No, you use the standard hyperthreading algorithm for when you have more cores than threads.
Schedule both threads on a single core. If either one uses its entire time slice, then move it to be alone in a core.
If two threads on different cores are each using "significantly" less than their entire timeslice, then try combining them on a single core.
You'll notice that none of these actually involve knowing anything about the processes themselves. While that may be useful for more fine-grained tuning, the simple algorithm above will cover 80% of cases in a "good enough" way.
At a secondary level, memory affinity, etc. can also drive what cores to try to put together to help with L2+ cache sharing, but this is to optimize the more difficult cases, or cases where there are more threads than cores.
1.) The world is BIG. Particularly the oceans. 2.) People are SMALL. Compared to the size of the Earth, people make up very little of it. 3.) People don't live very long (in a geologic sense.) 4.) Because of #1-#3, the affect appears to be very slow or even non-existent. It took lifetimes of global warming for people to even notice that it was happening. 5.) Because of #1-#4, it seems like a natural process. There have been lots of climate changes on geologic timescales, and people have trouble seeing the difference between "a few human generations" and tens of thousands of years. They both get lumped into the category of "a really long time." 6.) Conversely, for those who do understand the relative timescales, global warming may seem to be happening too fast. It's a bit fantastic to believe that man is doing in a few decades what takes nature centuries to do.
I can certainly believe reasonable people have a hard time believing in man-made global warming. We're so used to accepting it as fact that we don't realize how fantastic it is. People are changing the PLANET... in a matter of DECADES. We're ants moving a mountain, we just happen to be dumping that mountain right on top of ourselves.
dd is suboptimal for this because it doesn't zero out unused sectors. For linux filesystems, clone2fs is the equivalent command. I don't know if something similar is available on OS X.
I do this on my machines at home. Use ntfsclone and gzip to create compressed image files on a bootable linux partition. You can then create some custom scripts to ask for permission then restore the image automatically.
Actually, the quote wasn't quite that noble. What it's saying is that the person will only kill someone with the ability to defend themselves.
In a twisted way I see how they could have an argument.
If you dig a little deeper (like looking at the case on the FIRE site) the professor then put up a poster against fascism, indicating that fascism can lead to violence and death. Campus police took that one down too and got the dean involved, which is when this guy got a lawyer.
Seriously, Fascism?! Campus police has a problem with a poster against Fascism?!
Basically, what's going on here is that the professor had a poster that could, by a decidedly UNreasonable (but still sane and literate) person be construed to be a threat. Campus police took it down. The guy got upset and replaced it with a new poster which, while DEPICTING comic violence, constituted real political speech and clearly was NOT a threat of any kind. It was phrased as a warning that Fascism can lead to violence. This is where the story should have ended.
Campus police decided that since this guy was a "troublemaker" they would show him by taking down the new poster too and going after his job. This is where campus police went too far. The new poster was NOT a threat, and campus police knew it, or should have known it.
So, the professor got a lawyer.
And, the moral of the story is: Fear the police, they have public opinion, power, and guns on their side.:-(
A server you build yourself is NOT going to have similar power requirements to a mac mini. Quite frankly, 10w at idle is INSANELY low. You'll be lucky to be as low as 20w idle for just the processor and motherboard on a new build.
I spent $200 and upgraded my desktop from a 140w behemoth to a new machine that runs at 55w at idle, 40w at standby, including hard drives, fans, etc. I figure I will make the money back in 4 years, but the much nicer machine makes it worth it already.
For something like a server running on a Beagleboard, the hard drives will probably wind up being the main power hogs, and you might be surprised how much they draw.
Mr. Koomey, if we take your numbers from the attached article, which may not have been quoted correctly...
Feynman indicated that there was approximately 100 billion times efficiency improvement possible, and 40,000 times improvement has happened so far.
If we take Feynman's number at face value, this means that if computing efficiency improvements continue at the current rate (doubling every 18 months,) we will reach the theoretical maximum in 2043.
Based on that, do you believe that we will see a dramatic reduction in efficiency improvements in the next 10-20 years as we approach the theoretical limit, or do you think Feynman was conservative in his estimate?
I guess I was thinking JITs, or something like perl that is semi-compiled.
You're of course right that a simple straight-line interpreter can't do much optimization, but this isn't how modern javascript engines work. Getting a correctly formatted data set is part of the "write your code like this" idea I posted above.
My real point was, if it's possible to do, I would prefer the javascript engine do the work for me. That way I can write standards-compliant code that will compile and run on a machine without special vector language extensions, but that will be blazing fast on a machine supporting vector instructions.
Maybe I'm naive, but if I were a BIOS manufacturer, I would just have an option to "disable trusted boot" or "enable installation of new operating system" both with appropriate warnings about malware and lack of support. This setting would have to be protected after boot, but that should generally be possible. Only the truly experienced will go poking around with BIOS options, particularly those that are clearly labeled as being security related.
Windows 8 (at least) is going to allow an untrusted boot. Remember, they want people to install it on their existing PCs, as well as OEM ones. Windows 9 would be the first one to try this kind of game.
However, my guess is that this is being blown WAY out of proportion.
Then again, if it turns out to be true and Microsoft has just locked Linux out of all computers with an OEM copy of Windows, I will never buy ANY Microsoft product again. I'll miss buying new Xbox games.
I'll go you one better. I would hope that vector instructions are utilized by the compiler even for code that doesn't explicitly go after using them.
At a minimum, it should be possible to provide a "write your code like this, and it will be easy for the compiler to detect that what you really want is vector instructions."
However, sometimes interpreted languages can work against this due to things like strict exception checking.
