OK kids - here's the rundown on power generation given to seventh graders visiting power stations on field trips. Base load is the minimum demand and it's handy having stuff running 24/7/365 to produce it. Thermal power (coal/nukes/gas when gas is cheap) and hydro are good for that. Peak loads are when demand is higher and you switch in other generators as needed. If you only need something for a few hours a day it has a low capacity factor no matter whether it's capable of running 24/7/365 or not. It's that simple. Think of that when somebody uses "capacity factor" to push an agenda and pretends it's an indication of downtime due to mechanical failure or a lack of wind/sun/gas/water. It appears to be the term of choice for political opposition to various sources of electricity generation and the misuse probably came from some pimply Washington intern who thought he was being deviously clever instead of a manipulative prick.
Sorry about that. I misunderstood earlier and assumed the discussion was about reducing the process size instead of using a smaller area of wafer at the same size.
You are still trying to defend your snarky little attack on progress? The bit were you pretended to be stupid by bitching about costings for a process line for something that has only just been invented was the especially annoying insult to the intelligence of everyone here - most especially your own.
The largest irony is this thing is not breakthrough but incremental improvement, plus you are almost certain to have something powered by a battery based on the work of the inventor mentioned in the article already. This is no zero point energy scam or similar that turns up on this site from time to time.
Is that a way of saying with at the same process size? I have been writing about how smaller flaws matter when going from say 32 nm to 14 nm. If so sorry about the confusion.
If the traces or other items masked off are smaller then some small flaws that would not impact on larger things are enough to disable those small items.
And FYI, I work in the semiconductor industry
Then there must be some serious miscommunication here since it was looking like you'd missed something that we used to tell all of the first year engineering students in introductory materials science.
what you mean by "critical flaw size"
I meant the size at which the flaw matters (becomes critical) and not an actual technical term like "critical crack size" which is not the same and is about fracture. What the paper you linked does not go into is that there will be defects in the zone refined ingot of a wide range of sizes but below a certain size they just do not matter. That certain size is going to change if all of your traces are much smaller, which is beyond the scope of that paper because it's not about changing processes.
Good idea for some applications, once it's ready for prime time, but all the "X sux" shit from the fanboys and Daniel Stone detracts from it. Those losers (not the main dev community but a loud bunch of fanboys) are actually using the new, slow, buggy piece of shit gtk3 gedit as their "proof" that X is slow, while the gtk2 version of gedit has no trouble at all (and X is not the problem). Oh wait - you are just trolling aren't you? Please don't because my point about lazy programmers of toolkits relying on beefy graphics cards to get even average performance is an annoying issue and quite separate to both X and Wayland (and inflicts itself on both).
The very obvious one of pretending that a second very different question (with an obvious answer) was identical to the first one because you appeared to be unable to be critical of me without doing so. You pretended that I was denying the second undeniable thing you shifted the goalposts to. Shame on you! All for the sake of some mindless Eloi versus Moorlock shitfight.
Now shall I give you utter stupid advice on how to deal with databases in return or are you starting to get the idea of how ridiculous your initial post was?
shows that the militia is only one example of a use of the Right to Keep and Bear Arms
Exactly. It's a right that has nothing to do with the 2nd amendment no matter what the NRA are spouting by pretending to be a "militia" themselves. You don't have to hand your guns back in at 45 because it's not the second amendment that says you can have them in the first place.
Waaaaaayyyyy too rarely do they actually make it out of the lab.
Proof positive - cargo cult luddite that appears to assume that every successful product is born perfect instead of being the result of a process where less perfect stepping stones don't make it out of the lab.
The sort of mangled Chinese whispers from an MBA bullshit you were spouting initially isn't even taught to MBAs any more.
Which of course is something that is totally unknown at this point so is a ridiculous rhetorical question designed deliberately to push an agenda - as well as being wrong. If nobody is buying the old product the marginal gain is irrelevant. You adapt or die.
Why have you gone so far out of your depth on this one? Changing processes is not necessarily incredibly expensive. Shouldn't you be commenting about something you actually know something about instead? What's with the insults and "duh" when it's all incredibly speculative?
Your economic argument relies on the product being worth manufacturing somewhere, if not on current process lines, so your backtrack of pretending you never thought the product was worth considering is another blatant lie. Why the fuck are you doing this? What is it to you? Why do you want the kiddies to doubt based on numbers before the numbers are even available?
This sort of Porche will be safe in a crash and even better it's a 5 seater, plus sort of hybrid though the fuel economy sucks. https://en.wikipedia.org/wiki/VK_4501_(P) Might need a fire extinguisher of two.
