressors, but I didn't realize they were getting that competitive; I've been out of that industry for a while.
I wonder what the amortized emissions of the concrete used in the production of those hydro plants is - concrete production is a significant CO2 source.
I say we forget trying to sequester carbon. Instead, we should create artificial hurricanes. This will have two effects: more transfer of heat from the surface to the upper atmosphere where it can be radiated to space, and since hurricanes are massive heat engines, we can just use them as power plants.
Win-win!
I'm only half-joking.
'Safe' vs 'unsafe' I think is the wrong dichotomy.
Consider that the world actually needs 'unsafe' tools - for instance: knives, saws, fire. But what you want in tools like this is they always act in a nice deterministic way so, if you treat them with respect, you can obtain useful work from them.
Code is essentially the same thing. What is necessary is the proper amount of respect for the damage your tools can inflict, not suggesting everyone to use safety scissors (except where it's appropriate of course).
You shouldn't need electricity or some kind of communications infrastructure in order to carry out trade.
I think the bigger issue is that some people can sacrifice basically their entire standard of living and it still won't guarantee a particular outcome. And this isn't even on the backs of the "wealthy fossil fuel barons" - this is just the average everyday person.
So it boils down to "absolutely everyone must do something" and the backlash of "who are you to tell me what I should do."
So then you get into "since everyone isn't doing it, I'm not going to sacrifice..." and nothing happens. Basically you have to accept forced behaviors or penalties - which a substantial portion of the population (not just in the US, by the way) takes issue with.
The short response is, of course, "it depends."
There are essentially only two types of economic activity: creation of wealth and trade. Creation of wealth is actual manufacturing or agriculture that generates consumables or what I would call "productive infrastructure" like machines or homes or storage facilities. Trade - while it may indeed generate value for both parties involved in the trade, cannot generate wealth.
In the general economy is a combination of activities where some people / organizations do actually invest in new production to create wealth so in aggregate it is true that the economy is not "zero sum". But at some point things bifurcate: a portion of the population is no longer able (for some reason or another) to generate new wealth, but instead can only trade.
For the portion of the population that can only trade, it is indeed a zero-sum game, except for the instances where the actual wealth producers inject new wealth into the trade markets.
So yeah, a Gates or Buffet or whatever doesn't necessarily prevent new wealth from being created - but as wealth concentrates, there isn't necessarily any impetus for the wealthy to generate more wealth (in fact, diminishing marginal returns say at some point it won't). We are just assuming that the wealthy will continue to want more - but at some point, they are effectively self-contained and would no longer have any need for trade. At that point - what happens? This is the source of fear (rational or not) from globalization and automation - when the owners of productive capital no longer have a need or desire to trade with non-owners.
To summarize - and I do agree it is theoretical/hypothetical: super-wealthy entities harm non-wealthy when, instead of using that wealth to increase production (through productivity increases or just new capacity at the same productivity levels), that wealth is only used to acquire ownership of a larger portion of the total available wealth (e.g., rent-seeking).
Mea culpa - I was assuming you were talking hypothetical, not actually about Mr. Gates as a specific individual...
A VAT is actually counterproductive - you don't want to tax adding value!
What we really just need is an ownership tax - perhaps something like a property tax. The simplest form is this: your income tax rate is proportional to your ownership percentile.
This means if you own a lot, but have zero income, you get low tax - so you can keep your wealth. If you own nothing, and suddenly get income - you get to keep most of your income.
If you have massive wealth and massive income, you get taxed massively.
This solves most of the adverse incentive problems in any other form of taxation, but it has to be accepted that its unabashed purpose is to dampen wealth concentration effects.
If someone (or small enough group) has so much of the wealth that it's no longer possible for anyone else to obtain means of production, that's the problem.
Put another way - poverty is a result of being unable to produce something you need yourself or being unable to trade what you can produce for something you need. The reasons for that can be many, but the result is the same: if you can't produce what you need or get it by trade (or by gift), you're poor.
If all the wealth - that is, means of production and tradable goods - is owned by a small group, and you can't even provide a service for which those owners are interested in trading, and the government doesn't "tax" that wealth to redistribute it to you, you're SOL.
Funny thing with this rationale is that it fails to quantify the benefit of not having a monthly payment obligation.
Also - if you have $800 in your pocket now and you want to take out a 0%/24-month loan and "invest" the cash instead of buying outright, that means you have to have an investment where you can withdraw $33 a month to make the payment on the 0% loan. You can't just pay that $800, say, into your mortgage because that money is not really liquid and you can't use it to make the payments on that 0% loan.
I'm pessimistic and think, sadly, that you have too much faith in the patent office.
The correct response to this doesn't even have anything to do with "an abstract idea" - it should be "You have a device that is designed to take pictures and can identify objects in those photos, and that device has location information and storage. Using that device to keep track of the locations of objects is obvious - not patentable.
The only thing patentable would be if they had some novel way of storing location information perhaps, or some novel method of object recognition. The "idea" "keep track of objects' locations with a hololens" should not be patentable at all.
I suddenly just realized what you're saying I think - you're saying that a portion of people who currently could pay $300 a month in rent but don't because it's not stable and they might lose their homes, would be willing to pay that with a UBI because it is stable. And, knowing that there are now people with a stable income, builders would build properties at that lower price point (where they currently don't). Because this is new construction in a new segment of the market it would not therefore necessarily have an impact on price levels in other markets.
And I would agree with that assessment - when you have new production in a market that doesn't compete with other markets, then you simply have economic growth with minimal impact on price levels.