"Don't Worry About Global Warming, Say 16 Scientists" Subhead: 1. We're Screwed Anyway 2. Do Worry About Global Warming, Say The Rest 3. Exxon, British Petroleum Agree 4. Yes, Gerontologists are Scientists Too 5. Worry About Climate Change Instead 6. Pay No Attention to the Money Behind Our Curtain 7. There Is No Spoon
Microsoft and their partners sell operating systems and hardware. They can and should be able to do what they like with it, including locking it down so you can't do something you'd like to.
On the other hand, it's probably not the best business decision; mostly from the bad press. They probably won't lose a whole lot of sales from us nerds that want to install other OSes on it. But it's their decision to make. If you don't like their decision, then you shouldn't buy their products. If it's that big of a deal, then market pressure might force them to change.
Stupid? Yes. Illegal? Probably not, at least on the emerging tablet market.
First question: Why do you want it from Intel, versus anybody else? They've always struck me as moderately evil - the Microsoft of the chip world, looking out for their own sales numbers and not much else.
Second question: Which do you want in your chip; a fast CPU that can run your web browser and E-mail client, or a fast parallel computing unit that's good for gene sequencing, multimedia processing, etc? You can't have both. Well, you can, but both parts will be slower. The top-of-the-line chips you get these days are as big as can be reasonably manufactured, and produce as much heat as can be reasonably removed from the chip (without requiring you to supply a source of liquid nitrogen). You can combine them, but you won't get great performance from either the CPU or the GPU/"parallel unit" in that case. You're better off buying two chips. And Intel has a pretty poor history at making graphics chips with reasonable performance.
You should know that ECO mode does nothing other than remap the gas pedal a little bit (and turn down the maximum air conditioning level). If you floor it in ECO and floor it in PWR, it'll accelerate exactly as fast. It's more about reprogramming the driver - many new American cars have a very jumpy accelerator to make it feel like you have a powerful car (this is what PWR mode does). ECO mode does the opposite, making it feel like you really have to press hard to get acceleration. That way, for the car to work hard, you have to work hard too.
I love my Prius too. And, quite frankly, I don't care whether the added cost of the hybrid powertrain pays for itself. Will my car's bluetooth system or upgraded speakers ever pay for themselves? How about the radar cruise control?
For that matter, does the CAR even pay for itself? I could probably have just rented a car on the rare occasions I need one, or joined a car sharing organization, and spent less money than I do now for gas, registration, insurance, depreciation, etc. I got a car because I wanted one. I got Bluetooth and a better radio because I enjoy listening to music, and I like keeping it all on my phone. I got the radar cruise control because I'm a huge nerd, and (even though I don't drive long distances very often) I like supporting advancements in technology. I got a Prius because it's a good car from a reliable manufacturer that has low emissions.
Ask the guy with the Mustang how long it's going to take before he recoups the added cost over a Fiesta, and you'll get laughed at. I think we need to have the same mindset for saving fuel and reducing emissions - it's not just about what it costs me up front and what it's going to save me in the long run. That is just an added bonus for many of us.
While emissions and amount of fuel burned are correlated, they're not fixed ratios. For example, if not all the fuel is burned, there will be less CO2 and more hydrocarbons in the emissions. And if it uses a lean mixture to make sure that all the fuel gets burned, that increases the NOx emissions. Running in the engine's most efficient band makes it run hotter, also increasing NOx - but exhaust gas recirculation is used on many engines to dilute the fuel/air mixture with something that doesn't burn, cooling it down and reducing NOx. Additionally, the catalytic converter design varies by vehicle, and its performance will also affect the emissions. CO and hydrocarbon emissions are increased in cases of incomplete combustion.
The Prius was definitely primarily designed to have low emissions; good fuel economy is a byproduct of that. For example, the engine won't shut off (even if you're stopped) until it has warmed itself and the catalytic converter. This helps reduce emissions, but increases fuel consumption, especially in cold climates in the winter. (Extra power is produced that just charges the battery - losing some percentage of efficiency that it would not if the engine were powering the wheels instead.)