Only if you are male, under 45, and have the courage to join the National Guard. Try actually reading the thing some time instead of the Ollie North Hezbolla version.
How about we get back to the NRA instead of spouting their propaganda designed to make a poorly regulated rifle club look like something George Washington said we should all join.
Your luddite problem? It's called a cargo cult - a love of technology but not giving a shit about how it gets there, as shown by your rhetorical deliberately unanswerable question you asked to sow doubt.
It's not a "hard question" as you well know. It's an utterly pointless question (for reasons other than sowing doubt to push an agenda) since it's not going to have an answer until a process is developed and costed. The last time I met someone acting like you they also went on at great length about healing crystals and pyramid power.
**.- Before any of you that do some niche job like wave simulation or 3D rendering scream "But I needs moar power!"? You and your ilk are less than 2% max of the market, just too niche to be a big enough market to support the insane amount of R&D required to make that next leap.
Yes, but thankfully people getting 1337 gaming boxes means our stuff is really not all that different at the motherboard/CPU level and is a lot cheaper than having to get stuff from IBM or Oracle/Sun.
The critical flaw size increases as you shrink the die. Thus with all else being equal more flaws. Those flaws that previously were too small to matter now do. I hope I made that clear enough.
While that is true there are a lot of things where working in parallel is trivial. Image processing is one that most readers will be familiar with. How hard is "apply this filter to every pixel"? You don't care what order it gets done in so long as you get the entire result dumped somewhere. There are a lot of things in science and engineering with that same sort of approach of applying the same transformation to every item in a dataset. While there is plenty of stuff that can't be done in parallel there is a lot of untouched "low hanging fruit" that is. Currently the situation seems to be you pick Xeons for single threaded jobs and AMD if you need a huge number of cores (64 cores and 1TB of memory per node is cheaper than you would think). For those lucky enough to have tasks that don't need a lot of memory the GPU stuff gives you a vast number of cores.
This hints that you don't know what the economic term "marginal" means
You are seriously playing that "not educated enough" card as an adult? Do you really have that much contempt for the people around you?
Besides, "marginal" was not even mentioned until you engineered some petty "win" by moving goalposts. I did not question your use of it, I merely pointed out that you were changing the topic while claiming it was still the original topic.
Slashdot Mirror
OK kids - here's the rundown on power generation given to seventh graders visiting power stations on field trips.
Base load is the minimum demand and it's handy having stuff running 24/7/365 to produce it. Thermal power (coal/nukes/gas when gas is cheap) and hydro are good for that.
Peak loads are when demand is higher and you switch in other generators as needed.
If you only need something for a few hours a day it has a low capacity factor no matter whether it's capable of running 24/7/365 or not.
It's that simple.
Think of that when somebody uses "capacity factor" to push an agenda and pretends it's an indication of downtime due to mechanical failure or a lack of wind/sun/gas/water. It appears to be the term of choice for political opposition to various sources of electricity generation and the misuse probably came from some pimply Washington intern who thought he was being deviously clever instead of a manipulative prick.
Sorry about that. I misunderstood earlier and assumed the discussion was about reducing the process size instead of using a smaller area of wafer at the same size.
You are still trying to defend your snarky little attack on progress?
The bit were you pretended to be stupid by bitching about costings for a process line for something that has only just been invented was the especially annoying insult to the intelligence of everyone here - most especially your own.
The largest irony is this thing is not breakthrough but incremental improvement, plus you are almost certain to have something powered by a battery based on the work of the inventor mentioned in the article already. This is no zero point energy scam or similar that turns up on this site from time to time.
But you don't know a lot more and that is the problem. You are out of your depth and trying to drag others down.
Is that a way of saying with at the same process size? I have been writing about how smaller flaws matter when going from say 32 nm to 14 nm.
If so sorry about the confusion.
Then there must be some serious miscommunication here since it was looking like you'd missed something that we used to tell all of the first year engineering students in introductory materials science.
I meant the size at which the flaw matters (becomes critical) and not an actual technical term like "critical crack size" which is not the same and is about fracture. What the paper you linked does not go into is that there will be defects in the zone refined ingot of a wide range of sizes but below a certain size they just do not matter. That certain size is going to change if all of your traces are much smaller, which is beyond the scope of that paper because it's not about changing processes.
No assumptions just what you have written, a cry of doubt railing against the dread spirit of innovation.
You didn't wait - you threw shit on it years before a process line is even designed.