If half the work is unparallelizable then the max theoretical speedup is 2x.
This is a simple application of Amdahl's law:
speedup = 1 / ( (1-P) + (P/S) ) where P is the amount of the workload that is parallelizable and S is the number of cores. speedup = 1 / ( (1-0.5) + (0.5/S) ) lim S-> infinity (speedup) is 1/ 0.5 = 2x
The likely reason the speedup appears superlinear here is that there are actually two speedups.
1.) Speedup from parallelizing on multiple threads. From looking at the usage graphs, this is probably about 4x. 2.) Speedup from vectorizing to use processor AVX extensions: This could be another 4x.
Total speedup: 16x.
A 16x speedup is totally believable for vectorizing and parallelizing a simple scientific simulation like the one shown in the video.
Yes, Microsoft is still evil, or at least the management team is.
Realize one thing: a corporation like Microsoft doesn't do anything without thinking through the larger implications.
1.) This simplifies the browser. That's a win for Microsoft in many ways, and a good technical decision. 2.) This will drive websites away from Flash. That helps lock people out of what is, unfortunately, the best current cross-platform solution for web video. 3.) This will also prevent implementation of WebM, locking people out of the best mid-term solution for cross-platform web video. 4.) Long-term (20 years+) H.264 is the best solution for cross-platform web video, so this is a good long-term win.
I suspect what happened here is that somebody on the code team suggested #1. High-level management considered 2-3, and decided that this was a good way to try and kill other OSes.
Silverlight support may also have been considered as part of this decision, as a Microsoft-only replacement for Flash.
I think everyone's experience will be different, dependent on the HS they attended, their teachers and the College/University they attended.
In my case the University did not have any special English courses for non-English Majors. You were just expected to be able to read/write etc. at a competent level. IMO this is how it should be although it was a shock to many 1st year students that they were expected to read, analyse, discuss and write about a new book every 1-2 weeks. These were significantly more difficult than the courses in the final year of HS. There were also so-called "remedial courses" for the many who weren't adequately prepared - but they didn't count for degree credit or for Science students looking to fill their Arts requirements.
What did happen though, and most of the people I knew in Science thought this was wrong, was that there were special "Intro to Chemistry/Physics/Biology for Arts Students" courses which were highly diluted but still counted as degree credit - for Arts students.
I would say that a University that has to make the 1st year courses no better than the final year of high school is not a very good university. In the case of my undergrad school they were turning out people with Arts degrees that had an unacceptably low level knowledge of Science.
To be fair, the English courses were not for non-majors, it's just that only majors typically progressed past the freshman-level course, which was quite easy. This freshman-level English was required of ALL students. It was not an Arts breadth course. Based on some of the TAs I got, I doubt some of them would pass even the basic English course. This was at Carnegie Mellon, so I doubt my overall Engineering education was poor, even if my English was.
Math becomes a real problem for non-Majors in the sciences. Calculus is a must for even freshman-level in-major Physics, which means only Chemistry and Biology would be open to non-Majors without also taking on an advanced math requirement or adding a "Physics for non-majors" course. As far as Chemistry and Biology go, I see no reason why those should require non-major flavors. I took freshman Chemistry as a breadth course towards my Electrical Engineering degree, and didn't find it inappropriately difficult.
On a side note, why is book count so important to an English course? I realize this could be considered important for a Literature course. For an English course, the primary goal is to improve your ability to read, write, and understand the language. I don't see how reading 100-400 pages of source material a week (a 200-400 page book every 1-2 weeks) really accomplishes that goal. I do see how this is necessary in a Literature course, where you can't learn the literature unless you've read it.
Breyer's dissent, unlike Thomas's, is well worth reading.
Breyer correctly recognizes that restrictions on violence in videogames, even to minors, requires strict scrutiny.
Breyer disagrees with Alito on whether or not the statute is vague, which is the primary reason why his is a dissenting rather than a concurring opinion. In making this determination he relies on historical literary references such as the works of Homer or Lord of the Flies. He ignores more modern children's literature/cinema which depicts violence that would fall under what he considers to be an acceptably narrow definition of violence. On this point I consider Breyer to simply be wrong, and Alito to be correct. The definition of violence needs to include "Nightmare on Elm Street" and exclude "Mulan," before we even get to the "community standards" portion of the statute.
Breyer then appears to commit a judicial no-no by doing his own research to determine if the research shows that violence in video games is harmful to children. Ultimately, though, he only uses this to support his claim that the courts should defer to the Legislature in determining if there is a causal link between video game violence and psychological harm to children. This is in contrast to the majority who basically says that unless the State can get further research, it is the opinion of the court that no link exists. Breyer makes a convincing argument on this point, to the extent that I'm not sure whether I agree with him or with the majority.
Breyer then argues that since we're only limiting the *sale* of video games to minors that the restriction is sufficiently narrow to be acceptable under strict scrutiny. This is basically the point I was uncomfortable with in the majority opinion, so I agree with him. If a sufficiently narrow statute (see Alito's concurring opinion) were drafted, I agree with Breyer and Alito that it might pass strict scrutiny, even though I disagree with Breyer and agree with Alito that the statute under consideration fails.
Where Breyer does seem a bit confused is that he tries to shoehorn an "action" component into playing a video game. It always seems a bit of a stretch when he does it, and I can't tell whether that's a failure on his part to understand or a failure on his part to articulate the point he's trying to make. To me, he fails to show how using a controller to play a video game is any different than using a remote control to play a movie, or a pencil to write text or draw a picture. Ultimately, his argument doesn't appear to hinge on this, so I tend to just ignore these parts.