I live high in the hills, and by the time I'm at home the battery is usually on its last couple of bars. This is normal and it has no ill effects. In fact, the battery still retains about half of its charge at that time.
And on the third gen Prius (2010-), it's even less likely to deplete the battery going up a big hill. The engine is slightly larger (1.8L versus 1.5L), so it's a bit more efficient at providing a large amount of power to go up the hill. And in general, Toyota seems to have decided that the battery is best used for only very light demands, while I think the earlier models tended to use it at both very light demand and very heavy demand.
I regularly drive up some mountains at 45-50 mph; the battery is always at the same level at the top as it was at the bottom. Sometimes, that's actually a little frustrating, because I know I can recover some energy on the way back down. But not depleting the battery is much more efficient than depleting the battery and then having to recharge it from the engine.
It actually has two systems. One is the Lane Departure Warning that will beep and torque the steering wheel when you leave your lane. The other, Lane Keep Assist, is active when you turn on both LDW and the radar cruise control. It will keep you more towards the center of the lane, and is perhaps most noticeable when you are in a long curve, as the car will hold the steering wheel for you; no effort is required to hold the turn. However, it's obviously not trying to be anything more than a safety system and fatigue reducer; if you let it steer itself, you will bobble annoyingly back and forth within the lane, rather than stay perfectly in the middle. A big downside is that gradual exit ramps, shadows from overpasses, poor lane markings, and many other things cause it to lose sight of the lanes (it beeps twice and stops running for at least 3 seconds or so, or until it can find the lane again). Also, it only has enough torque to steer a very gentle turn, though it can hold a much steeper turn by itself.
With the radar cruise control, it's fairly effortless. It slows down for traffic, and points you in the right-ish direction. It needs a lot of improvement before it could ever be self-driving, and it was obviously designed to assist, not replace, the driver.
In my car, yes, that is the case. Also, the torque applied to the steering wheel to keep you in your lane is pretty minimal; even grandma would have no trouble overpowering the motor to, say, make an emergency lane change to avoid an accident.
My Prius already does this. But I guess maybe that's a "hatchback", not a "sedan". And admittedly, it's an option available only on the premium configuration, so there might be an argument to be made of whether it's "mainstream" or not. Or maybe they don't consider it a "midsize" vehicle; I hear it has the interior volume of a midsize, but a smaller wheelbase than many midsize cars. It is, at least, "North American", though.
Point being, every one of those qualifiers is there because somebody else really did it first. Not that I'm claiming the Prius did it first, mind you.
It is a nice feature though, and I hope a baby step towards self-driving cars.
Agreed. This guy wasn't saved by a miracle, just by statistics. Even in the smallest of homes, a falling object is more likely to hit unoccupied space than a person, even if it's quite crowded.
For that matter, it doesn't even sound like the object penetrated the roof; the story says it "crashed into the roof" and the owner had to go outside before he saw the damage.
34 is a perfectly reasonable number with which to say "hey, there's something interesting going on here, maybe we could learn something interesting if we studied it more." Hence why the interviewees note several curiosities they have, such as whether it's a generational difference (they suspect it might be), and whether radio station personalities do it or not based on their target demographic. You're right, if you're trying to just make a be all and end all study, 34 is fairly low.
It seems bafflingly weird that PR would provide a number so grossly wrong, since the fanboys and the haters basically make no real use of the number and the people who really care should be able to easily detect a lie of that magnitude; but I'd be somewhat surprised if the original PR numbers meant all that much.
IANALE (I am not a layout engineer), but it's my understanding that it is not an easy task to actually figure out how many transistors are contained within a modern chip. The CAD tools used aren't anything like Photoshop, where you can pop up an info window and see how many pixels it has. There are many different pieces within a chip - some might be standard library cells (like building a chip out of legos, rather than making a custom injection mold and filling it with plastic - where each building block is a few to a few thousand transistors (or more?). Other parts might be a full-custom layout, where somebody hand-placed every transistor to get the highest speed, lowest power, smallest area, or some combination of all of those. The chip might also include some hard macros, IP that is purchased from another company (like a memory controller or power manager) and just plopped onto the chip, with no insight to what is actually inside. There are hierarchies, and some parts (like cores or cache sub-blocks) are replicated a couple times, or a couple thousand times.
