Half of these sound extremely fake. Most of these look like it's VERY unlikely they would get themselves on this list if it didn't gain them money...
Not that I tell myself these guys collectively contributed even 1% of those amounts... very strange names here. Were the pressured into signing this ?
Yes, basically.
The teamsters would have signed up because they get work from the motion picture industry. Union solidarity pulls in AFL-CIO, electrical workers, fire fighters, construction workers, and law enforcement.
The rest sound fake because they are. Mostly large businesses and rich people who fund private associations to commit their sins, so they don't get literally executed when the revolution comes.
The object files will contain redundancies, such as symbols and template instantiations. Without good header management the linker (and profiler, since this is done per translation unit) will need vastly more memory.
Thank you for posting this. Filling the virtual address space of the linker probably does indicate some problems with the Firefox source code - crazy big translation units, for example - but it doesn't imply anything about the size or quality of the resulting binary.
I thought people on this site were supposed to know something about computers.
Hey you businesses! Any of you want to pay a decent wage for all those vocational/technical jobs you're screaming for?
(crickets)
(Hmmm...I wonder if there's a correlation there...)
Seriously.
If there's a shortage of qualified people in a field, the answer isn't to "encourage" (read: throw money at) the schools teaching in the field. The answer is for employers to man up, quit whining to the government, and pay the clearing wage.
EA, Activision/Blizzard,... anybody else who has a stick up their butt about Steam.
I use one game service: Steam. I have a lot of games on it. I'm also not huge into games any more, so almost everything I download is an impulse buy (usually during Steam sales.) There is absolutely zero chance that EA or its subsidiaries will ever make an Arab hunting simulation good enough that I would actually go out of my way to sign up for their service.
Steam allows it to persist because they want to serve the interests of their customers and their shareholders.
Steam is only a good service if it carries the games that people want to play. Most people want games from big publishers, and the big publishers won't sell to a store that won't let them conduct business as usual. They just aren't smart enough.
The CC info was encrypted in the database, and Sony used a separate internal-facing server to handle credit card transactions. The problem is, the transaction server wasn't configured properly; unencrypted credit card numbers and billing information were being recorded in Apache logs.
No. Changes, e.g. in transistor and wafer size, require new equipment almost everywhere. Changes in transistor size (or technology) require higher accuracy and sometimes totally new lithography technologies. Changes in wafer size require new cleaning, dicing and packaging machines. These machines are extremely precise and purpose-built. They are not the kind of thing you can strip down and re-purpose.
Yes, seconding this. People just don't seem to realize the completely insane amount of money it takes. It's also not a one-time payment; any time you change technologies you literally need to build an entire new facility.
There are a ton of American semiconductor manufacturers which cannot reasonably afford to run their own fabs. Qualcomm, Broadcom, Conexant, Marvell, NVIDIA and Apple are all fabless. Even AMD and Intel outsource some work to TSMC. When Intel, the world's largest semiconductor company, balks at the idea of building enough manufacturing capacity, it should give you an idea about how much it would cost.
That's certainly possible, or two 256-bit SIMD units (one add/div/sqrt, one mult) at double precision. That means a lot of overhead is showing up in the supercomputer benchmark, though. Usually they aren't so honest.
If politicians understood the reason for scientific and cultural pursuits, they wouldn't need to be politicians. It's almost tautological that a country would have bad laws regarding such things.
According to my back-of-the-napkin calculations, and their claimed performance, each CPU in this supercomputer is comitting 6.5 floating point instructions per clock. I don't know the specific engineering details of this new CPU, but the one it's based on is four-issue superscalar with two floating point units, one for multiplication and the other for all other operations. Either they made extensive modifications, in which case I would question why they needed to license the design in the first place, or their performance measurements are fabricated. I'm leaning toward the latter. Either way, their floating point performance is a complete joke compared to even low-end GPUs.
One thing to note is that FLOPS is completely useless as a measure for performance. Basically it is up to the supercomputer vendor to choose between measuring only the fastest operations (adds,) or to use some arbitrary convention for weighting the other operations (multiply, divide, sqrt.) They also eliminate all overhead; they run benchmarks with all operands in registers, eliminate read-after-write dependencies, and only measure after the branch predictor and I-cache/trace cache are in steady state. Completely unrealistic. The entire supercomputer industry is like this, by the way. (Even if they really did make their CPU 7- or 8-way superscalar, any realistic workload is going to cut the instruction throughput by more than half.)