In summary, while I disagree with Breyer on most points, his opinion is well worth reading to get a second look at how the case could have been considered.
Thomas's argument is that the First Amendment only protects you if you think like a Puritan, and it's horrific to me to see anyone on the Supreme Court who basically believes that children have no rights except those their parents give them.
The court's opinion shows that they "get it." Basically, they argue (convincingly) that violence, both by legal precedent and by community standards, is not allowed the same liberal restrictions as obscenity. Furthermore, the California law does not restrict speech in a specific way that sufficiently furthers any compelling state interest. While I think they go a little far in some places (particularly in their over-broad argument about the state interest for helping parents) I think they do an excellent job of laying out the issues involved and ruling correctly.
Justice Alito, in his concurring opinion, also makes some excellent points. I think he "gets it" too, he's just not willing to rule on the free-speech argument if he doesn't have to. In many ways, I think I would prefer his ruling to be the controlling one, as his would allow the court the leeway to allow a specific, well-defined law to be considered alone, instead of the current poorly-defined one.
Thomas' dissent is frightening. From his points, I wouldn't want this guy anywhere near the bench, not just on the Supreme Court. His standard would allow any restriction on speech to minors, up to and including a law that stated "It is illegal to speak to a minor without prior consent from their parent or legal guardian." The only thing of value in his dissent is a comment to the effect that Alito may not have been allowed to make the decision he did, based on procedure, and that only the majority opinion could be used to strike down the law at the Supreme Court level. Instead, Alito should have suggested remanding to the lower-level courts to decide. On that point I suspect he may be correct.
I haven't gotten to Breyer's dissent yet, and may comment on it later. From what I see, I disagree, but it is not nearly so dangerous of a view as Thomas.
For those saying they "got all that" in high school - there is just no comparing a university level English course and a high school English course.
While I generally agree with the point you're making, my high school English courses were significantly more difficult than my college English course. The college course was designed to be the minimum to ensure that you were a functional speaker, whereas the high school course (an AP course) was significantly more difficult. I specifically remember being praised highly for a paper in college that, upon re-reading, I would have been embarrassed to turn in to my high school English teacher.
In-major English was significantly more difficult, but those courses weren't required for my engineering degree.
I wound up taking freshman economics even though I APed out of it (basically to fill a graduation requirement) and I found it about the same level of difficulty as, or possibly a little easier than, my high school courses. This behavior isn't limited to freshman English courses.
My point is this: While I agree with you that non-major courses are important, it is quite possible to get a good grounding on these subjects in high school. Due to the desire for all reasonably qualified university students to pass these courses, they usually teach little more than was covered in high school, particularly AP-level high school courses. In many ways these wind up being "paying your dues" rather than really teaching anything of substance for students that went to a good high school, but are there to ensure a solid base for all students, including those whose high school education might not have been quite as good.
Yes, turning it off is THE solution. "Airplane mode" might count as turning it off, or removing the battery might be required, it depends how paranoid you are.
As far as waiting an extra minute in an emergency... yes, waiting one minute to turn on a cell phone is perfectly reasonable. We forget that 10 years ago, when cell phones were less common, the solution was:
"It's Sunday so everything's closed. I think I saw a gas station a half mile up the road. I'll either try to flag down a passing car and get them to go there and call 911, or I'll hike the half mile and use their phone." If 911 had been called within 5 minutes of the accident that was good even for an urban area.
In your heart attack case it's even easier. If you're at her place then use HER phone, she has no reason not to be at her own house and thus no reason to turn off her phone. If you're at a hotel use the hotel phone, even the sleazy hotels tend to have phones in the rooms. If it's at the office then use the office phone.
911 response times are realistically measured in minutes, not seconds. Waiting a minute for a phone to boot would seldom be the difference between life and death.
If the phone is on you can be tracked, period. It's left to the reader whether or not you think this requires a subpoena or warrant.
Traffic was diverted from a major highway onto a 2-lane road. This caused the buses to run late.
Because the buses were running late, everyone decided to take their own car to work. This further increased the amount of traffic on the tiny road.
The cops figured out that everyone was on the wrong road, and diverted traffic onto another freeway. However, by this point everyone was already taking their cars, so diverting to the other freeway didn't completely fix the problem.
All this traffic indirectly caused minor traffic problems in neighboring cities, because all the traffic cops in those cities were covering the traffic nightmare in this city.
Eventually, the cops got everyone to stop getting on the roads, and piecemeal managed to get people where they were going, which eventually cleaned things up.
Thank you for your explanation, which really is quite a good example.
However, the patronizing attitude is unnecessary. There are good reasons to be liberal (or maybe even socialist) even though you understand this productivity concept, and a person's misunderstanding of it may have nothing to do with their reasons for being liberal, such as:
#1) The "fiscal conservatives" have aligned themselves with the "religious conservatives" for so long that if I don't want my children being taught evolution in school I can't vote for a Republican. #2) Tragedy of the Commons means government regulation is sometimes required. #3) Disagreement about the need for and implementation of a "social safety net," to both protect those unable to work and to ensure the ability of people to return to productive work in the future.
Only #3 has anything to do with productivity, and represents an honest disagreement. "Most" people agree that we as a society should protect people who need protecting and give a kick in the pants to those who should be working. We however disagree on how to accomplish that goal and even about which part of the goal is most important.
And, to swing this around to the original topic... I don't know why anyone would be against cleaning up the mail-order sales tax loophole for the largest online retailers. I realize implementation would be a bit onerous, but if you're amazon you simply refuse to ship to a jurisdiction to which you don't know the correct sales tax rules. Online liquor sales already have to navigate the complexities of state-by-state liquor laws, so it's not like this is without precedent.