So it's my indication that any time you hear one of these numbers, it's really just an estimate anyway. Probably some engineer at AMD heard the 2B number after PR trumpeted it, thought it sounded a little high, and found a mistake in somebody's estimate.
Or, maybe more likely, marketing just made crap up without actually asking any of the engineers. That happens a lot too, and it pisses us engineers off to no end. At least when they do it after the product is made, PR has to fess up. When they do it before a product is finalized, it usually means engineering has to scramble and actually make it do whatever marketing promised.
For most devices these days (even desktops), there's a TDP ceiling for the design. If a chip is a watt over that (or in smartphones, a milliwatt over - no joke), you have no product. Only if you fit within that TDP does performance matter. In that sense, power consumption is a far greater factor. If your power consumption is too high, you don't even get benchmarked. You don't just lose the game, you never even make it on the field.
That said, SemiAccurate and its owner are in general full of crap. I'm thankful when sites cite the original source, so I can go find that grain of salt and not follow the link to give him more pageviews.
In any case, as one can see from the list, the best GPU machine manages to beat the K machines by a factor of 1.66, a far cry from the factor of 3-6 you originally claimed. And most GPU machines fall behind the K.
I was going by the per-chip benchmarks of both the SPARC and of GPUs, since that was what was at hand.
I think most of these GPU clusters have very high-powered CPUs (eg Xeons) in them. I wonder what the perf per watt would be if you replaced the Xeon with, say, a low-power embedded chip, like an ARM (or an Atom, if you must). Granted, that further limits the types of computations that your supercomputer can do, but it might get you a lot closer to that 3-6x factor for the chip alone.
Large GPGPU clusters are still a relatively new phenomenon, give it a few years and I suspect you'll see a lot more of them.
No doubt. When I was doing my master's research, they were just starting to look interesting. I was working in reconfigurable hardware, but had one or two peers playing around with GPUs. That's one of those things that people keep looking at, and occasionally use for something, but has never really taken off like GPGPU has.
What you're looking for is the Green500 list, where the K computer is #6; behind an ATI GPU+Intel i5 cluster and two NVIDIA Fermi+Intel Xeon clusters (two IBM Blue Gene Q prototypes sit at the top). The first three are fairly small (100+ on the top500), but the NVIDIA systems sit at #5 and #54 on the top500 as well, so it doesn't appear to be a scalability issue. I have no knowledge of the design tradeoffs of the individual systems, but I'd say that it's fairly impressive that both the top500 and the Green500 have so many GPUs in the top 10, given that they're both CPU-dominated lists.
A high-end GPU will do 5 to 20 times as many FLOPS, at 3-6x the efficiency. Quite possibly at a lower cost, too, given that they're relatively high-volume parts, compared to a SPARC. And before anybody complains about "well yeah, but GPUs can't do X", well, that's what they get for using FLOPS for a benchmark. Unless what they're interested in is FLOPS, they really shouldn't use them as a benchmark!
Testing somebody on something they should have learned is a better way to hire a trivia expert, if you ask me. When employers give this sort of a test, they don't really care if you already know why manhole covers are round, or how many dump trucks it would take to move a mountain - they want to see how you tackle a problem you've never seen before. The whole point is that there's no way to prepare for this sort of thing - either you can handle a new problem, know when to ask questions, and break it down into solvable pieces; or you look at the problem as an unsolvable whole, and give up without even really trying. After hiring somebody, I'm not going to be asking them to write down a data structure they learned from a book in college; I'm going to give them a problem to solve, and as a part of it, they need to recognize a need for a data structure, and make an intelligent decision about which one is the best fit for the job.
Does this approach flunk good coders? Absolutely. Does it pass bad coders? Certainly. But, used as one tool among several in the interview process, it has a place. Those who do well with puzzles like this are more likely to be good at coding than those who fail these puzzles, so it's a good way to filter out candidates. If your company doesn't have enough applicants, then maybe you can't afford to lose a single good coder - but these high-profile employers often attract hundreds or thousands of resumes for each open position. They can afford to be picky, and to filter out some good candidates as a cost of filtering out many bad candidates.