Good idea for some applications, once it's ready for prime time, but all the "X sux" shit from the fanboys and Daniel Stone detracts from it. Those losers (not the main dev community but a loud bunch of fanboys) are actually using the new, slow, buggy piece of shit gtk3 gedit as their "proof" that X is slow, while the gtk2 version of gedit has no trouble at all (and X is not the problem).
Oh wait - you are just trolling aren't you? Please don't because my point about lazy programmers of toolkits relying on beefy graphics cards to get even average performance is an annoying issue and quite separate to both X and Wayland (and inflicts itself on both).
Just use something without gtk3 and you'll have something optimised for X.
Stop using the current gnome stuff and you'll see the speed difference.
The very obvious one of pretending that a second very different question (with an obvious answer) was identical to the first one because you appeared to be unable to be critical of me without doing so. You pretended that I was denying the second undeniable thing you shifted the goalposts to.
Shame on you!
All for the sake of some mindless Eloi versus Moorlock shitfight.
Now shall I give you utter stupid advice on how to deal with databases in return or are you starting to get the idea of how ridiculous your initial post was?
Exactly. It's a right that has nothing to do with the 2nd amendment no matter what the NRA are spouting by pretending to be a "militia" themselves. You don't have to hand your guns back in at 45 because it's not the second amendment that says you can have them in the first place.
Proof positive - cargo cult luddite that appears to assume that every successful product is born perfect instead of being the result of a process where less perfect stepping stones don't make it out of the lab.
The sort of mangled Chinese whispers from an MBA bullshit you were spouting initially isn't even taught to MBAs any more.
Yet your "question" relied on the device being worth mass producing in the first place.
You can't have it both ways without being a liar.
Which of course is something that is totally unknown at this point so is a ridiculous rhetorical question designed deliberately to push an agenda - as well as being wrong. If nobody is buying the old product the marginal gain is irrelevant. You adapt or die.
Why have you gone so far out of your depth on this one? Changing processes is not necessarily incredibly expensive. Shouldn't you be commenting about something you actually know something about instead? What's with the insults and "duh" when it's all incredibly speculative?
Your economic argument relies on the product being worth manufacturing somewhere, if not on current process lines, so your backtrack of pretending you never thought the product was worth considering is another blatant lie.
Why the fuck are you doing this? What is it to you?
Why do you want the kiddies to doubt based on numbers before the numbers are even available?
This sort of Porche will be safe in a crash and even better it's a 5 seater, plus sort of hybrid though the fuel economy sucks.
https://en.wikipedia.org/wiki/VK_4501_(P)
Might need a fire extinguisher of two.
Only if you are male, under 45, and have the courage to join the National Guard. Try actually reading the thing some time instead of the Ollie North Hezbolla version.
How about we get back to the NRA instead of spouting their propaganda designed to make a poorly regulated rifle club look like something George Washington said we should all join.
Your luddite problem? It's called a cargo cult - a love of technology but not giving a shit about how it gets there, as shown by your rhetorical deliberately unanswerable question you asked to sow doubt.
It's not a "hard question" as you well know. It's an utterly pointless question (for reasons other than sowing doubt to push an agenda) since it's not going to have an answer until a process is developed and costed. The last time I met someone acting like you they also went on at great length about healing crystals and pyramid power.
"marginal" was not even mentioned until...
It was embedded in my original question.
Really? Let's see that question again?
No - not there in any way at all.
What's with the transparent lies?
Yes, but thankfully people getting 1337 gaming boxes means our stuff is really not all that different at the motherboard/CPU level and is a lot cheaper than having to get stuff from IBM or Oracle/Sun.
The critical flaw size increases as you shrink the die.
Thus with all else being equal more flaws. Those flaws that previously were too small to matter now do.
I hope I made that clear enough.
While that is true there are a lot of things where working in parallel is trivial. Image processing is one that most readers will be familiar with. How hard is "apply this filter to every pixel"? You don't care what order it gets done in so long as you get the entire result dumped somewhere. There are a lot of things in science and engineering with that same sort of approach of applying the same transformation to every item in a dataset. While there is plenty of stuff that can't be done in parallel there is a lot of untouched "low hanging fruit" that is.
Currently the situation seems to be you pick Xeons for single threaded jobs and AMD if you need a huge number of cores (64 cores and 1TB of memory per node is cheaper than you would think). For those lucky enough to have tasks that don't need a lot of memory the GPU stuff gives you a vast number of cores.
You are seriously playing that "not educated enough" card as an adult? Do you really have that much contempt for the people around you?
Besides, "marginal" was not even mentioned until you engineered some petty "win" by moving goalposts. I did not question your use of it, I merely pointed out that you were changing the topic while claiming it was still the original topic.