Congress gave the president both the authority and the responsibility to propose a federal budget to the President in The General Accounting Act of 1921 (the Budget Act.)
Congress has the power to take away that authority any time they like, but would have to establish new organizations to take over many of the necessary budgeting responsibilities of the OMB and the GAO.
I'm going to ignore the obvious economic collapse that would occur if all income earned by all Americans for an entire year went to the US government. (i.e. all to the government means no money for food, etc.) I think it's safe to assume this isn't the effect that the original poster was referring to, and they really just meant "if the amount collected by the government in a year was equivalent to the income of all Americans and American corporations in a year."
Increasing income taxes to 100% would most decidedly keep up with spending. Every dollar the government spends would be income for someone, who would probably also be paying some worker for their labor. That worker would then buy goods like food and clothes. Other workers at the retail store would be paid as a result of the goods purchases. Some of that money would be deposited in a bank, which would loan it to someone else and collect interest income. The recipient of the loan would also spend the money. With interest rates as low as they are now, this money would change hands many times over and be income many times over. The total income would be several times the amount spent by the government.
I'm not advocating a 100% tax as any kind of solution, but to imply that much money would be insufficient to cover spending is just incorrect.
Why would 4 cores be more than 4x faster than a single core? Cache.
Generally 4 cores is slower than a single core that is 4x as fast, but there can be exceptions where the effectively 4x as big L1 (sometimes L2) cache in the 4-core design makes it faster.
However, this isn't why four cores is a good thing in either a desktop or a server. The reason 4 cores is a good thing is that, given a fixed power envelope, you can only make a single core so fast. After that, it becomes both cheaper and more power efficient to build multiple cores than to try and build a really fast core. We've been in that space for a long time with servers (POWER4) and desktops (Athlon 64 X2), and we're realizing that the same thing makes sense for mobile processing. It's happening a lot faster with mobile because it's only ultra-low-power that allowed us to get into the mobile space in the first place, which makes upping the clock speed difficult.
Though I don't know any specifics, you probably pay a VERY small penalty for having extra turned-off cores on a mobile processor. The cores should be independently voltage and frequency controlled, and the clocks should be able to be stopped on an unused core. It may even be possible to turn off power to cores completely and have them wake back up in a clean state.
Leakage in a mobile processor is already going to be well-controlled, so the power cost of all that extra dormant silicon is going to be minimal. Overall, multi-core is the only way to significantly improve mobile performance, though it will take a fair amount of effort for the applications to be able to take advantage of it.
I really liked the proposal above to tax tires. Since generally tires wear due to the same forces that damage roads, tire wear costs would likely be proportional to road wear costs. This is probably even more true than the existing system that uses gasoline usage to approximate road wear costs.
Mileage is a HORRIBLE proxy for road wear costs. Heavy trucks do so much more damage per mile than passenger cars that a mileage tax is silly.
Actualy, it falls apart for a completely different reason.
In Omega vs. Costco the Supreme Court made it pretty clear that it isn't within the US court's ability to decide whether or not an overseas company was violating copyright law. The same reasoning would apply here, in that a US company should not be liable for something that an overseas supplier did that may or may not have been legal (a decision outside the jurisdiction of a US court) in their own country.
You can potentially modify the import laws to disallow imports of these kinds of goods, but import laws are federal, not state laws, and even there the reasoning would make this difficult.
Now, it should be possible to put this restriction on domestic suppliers, as in that case the court is qualified to make a decision on whether or not the supplier violated copyright law. This is just another instance where Omega vs. Costco creates an advantage for overseas suppliers that is not present for domestic suppliers.
And, of course, IANAL, and this is not legal advice.
As far as the privacy violations go, kids need to learn something to say something that very few do because of fear:
NO. Call my parents.
It's amazing how many examples of an over-reaching school could be significantly reduced by the student steadfastly refusing whatever injustice the school is perpetrating, and getting their own adult advocate (in the form of their parents) involved.
Unfortunately, most children are afraid of the "second punishment" that will occur once their parents get involved, and so usually try to keep the discipline confined to school. However, when the school oversteps its bounds (like requiring you to log in to your facebook account) it's time to get the parents involved FIRST.
Either the principal had evidence already or was fishing. If the principal already had evidence, then punish the kids who you have evidence on. If s/he's just fishing, then once the parents get involved, they can decide how to handle it. My guess is that the principal had evidence on the primary poster, but was fishing for more names that they could punish.
That said, as a student's parent I would accept action from the school ONLY with a written covenant not to sue from the teacher. In this case where no actual damages occurred suspension may be appropriate, expulsion is not. If expulsion got me a covenant not to sue I still might take it, but it's over-reaching for just mouthing off. The moment the kid is fraudulently telling law enforcement (or even trying to convince a school administrator) that the teacher is a pedophile, throw the book at them. At this age, the most appropriate response would seem to be explaining just how bad of an accusation pedophile is, along with an appropriate punishment to make it clear that it won't be tolerated.
Now, we don't have the actual facebook post. If it was of the form: "how can I accuse Mr. D of being a pedophile so that I can get him fired?" and other students were "helping," then now you have a conspiracy to commit fraud, and the punishments start to look more appropriate. I doubt that is what is going on here, based primarily on the number of punishments doled out for multiple independent infractions.
If my child went to that school, I would make very clear to them that the administration were a bunch of jerks, and that saying anything against them would likely get you severely punished, and have potential repercussions against them in the future. I then might explain to them what are appropriate complaints, and how to express them: up to and including peaceful protest of an overbearing school administrator.