In a similar vein, my employer has a strict 3.5 minimum GPA for new college hires (I assume this doesn't apply to experienced professionals, but I don't know). Are there good candidates that got a 3.4? Definitely. There are probably a couple that just don't do well in school, and get a 2-point-something, but would be fantastic coders. But filtering by GPA is a good way to eliminate most of the bottom 50% of candidates when you have 100 resumes per position, even if it takes out a few of the top 50% along the way.
Oh, I'm not skeptical of whether he parked in handicapped spaces, or didn't have a license plate, or was in general an ass that violated many typical social norms, or anything else Andy Hertzfeld says in his articles - I'm skeptical that he found a way to park in handicapped spaces legally, which is not one of Andy's claims.
As a random anecdote, I saw a plateless Mercedes parked outside a big-name tech company (whose products Apple uses) in the valley once... I figure it couldn't have been Steve, because the car wasn't parked in a handicapped space.
I'm still not convinced that Steve even arranged this "lease for 6 months" thing - it wouldn't be the first time a story had been made up about how Steve did this legally. It's not hard to see how - people hear about Steve's habit, then think up a way that it could be done legally, it gets repeated a few times, and eventually the game of telephone means that the story about how it could be done legally is no longer somebody's theory, it's fact about how Steve does it.
I'm additionally skeptical because the article says its source for the story didn't start working for Apple until 2010, though Linkedin says he worked there from 1995-1997 and omits the more recent experience, and during both of those periods, Steve was not as omnipresent as he might have been at other times. Additionally, there's no indication that he had any reason to be close to Steve. At best, this story has gone through several other people first. Another final stab - the California Vehicle Code says that you can operate a vehicle without plates for 6 months, or until the plates are received by the owner, whichever comes first; typically, that'd be less than 2 months.
The most notable previous explanation I've heard was that his license plate was stolen frequently, so California gave him a bar code instead (which is visible in the photos where the license plate would normally be). That story doesn't hold up for very long though, because every Mercedes has such a bar code - it's not from the state.
My money is still just on the fact that he didn't care. If you feel the laws don't apply to you, you don't find legal ways to get around the laws - you ignore them. Finding a way to make it legal is an exercise reserved for the rest of us, who follow the rules.
When I was young, I visited my sister, who trains astronauts at NASA, and an astronaut gave me some of their food - a packet of cocoa, creamed spinach, and a brownie. The cocoa is pretty normal - powder in a silver bag. The spinach looks absolutely disgusting, even more so than spinach usually does, from the plastic crinkled around its odd texture when it was vacuum sealed. And the brownie is one of the rare foods that wasn't freeze dried; just sealed.
It was pretty cool getting the insider's tour; we'd see the official tours go by, two floors up in a glass-sided hallway. But we'd be down on the floor, looking at everything close-up. I even got to try one of their simulators briefly (I think playing with the fake robotic arm), and sat in the back seat while a couple astronauts trained in a different simulator.
And in the same strain of disappointing mainstream science reporting, basically everything I've read on this seems to focus on the "we made cats glow green" part, rather than the more serious part of the research, which is to protect against AIDS. The gene to make the cat's fur fluoresce green is just a marker so that they know which cats will also have the gene that could protect against AIDS, and can then run experiments to see if that gene works as they hope. That article seems to be better than average, in that it largely focuses on the AIDS part, and keeps the "omg glowing cats" restrained to the headline and a few spots in the body.
I'm just waiting for the article about the research dog that harbors a cure for cancer, but more importantly, it can skateboard!
Slashdot Mirror
"Don't Worry About Global Warming, Say 16 Scientists"
Subhead:
1. We're Screwed Anyway
2. Do Worry About Global Warming, Say The Rest
3. Exxon, British Petroleum Agree
4. Yes, Gerontologists are Scientists Too
5. Worry About Climate Change Instead
6. Pay No Attention to the Money Behind Our Curtain
7. There Is No Spoon
Maybe we can get an autonomous congress at the same time, taking the worst part of government out of the equation - the politician.