Slashdot Mirror
No, you use the standard hyperthreading algorithm for when you have more cores than threads.
Schedule both threads on a single core. If either one uses its entire time slice, then move it to be alone in a core.
If two threads on different cores are each using "significantly" less than their entire timeslice, then try combining them on a single core.
You'll notice that none of these actually involve knowing anything about the processes themselves. While that may be useful for more fine-grained tuning, the simple algorithm above will cover 80% of cases in a "good enough" way.
At a secondary level, memory affinity, etc. can also drive what cores to try to put together to help with L2+ cache sharing, but this is to optimize the more difficult cases, or cases where there are more threads than cores.
Why is it so hard to believe?
1.) The world is BIG. Particularly the oceans.
2.) People are SMALL. Compared to the size of the Earth, people make up very little of it.
3.) People don't live very long (in a geologic sense.)
4.) Because of #1-#3, the affect appears to be very slow or even non-existent. It took lifetimes of global warming for people to even notice that it was happening.
5.) Because of #1-#4, it seems like a natural process. There have been lots of climate changes on geologic timescales, and people have trouble seeing the difference between "a few human generations" and tens of thousands of years. They both get lumped into the category of "a really long time."
6.) Conversely, for those who do understand the relative timescales, global warming may seem to be happening too fast. It's a bit fantastic to believe that man is doing in a few decades what takes nature centuries to do.
I can certainly believe reasonable people have a hard time believing in man-made global warming. We're so used to accepting it as fact that we don't realize how fantastic it is. People are changing the PLANET... in a matter of DECADES. We're ants moving a mountain, we just happen to be dumping that mountain right on top of ourselves.
dd is suboptimal for this because it doesn't zero out unused sectors. For linux filesystems, clone2fs is the equivalent command. I don't know if something similar is available on OS X.
I do this on my machines at home. Use ntfsclone and gzip to create compressed image files on a bootable linux partition. You can then create some custom scripts to ask for permission then restore the image automatically.
Actually, the quote wasn't quite that noble. What it's saying is that the person will only kill someone with the ability to defend themselves.
:-(
In a twisted way I see how they could have an argument.
If you dig a little deeper (like looking at the case on the FIRE site) the professor then put up a poster against fascism, indicating that fascism can lead to violence and death. Campus police took that one down too and got the dean involved, which is when this guy got a lawyer.
Seriously, Fascism?! Campus police has a problem with a poster against Fascism?!
Basically, what's going on here is that the professor had a poster that could, by a decidedly UNreasonable (but still sane and literate) person be construed to be a threat. Campus police took it down. The guy got upset and replaced it with a new poster which, while DEPICTING comic violence, constituted real political speech and clearly was NOT a threat of any kind. It was phrased as a warning that Fascism can lead to violence. This is where the story should have ended.
Campus police decided that since this guy was a "troublemaker" they would show him by taking down the new poster too and going after his job. This is where campus police went too far. The new poster was NOT a threat, and campus police knew it, or should have known it.
So, the professor got a lawyer.
And, the moral of the story is: Fear the police, they have public opinion, power, and guns on their side.
A server you build yourself is NOT going to have similar power requirements to a mac mini. Quite frankly, 10w at idle is INSANELY low. You'll be lucky to be as low as 20w idle for just the processor and motherboard on a new build.
I spent $200 and upgraded my desktop from a 140w behemoth to a new machine that runs at 55w at idle, 40w at standby, including hard drives, fans, etc. I figure I will make the money back in 4 years, but the much nicer machine makes it worth it already.
For something like a server running on a Beagleboard, the hard drives will probably wind up being the main power hogs, and you might be surprised how much they draw.
Mr. Koomey, if we take your numbers from the attached article, which may not have been quoted correctly...
Feynman indicated that there was approximately 100 billion times efficiency improvement possible, and 40,000 times improvement has happened so far.
If we take Feynman's number at face value, this means that if computing efficiency improvements continue at the current rate (doubling every 18 months,) we will reach the theoretical maximum in 2043.
Based on that, do you believe that we will see a dramatic reduction in efficiency improvements in the next 10-20 years as we approach the theoretical limit, or do you think Feynman was conservative in his estimate?
Thanks!
I guess I was thinking JITs, or something like perl that is semi-compiled.
You're of course right that a simple straight-line interpreter can't do much optimization, but this isn't how modern javascript engines work. Getting a correctly formatted data set is part of the "write your code like this" idea I posted above.
My real point was, if it's possible to do, I would prefer the javascript engine do the work for me. That way I can write standards-compliant code that will compile and run on a machine without special vector language extensions, but that will be blazing fast on a machine supporting vector instructions.
Maybe I'm naive, but if I were a BIOS manufacturer, I would just have an option to "disable trusted boot" or "enable installation of new operating system" both with appropriate warnings about malware and lack of support. This setting would have to be protected after boot, but that should generally be possible. Only the truly experienced will go poking around with BIOS options, particularly those that are clearly labeled as being security related.
Windows 8 (at least) is going to allow an untrusted boot. Remember, they want people to install it on their existing PCs, as well as OEM ones. Windows 9 would be the first one to try this kind of game.
However, my guess is that this is being blown WAY out of proportion.
Then again, if it turns out to be true and Microsoft has just locked Linux out of all computers with an OEM copy of Windows, I will never buy ANY Microsoft product again. I'll miss buying new Xbox games.
I'll go you one better. I would hope that vector instructions are utilized by the compiler even for code that doesn't explicitly go after using them.
At a minimum, it should be possible to provide a "write your code like this, and it will be easy for the compiler to detect that what you really want is vector instructions."
However, sometimes interpreted languages can work against this due to things like strict exception checking.
No.
If half the work is unparallelizable then the max theoretical speedup is 2x.
This is a simple application of Amdahl's law:
speedup = 1 / ( (1-P) + (P/S) )
where P is the amount of the workload that is parallelizable and S is the number of cores.
speedup = 1 / ( (1-0.5) + (0.5/S) )
lim S-> infinity (speedup) is 1/ 0.5 = 2x
The likely reason the speedup appears superlinear here is that there are actually two speedups.
1.) Speedup from parallelizing on multiple threads. From looking at the usage graphs, this is probably about 4x.
2.) Speedup from vectorizing to use processor AVX extensions: This could be another 4x.
Total speedup: 16x.
A 16x speedup is totally believable for vectorizing and parallelizing a simple scientific simulation like the one shown in the video.
Yes, Microsoft is still evil, or at least the management team is.
Realize one thing: a corporation like Microsoft doesn't do anything without thinking through the larger implications.
1.) This simplifies the browser. That's a win for Microsoft in many ways, and a good technical decision.
2.) This will drive websites away from Flash. That helps lock people out of what is, unfortunately, the best current cross-platform solution for web video.
3.) This will also prevent implementation of WebM, locking people out of the best mid-term solution for cross-platform web video.
4.) Long-term (20 years+) H.264 is the best solution for cross-platform web video, so this is a good long-term win.
I suspect what happened here is that somebody on the code team suggested #1. High-level management considered 2-3, and decided that this was a good way to try and kill other OSes.
Silverlight support may also have been considered as part of this decision, as a Microsoft-only replacement for Flash.
I think everyone's experience will be different, dependent on the HS they attended, their teachers and the College/University they attended.
In my case the University did not have any special English courses for non-English Majors. You were just expected to be able to read/write etc. at a competent level. IMO this is how it should be although it was a shock to many 1st year students that they were expected to read, analyse, discuss and write about a new book every 1-2 weeks. These were significantly more difficult than the courses in the final year of HS. There were also so-called "remedial courses" for the many who weren't adequately prepared - but they didn't count for degree credit or for Science students looking to fill their Arts requirements.
What did happen though, and most of the people I knew in Science thought this was wrong, was that there were special "Intro to Chemistry/Physics/Biology for Arts Students" courses which were highly diluted but still counted as degree credit - for Arts students.
I would say that a University that has to make the 1st year courses no better than the final year of high school is not a very good university. In the case of my undergrad school they were turning out people with Arts degrees that had an unacceptably low level knowledge of Science.
To be fair, the English courses were not for non-majors, it's just that only majors typically progressed past the freshman-level course, which was quite easy. This freshman-level English was required of ALL students. It was not an Arts breadth course. Based on some of the TAs I got, I doubt some of them would pass even the basic English course. This was at Carnegie Mellon, so I doubt my overall Engineering education was poor, even if my English was.
Math becomes a real problem for non-Majors in the sciences. Calculus is a must for even freshman-level in-major Physics, which means only Chemistry and Biology would be open to non-Majors without also taking on an advanced math requirement or adding a "Physics for non-majors" course. As far as Chemistry and Biology go, I see no reason why those should require non-major flavors. I took freshman Chemistry as a breadth course towards my Electrical Engineering degree, and didn't find it inappropriately difficult.
On a side note, why is book count so important to an English course? I realize this could be considered important for a Literature course. For an English course, the primary goal is to improve your ability to read, write, and understand the language. I don't see how reading 100-400 pages of source material a week (a 200-400 page book every 1-2 weeks) really accomplishes that goal. I do see how this is necessary in a Literature course, where you can't learn the literature unless you've read it.
Breyer's dissent, unlike Thomas's, is well worth reading.
Breyer correctly recognizes that restrictions on violence in videogames, even to minors, requires strict scrutiny.
Breyer disagrees with Alito on whether or not the statute is vague, which is the primary reason why his is a dissenting rather than a concurring opinion. In making this determination he relies on historical literary references such as the works of Homer or Lord of the Flies. He ignores more modern children's literature/cinema which depicts violence that would fall under what he considers to be an acceptably narrow definition of violence. On this point I consider Breyer to simply be wrong, and Alito to be correct. The definition of violence needs to include "Nightmare on Elm Street" and exclude "Mulan," before we even get to the "community standards" portion of the statute.
Breyer then appears to commit a judicial no-no by doing his own research to determine if the research shows that violence in video games is harmful to children. Ultimately, though, he only uses this to support his claim that the courts should defer to the Legislature in determining if there is a causal link between video game violence and psychological harm to children. This is in contrast to the majority who basically says that unless the State can get further research, it is the opinion of the court that no link exists. Breyer makes a convincing argument on this point, to the extent that I'm not sure whether I agree with him or with the majority.
Breyer then argues that since we're only limiting the *sale* of video games to minors that the restriction is sufficiently narrow to be acceptable under strict scrutiny. This is basically the point I was uncomfortable with in the majority opinion, so I agree with him. If a sufficiently narrow statute (see Alito's concurring opinion) were drafted, I agree with Breyer and Alito that it might pass strict scrutiny, even though I disagree with Breyer and agree with Alito that the statute under consideration fails.
Where Breyer does seem a bit confused is that he tries to shoehorn an "action" component into playing a video game. It always seems a bit of a stretch when he does it, and I can't tell whether that's a failure on his part to understand or a failure on his part to articulate the point he's trying to make. To me, he fails to show how using a controller to play a video game is any different than using a remote control to play a movie, or a pencil to write text or draw a picture. Ultimately, his argument doesn't appear to hinge on this, so I tend to just ignore these parts.