Microsoft and their partners sell operating systems and hardware. They can and should be able to do what they like with it, including locking it down so you can't do something you'd like to.
On the other hand, it's probably not the best business decision; mostly from the bad press. They probably won't lose a whole lot of sales from us nerds that want to install other OSes on it. But it's their decision to make. If you don't like their decision, then you shouldn't buy their products. If it's that big of a deal, then market pressure might force them to change.
Stupid? Yes. Illegal? Probably not, at least on the emerging tablet market.
First question: Why do you want it from Intel, versus anybody else? They've always struck me as moderately evil - the Microsoft of the chip world, looking out for their own sales numbers and not much else.
Second question: Which do you want in your chip; a fast CPU that can run your web browser and E-mail client, or a fast parallel computing unit that's good for gene sequencing, multimedia processing, etc? You can't have both. Well, you can, but both parts will be slower. The top-of-the-line chips you get these days are as big as can be reasonably manufactured, and produce as much heat as can be reasonably removed from the chip (without requiring you to supply a source of liquid nitrogen). You can combine them, but you won't get great performance from either the CPU or the GPU/"parallel unit" in that case. You're better off buying two chips. And Intel has a pretty poor history at making graphics chips with reasonable performance.
You should know that ECO mode does nothing other than remap the gas pedal a little bit (and turn down the maximum air conditioning level). If you floor it in ECO and floor it in PWR, it'll accelerate exactly as fast. It's more about reprogramming the driver - many new American cars have a very jumpy accelerator to make it feel like you have a powerful car (this is what PWR mode does). ECO mode does the opposite, making it feel like you really have to press hard to get acceleration. That way, for the car to work hard, you have to work hard too.
I love my Prius too. And, quite frankly, I don't care whether the added cost of the hybrid powertrain pays for itself. Will my car's bluetooth system or upgraded speakers ever pay for themselves? How about the radar cruise control?
For that matter, does the CAR even pay for itself? I could probably have just rented a car on the rare occasions I need one, or joined a car sharing organization, and spent less money than I do now for gas, registration, insurance, depreciation, etc. I got a car because I wanted one. I got Bluetooth and a better radio because I enjoy listening to music, and I like keeping it all on my phone. I got the radar cruise control because I'm a huge nerd, and (even though I don't drive long distances very often) I like supporting advancements in technology. I got a Prius because it's a good car from a reliable manufacturer that has low emissions.
Ask the guy with the Mustang how long it's going to take before he recoups the added cost over a Fiesta, and you'll get laughed at. I think we need to have the same mindset for saving fuel and reducing emissions - it's not just about what it costs me up front and what it's going to save me in the long run. That is just an added bonus for many of us.
While emissions and amount of fuel burned are correlated, they're not fixed ratios. For example, if not all the fuel is burned, there will be less CO2 and more hydrocarbons in the emissions. And if it uses a lean mixture to make sure that all the fuel gets burned, that increases the NOx emissions. Running in the engine's most efficient band makes it run hotter, also increasing NOx - but exhaust gas recirculation is used on many engines to dilute the fuel/air mixture with something that doesn't burn, cooling it down and reducing NOx. Additionally, the catalytic converter design varies by vehicle, and its performance will also affect the emissions. CO and hydrocarbon emissions are increased in cases of incomplete combustion.
The Prius was definitely primarily designed to have low emissions; good fuel economy is a byproduct of that. For example, the engine won't shut off (even if you're stopped) until it has warmed itself and the catalytic converter. This helps reduce emissions, but increases fuel consumption, especially in cold climates in the winter. (Extra power is produced that just charges the battery - losing some percentage of efficiency that it would not if the engine were powering the wheels instead.)
I live high in the hills, and by the time I'm at home the battery is usually on its last couple of bars. This is normal and it has no ill effects. In fact, the battery still retains about half of its charge at that time.