In summary, while I disagree with Breyer on most points, his opinion is well worth reading to get a second look at how the case could have been considered.
Thomas's argument is that the First Amendment only protects you if you think like a Puritan, and it's horrific to me to see anyone on the Supreme Court who basically believes that children have no rights except those their parents give them.
Agreed, Hear hear!
The court's opinion shows that they "get it." Basically, they argue (convincingly) that violence, both by legal precedent and by community standards, is not allowed the same liberal restrictions as obscenity. Furthermore, the California law does not restrict speech in a specific way that sufficiently furthers any compelling state interest. While I think they go a little far in some places (particularly in their over-broad argument about the state interest for helping parents) I think they do an excellent job of laying out the issues involved and ruling correctly.
Justice Alito, in his concurring opinion, also makes some excellent points. I think he "gets it" too, he's just not willing to rule on the free-speech argument if he doesn't have to. In many ways, I think I would prefer his ruling to be the controlling one, as his would allow the court the leeway to allow a specific, well-defined law to be considered alone, instead of the current poorly-defined one.
Thomas' dissent is frightening. From his points, I wouldn't want this guy anywhere near the bench, not just on the Supreme Court. His standard would allow any restriction on speech to minors, up to and including a law that stated "It is illegal to speak to a minor without prior consent from their parent or legal guardian." The only thing of value in his dissent is a comment to the effect that Alito may not have been allowed to make the decision he did, based on procedure, and that only the majority opinion could be used to strike down the law at the Supreme Court level. Instead, Alito should have suggested remanding to the lower-level courts to decide. On that point I suspect he may be correct.
I haven't gotten to Breyer's dissent yet, and may comment on it later. From what I see, I disagree, but it is not nearly so dangerous of a view as Thomas.
For those saying they "got all that" in high school - there is just no comparing a university level English course and a high school English course.
While I generally agree with the point you're making, my high school English courses were significantly more difficult than my college English course. The college course was designed to be the minimum to ensure that you were a functional speaker, whereas the high school course (an AP course) was significantly more difficult. I specifically remember being praised highly for a paper in college that, upon re-reading, I would have been embarrassed to turn in to my high school English teacher.
In-major English was significantly more difficult, but those courses weren't required for my engineering degree.
I wound up taking freshman economics even though I APed out of it (basically to fill a graduation requirement) and I found it about the same level of difficulty as, or possibly a little easier than, my high school courses. This behavior isn't limited to freshman English courses.
My point is this: While I agree with you that non-major courses are important, it is quite possible to get a good grounding on these subjects in high school. Due to the desire for all reasonably qualified university students to pass these courses, they usually teach little more than was covered in high school, particularly AP-level high school courses. In many ways these wind up being "paying your dues" rather than really teaching anything of substance for students that went to a good high school, but are there to ensure a solid base for all students, including those whose high school education might not have been quite as good.
Yes, turning it off is THE solution. "Airplane mode" might count as turning it off, or removing the battery might be required, it depends how paranoid you are.
As far as waiting an extra minute in an emergency... yes, waiting one minute to turn on a cell phone is perfectly reasonable. We forget that 10 years ago, when cell phones were less common, the solution was:
"It's Sunday so everything's closed. I think I saw a gas station a half mile up the road. I'll either try to flag down a passing car and get them to go there and call 911, or I'll hike the half mile and use their phone." If 911 had been called within 5 minutes of the accident that was good even for an urban area.
In your heart attack case it's even easier. If you're at her place then use HER phone, she has no reason not to be at her own house and thus no reason to turn off her phone. If you're at a hotel use the hotel phone, even the sleazy hotels tend to have phones in the rooms. If it's at the office then use the office phone.
911 response times are realistically measured in minutes, not seconds. Waiting a minute for a phone to boot would seldom be the difference between life and death.
If the phone is on you can be tracked, period. It's left to the reader whether or not you think this requires a subpoena or warrant.
Traffic was diverted from a major highway onto a 2-lane road. This caused the buses to run late.
Because the buses were running late, everyone decided to take their own car to work. This further increased the amount of traffic on the tiny road.
The cops figured out that everyone was on the wrong road, and diverted traffic onto another freeway. However, by this point everyone was already taking their cars, so diverting to the other freeway didn't completely fix the problem.
All this traffic indirectly caused minor traffic problems in neighboring cities, because all the traffic cops in those cities were covering the traffic nightmare in this city.
Eventually, the cops got everyone to stop getting on the roads, and piecemeal managed to get people where they were going, which eventually cleaned things up.
Thank you for your explanation, which really is quite a good example.
However, the patronizing attitude is unnecessary. There are good reasons to be liberal (or maybe even socialist) even though you understand this productivity concept, and a person's misunderstanding of it may have nothing to do with their reasons for being liberal, such as:
#1) The "fiscal conservatives" have aligned themselves with the "religious conservatives" for so long that if I don't want my children being taught evolution in school I can't vote for a Republican.
#2) Tragedy of the Commons means government regulation is sometimes required.
#3) Disagreement about the need for and implementation of a "social safety net," to both protect those unable to work and to ensure the ability of people to return to productive work in the future.
Only #3 has anything to do with productivity, and represents an honest disagreement. "Most" people agree that we as a society should protect people who need protecting and give a kick in the pants to those who should be working. We however disagree on how to accomplish that goal and even about which part of the goal is most important.
And, to swing this around to the original topic...
I don't know why anyone would be against cleaning up the mail-order sales tax loophole for the largest online retailers. I realize implementation would be a bit onerous, but if you're amazon you simply refuse to ship to a jurisdiction to which you don't know the correct sales tax rules. Online liquor sales already have to navigate the complexities of state-by-state liquor laws, so it's not like this is without precedent.
Congress gave the president both the authority and the responsibility to propose a federal budget to the President in The General Accounting Act of 1921 (the Budget Act.)
Congress has the power to take away that authority any time they like, but would have to establish new organizations to take over many of the necessary budgeting responsibilities of the OMB and the GAO.
I'm wondering how this got modded up.
I'm going to ignore the obvious economic collapse that would occur if all income earned by all Americans for an entire year went to the US government. (i.e. all to the government means no money for food, etc.) I think it's safe to assume this isn't the effect that the original poster was referring to, and they really just meant "if the amount collected by the government in a year was equivalent to the income of all Americans and American corporations in a year."
Increasing income taxes to 100% would most decidedly keep up with spending. Every dollar the government spends would be income for someone, who would probably also be paying some worker for their labor. That worker would then buy goods like food and clothes. Other workers at the retail store would be paid as a result of the goods purchases. Some of that money would be deposited in a bank, which would loan it to someone else and collect interest income. The recipient of the loan would also spend the money. With interest rates as low as they are now, this money would change hands many times over and be income many times over. The total income would be several times the amount spent by the government.
I'm not advocating a 100% tax as any kind of solution, but to imply that much money would be insufficient to cover spending is just incorrect.
Why would 4 cores be more than 4x faster than a single core? Cache.
Generally 4 cores is slower than a single core that is 4x as fast, but there can be exceptions where the effectively 4x as big L1 (sometimes L2) cache in the 4-core design makes it faster.
However, this isn't why four cores is a good thing in either a desktop or a server. The reason 4 cores is a good thing is that, given a fixed power envelope, you can only make a single core so fast. After that, it becomes both cheaper and more power efficient to build multiple cores than to try and build a really fast core. We've been in that space for a long time with servers (POWER4) and desktops (Athlon 64 X2), and we're realizing that the same thing makes sense for mobile processing. It's happening a lot faster with mobile because it's only ultra-low-power that allowed us to get into the mobile space in the first place, which makes upping the clock speed difficult.
Though I don't know any specifics, you probably pay a VERY small penalty for having extra turned-off cores on a mobile processor. The cores should be independently voltage and frequency controlled, and the clocks should be able to be stopped on an unused core. It may even be possible to turn off power to cores completely and have them wake back up in a clean state.
Leakage in a mobile processor is already going to be well-controlled, so the power cost of all that extra dormant silicon is going to be minimal. Overall, multi-core is the only way to significantly improve mobile performance, though it will take a fair amount of effort for the applications to be able to take advantage of it.
I really liked the proposal above to tax tires. Since generally tires wear due to the same forces that damage roads, tire wear costs would likely be proportional to road wear costs. This is probably even more true than the existing system that uses gasoline usage to approximate road wear costs.
Mileage is a HORRIBLE proxy for road wear costs. Heavy trucks do so much more damage per mile than passenger cars that a mileage tax is silly.
Actualy, it falls apart for a completely different reason.
In Omega vs. Costco the Supreme Court made it pretty clear that it isn't within the US court's ability to decide whether or not an overseas company was violating copyright law. The same reasoning would apply here, in that a US company should not be liable for something that an overseas supplier did that may or may not have been legal (a decision outside the jurisdiction of a US court) in their own country.
You can potentially modify the import laws to disallow imports of these kinds of goods, but import laws are federal, not state laws, and even there the reasoning would make this difficult.
Now, it should be possible to put this restriction on domestic suppliers, as in that case the court is qualified to make a decision on whether or not the supplier violated copyright law. This is just another instance where Omega vs. Costco creates an advantage for overseas suppliers that is not present for domestic suppliers.
And, of course, IANAL, and this is not legal advice.
NO. Call my parents.
It's amazing how many examples of an over-reaching school could be significantly reduced by the student steadfastly refusing whatever injustice the school is perpetrating, and getting their own adult advocate (in the form of their parents) involved.
Unfortunately, most children are afraid of the "second punishment" that will occur once their parents get involved, and so usually try to keep the discipline confined to school. However, when the school oversteps its bounds (like requiring you to log in to your facebook account) it's time to get the parents involved FIRST.
Either the principal had evidence already or was fishing. If the principal already had evidence, then punish the kids who you have evidence on. If s/he's just fishing, then once the parents get involved, they can decide how to handle it. My guess is that the principal had evidence on the primary poster, but was fishing for more names that they could punish.
That said, as a student's parent I would accept action from the school ONLY with a written covenant not to sue from the teacher. In this case where no actual damages occurred suspension may be appropriate, expulsion is not. If expulsion got me a covenant not to sue I still might take it, but it's over-reaching for just mouthing off. The moment the kid is fraudulently telling law enforcement (or even trying to convince a school administrator) that the teacher is a pedophile, throw the book at them. At this age, the most appropriate response would seem to be explaining just how bad of an accusation pedophile is, along with an appropriate punishment to make it clear that it won't be tolerated.
Now, we don't have the actual facebook post. If it was of the form: "how can I accuse Mr. D of being a pedophile so that I can get him fired?" and other students were "helping," then now you have a conspiracy to commit fraud, and the punishments start to look more appropriate. I doubt that is what is going on here, based primarily on the number of punishments doled out for multiple independent infractions.
If my child went to that school, I would make very clear to them that the administration were a bunch of jerks, and that saying anything against them would likely get you severely punished, and have potential repercussions against them in the future. I then might explain to them what are appropriate complaints, and how to express them: up to and including peaceful protest of an overbearing school administrator.