And on the third gen Prius (2010-), it's even less likely to deplete the battery going up a big hill. The engine is slightly larger (1.8L versus 1.5L), so it's a bit more efficient at providing a large amount of power to go up the hill. And in general, Toyota seems to have decided that the battery is best used for only very light demands, while I think the earlier models tended to use it at both very light demand and very heavy demand.
I regularly drive up some mountains at 45-50 mph; the battery is always at the same level at the top as it was at the bottom. Sometimes, that's actually a little frustrating, because I know I can recover some energy on the way back down. But not depleting the battery is much more efficient than depleting the battery and then having to recharge it from the engine.
...because I hate giving pageviews to morons so that I can see what half-baked ideas they have.
http://webcache.googleusercontent.com/search?q=cache:ZxS4pENaxJ0J:spectator.org/archives/2012/01/06/honda-civic-lesson/print+http://spectator.org/archives/2012/01/06/honda-civic-lesson
Of course, given that all the images and other resources still load, it probably still counts as a pageview as far as they're concerned.
It actually has two systems. One is the Lane Departure Warning that will beep and torque the steering wheel when you leave your lane. The other, Lane Keep Assist, is active when you turn on both LDW and the radar cruise control. It will keep you more towards the center of the lane, and is perhaps most noticeable when you are in a long curve, as the car will hold the steering wheel for you; no effort is required to hold the turn. However, it's obviously not trying to be anything more than a safety system and fatigue reducer; if you let it steer itself, you will bobble annoyingly back and forth within the lane, rather than stay perfectly in the middle. A big downside is that gradual exit ramps, shadows from overpasses, poor lane markings, and many other things cause it to lose sight of the lanes (it beeps twice and stops running for at least 3 seconds or so, or until it can find the lane again). Also, it only has enough torque to steer a very gentle turn, though it can hold a much steeper turn by itself.
With the radar cruise control, it's fairly effortless. It slows down for traffic, and points you in the right-ish direction. It needs a lot of improvement before it could ever be self-driving, and it was obviously designed to assist, not replace, the driver.
In my car, yes, that is the case. Also, the torque applied to the steering wheel to keep you in your lane is pretty minimal; even grandma would have no trouble overpowering the motor to, say, make an emergency lane change to avoid an accident.
My Prius already does this. But I guess maybe that's a "hatchback", not a "sedan". And admittedly, it's an option available only on the premium configuration, so there might be an argument to be made of whether it's "mainstream" or not. Or maybe they don't consider it a "midsize" vehicle; I hear it has the interior volume of a midsize, but a smaller wheelbase than many midsize cars. It is, at least, "North American", though.
Point being, every one of those qualifiers is there because somebody else really did it first. Not that I'm claiming the Prius did it first, mind you.
It is a nice feature though, and I hope a baby step towards self-driving cars.
Agreed. This guy wasn't saved by a miracle, just by statistics. Even in the smallest of homes, a falling object is more likely to hit unoccupied space than a person, even if it's quite crowded.
For that matter, it doesn't even sound like the object penetrated the roof; the story says it "crashed into the roof" and the owner had to go outside before he saw the damage.
Sounds like it could be a profitable job. How does one apply?
Purely for satisfying curiosity, of course. I'd never steal money from children who can't count. It's much more fun to steal it from the ones who can.
34 is a perfectly reasonable number with which to say "hey, there's something interesting going on here, maybe we could learn something interesting if we studied it more." Hence why the interviewees note several curiosities they have, such as whether it's a generational difference (they suspect it might be), and whether radio station personalities do it or not based on their target demographic. You're right, if you're trying to just make a be all and end all study, 34 is fairly low.
It seems bafflingly weird that PR would provide a number so grossly wrong, since the fanboys and the haters basically make no real use of the number and the people who really care should be able to easily detect a lie of that magnitude; but I'd be somewhat surprised if the original PR numbers meant all that much.
IANALE (I am not a layout engineer), but it's my understanding that it is not an easy task to actually figure out how many transistors are contained within a modern chip. The CAD tools used aren't anything like Photoshop, where you can pop up an info window and see how many pixels it has. There are many different pieces within a chip - some might be standard library cells (like building a chip out of legos, rather than making a custom injection mold and filling it with plastic - where each building block is a few to a few thousand transistors (or more?). Other parts might be a full-custom layout, where somebody hand-placed every transistor to get the highest speed, lowest power, smallest area, or some combination of all of those. The chip might also include some hard macros, IP that is purchased from another company (like a memory controller or power manager) and just plopped onto the chip, with no insight to what is actually inside. There are hierarchies, and some parts (like cores or cache sub-blocks) are replicated a couple times, or a couple thousand times.
So it's my indication that any time you hear one of these numbers, it's really just an estimate anyway. Probably some engineer at AMD heard the 2B number after PR trumpeted it, thought it sounded a little high, and found a mistake in somebody's estimate.
Or, maybe more likely, marketing just made crap up without actually asking any of the engineers. That happens a lot too, and it pisses us engineers off to no end. At least when they do it after the product is made, PR has to fess up. When they do it before a product is finalized, it usually means engineering has to scramble and actually make it do whatever marketing promised.
For most devices these days (even desktops), there's a TDP ceiling for the design. If a chip is a watt over that (or in smartphones, a milliwatt over - no joke), you have no product. Only if you fit within that TDP does performance matter. In that sense, power consumption is a far greater factor. If your power consumption is too high, you don't even get benchmarked. You don't just lose the game, you never even make it on the field.
That said, SemiAccurate and its owner are in general full of crap. I'm thankful when sites cite the original source, so I can go find that grain of salt and not follow the link to give him more pageviews.
In any case, as one can see from the list, the best GPU machine manages to beat the K machines by a factor of 1.66, a far cry from the factor of 3-6 you originally claimed. And most GPU machines fall behind the K.
I was going by the per-chip benchmarks of both the SPARC and of GPUs, since that was what was at hand.
I think most of these GPU clusters have very high-powered CPUs (eg Xeons) in them. I wonder what the perf per watt would be if you replaced the Xeon with, say, a low-power embedded chip, like an ARM (or an Atom, if you must). Granted, that further limits the types of computations that your supercomputer can do, but it might get you a lot closer to that 3-6x factor for the chip alone.
Large GPGPU clusters are still a relatively new phenomenon, give it a few years and I suspect you'll see a lot more of them.
No doubt. When I was doing my master's research, they were just starting to look interesting. I was working in reconfigurable hardware, but had one or two peers playing around with GPUs. That's one of those things that people keep looking at, and occasionally use for something, but has never really taken off like GPGPU has.
What you're looking for is the Green500 list, where the K computer is #6; behind an ATI GPU+Intel i5 cluster and two NVIDIA Fermi+Intel Xeon clusters (two IBM Blue Gene Q prototypes sit at the top). The first three are fairly small (100+ on the top500), but the NVIDIA systems sit at #5 and #54 on the top500 as well, so it doesn't appear to be a scalability issue. I have no knowledge of the design tradeoffs of the individual systems, but I'd say that it's fairly impressive that both the top500 and the Green500 have so many GPUs in the top 10, given that they're both CPU-dominated lists.
A high-end GPU will do 5 to 20 times as many FLOPS, at 3-6x the efficiency. Quite possibly at a lower cost, too, given that they're relatively high-volume parts, compared to a SPARC. And before anybody complains about "well yeah, but GPUs can't do X", well, that's what they get for using FLOPS for a benchmark. Unless what they're interested in is FLOPS, they really shouldn't use them as a benchmark!
Testing somebody on something they should have learned is a better way to hire a trivia expert, if you ask me. When employers give this sort of a test, they don't really care if you already know why manhole covers are round, or how many dump trucks it would take to move a mountain - they want to see how you tackle a problem you've never seen before. The whole point is that there's no way to prepare for this sort of thing - either you can handle a new problem, know when to ask questions, and break it down into solvable pieces; or you look at the problem as an unsolvable whole, and give up without even really trying. After hiring somebody, I'm not going to be asking them to write down a data structure they learned from a book in college; I'm going to give them a problem to solve, and as a part of it, they need to recognize a need for a data structure, and make an intelligent decision about which one is the best fit for the job.
Does this approach flunk good coders? Absolutely. Does it pass bad coders? Certainly. But, used as one tool among several in the interview process, it has a place. Those who do well with puzzles like this are more likely to be good at coding than those who fail these puzzles, so it's a good way to filter out candidates. If your company doesn't have enough applicants, then maybe you can't afford to lose a single good coder - but these high-profile employers often attract hundreds or thousands of resumes for each open position. They can afford to be picky, and to filter out some good candidates as a cost of filtering out many bad candidates.
In a similar vein, my employer has a strict 3.5 minimum GPA for new college hires (I assume this doesn't apply to experienced professionals, but I don't know). Are there good candidates that got a 3.4? Definitely. There are probably a couple that just don't do well in school, and get a 2-point-something, but would be fantastic coders. But filtering by GPA is a good way to eliminate most of the bottom 50% of candidates when you have 100 resumes per position, even if it takes out a few of the top 50% along the way.
Oh, I'm not skeptical of whether he parked in handicapped spaces, or didn't have a license plate, or was in general an ass that violated many typical social norms, or anything else Andy Hertzfeld says in his articles - I'm skeptical that he found a way to park in handicapped spaces legally, which is not one of Andy's claims.
As a random anecdote, I saw a plateless Mercedes parked outside a big-name tech company (whose products Apple uses) in the valley once... I figure it couldn't have been Steve, because the car wasn't parked in a handicapped space.
I'm still not convinced that Steve even arranged this "lease for 6 months" thing - it wouldn't be the first time a story had been made up about how Steve did this legally. It's not hard to see how - people hear about Steve's habit, then think up a way that it could be done legally, it gets repeated a few times, and eventually the game of telephone means that the story about how it could be done legally is no longer somebody's theory, it's fact about how Steve does it.
I'm additionally skeptical because the article says its source for the story didn't start working for Apple until 2010, though Linkedin says he worked there from 1995-1997 and omits the more recent experience, and during both of those periods, Steve was not as omnipresent as he might have been at other times. Additionally, there's no indication that he had any reason to be close to Steve. At best, this story has gone through several other people first. Another final stab - the California Vehicle Code says that you can operate a vehicle without plates for 6 months, or until the plates are received by the owner, whichever comes first; typically, that'd be less than 2 months.
The most notable previous explanation I've heard was that his license plate was stolen frequently, so California gave him a bar code instead (which is visible in the photos where the license plate would normally be). That story doesn't hold up for very long though, because every Mercedes has such a bar code - it's not from the state.
My money is still just on the fact that he didn't care. If you feel the laws don't apply to you, you don't find legal ways to get around the laws - you ignore them. Finding a way to make it legal is an exercise reserved for the rest of us, who follow the rules.
When I was young, I visited my sister, who trains astronauts at NASA, and an astronaut gave me some of their food - a packet of cocoa, creamed spinach, and a brownie. The cocoa is pretty normal - powder in a silver bag. The spinach looks absolutely disgusting, even more so than spinach usually does, from the plastic crinkled around its odd texture when it was vacuum sealed. And the brownie is one of the rare foods that wasn't freeze dried; just sealed.
It was pretty cool getting the insider's tour; we'd see the official tours go by, two floors up in a glass-sided hallway. But we'd be down on the floor, looking at everything close-up. I even got to try one of their simulators briefly (I think playing with the fake robotic arm), and sat in the back seat while a couple astronauts trained in a different simulator.
And in the same strain of disappointing mainstream science reporting, basically everything I've read on this seems to focus on the "we made cats glow green" part, rather than the more serious part of the research, which is to protect against AIDS. The gene to make the cat's fur fluoresce green is just a marker so that they know which cats will also have the gene that could protect against AIDS, and can then run experiments to see if that gene works as they hope. That article seems to be better than average, in that it largely focuses on the AIDS part, and keeps the "omg glowing cats" restrained to the headline and a few spots in the body.
I'm just waiting for the article about the research dog that harbors a cure for cancer, but more importantly, it can skateboard!