The classical treatments argue that mathematics is discovered because there is some ephemeral "world" of all possible abstract thoughts, including all possible systems of mathematics for all possible choices of consistent axioms. It's a useful thought experiment, but that's about it. For example, this world would necessarily include all possible novels, because all novels of finite length can be uniquely represented by natural numbers. We can't "discover" a novel, because randomly choosing novels will never yield one that has any meaning. In order to choose a novel that has meaning, we need to invent it first. (If you don't quite understand what I mean, try reading the short story "The Library of Babel" by Jorge Luis Borges.)
Note that the classical interpretation differs from the contemporary interpretation. The modern opinion isn't based on an epistemological argument, it's based on the layperson's misconception of mathematics: that mathematics is some sort of transcendental model of the natural world. You can't blame people for thinking this way, since most people - even at the university level - never encounter mathematics outside of the rote.
The fact is, we choose axioms. We choose them all the time. Sometimes we choose them because they self-evidently reflect reality (e.g. Euclid's axioms) and sometimes we choose them because it is convenient (e.g. completeness, the axiom of choice.) There's nothing stopping you from devising your own totally original system of mathematics, especially now that you no longer have a teacher around to call you stupid for doing so.
Re:Why not focus on quality instead of major revs?
on
Ubuntu Turns 7
·
· Score: 1
Really.
For the past two versions of Ubuntu I have had issues with mouse focus and clicks on my desktop. It's due to a bug that was first reported and confirmed in 2006, but nobody has ever given enough of a shit to fix it.
If I'm stuck with patching and compiling xorg, why am I running Ubuntu? I might as well just run Gentoo. Better yet, I might as well run OSX or Windows, where I know I will never have a problem this stupid.
The median is, itself, a statement about the probability distribution for the data. You can't say anything intelligent about a median based on a probability distribution you assumed, because the act of assuming a probability distribution also assumes the median.
They haven't provided enough data to make that kind of conclusion. I wish they would release more, but if I ran the thing I definitely wouldn't. If the humble bundle people are intelligent at all they know how much of a competitive advantage this sort of market research can give them.
Some tl;dr about why this data can't be used to estimate potential revenues:
The data they've shown can only be used to make a make a conclusion if you can show the following: 1.) the people who bought the bundle are representative of potential customers as a whole, 2.) the charity aspect does not affect sales or prices, and 3.) it is possible to engage in perfect price discrimination, i.e. choose the highest price that all customers are willing to pay.
I don't think the first two are correct, but they might be reasonable assumptions. The third point is completely impossible. The reason this is a problem is because the mean is strongly influenced by outliers: if the mean price is being inflated by a very small number of extremely generous people, it will be much higher than the profit-maximizing price in general. If the sales prices are skewed left, you'll make a lot more money by lowering the price than you will by catering to a few rich people, but the total amount of money isn't going to match the estimate you see here. You'd need to follow rich people around all day, forcing them to pay $2000 for every game they buy, and that's not realistically possible.
Some more considerations: modern games can be developed relatively cheaply because they rely on middleware, but sometimes those packages require the purchase of additional licenses for other platforms. Each new platform also adds a lot of QA and support costs, especially for a platform like Linux which is so nebulous. (It's nice to say that the Linux community will support the game for you but you can't rely on volunteer tech support. You can't even endorse it, because that might be illegal - e.g. AOL CL class action judgement.)
Notch single-handedly raised the grand average purchase price by a cent. That's a pretty big difference for one person.
The mean is strongly influenced by outliers, which is why it's generally avoided in academic discussions about things like income. With the mean I can make a reasonable prediction about a *group* of buyers, but I can't make a reasonable prediction about a *single* buyer. For example, it's true that Linux users gave proportionally the most as a group, but you can't use the same evidence to say that Linux users are more generous in general.
Example: suppose you are given a mean sale of $9.20 spread across 70,000 buyers. You could achieve this average if each of the 70,000 paid $9.20. You could also achieve this average if 322 people spent $2000, and the remaining 69,678 people only spent $0.01. The averages are the same, but the people in the first group are much, much more generous than the people in the second.
The median is more constructive for this discussion. The median means that half of the people are above it, and half of the people are below it. If I know $9.20 is the median for Linux and $4.11 is the median for Windows, I *can* say that Linux users are generally more generous, because they are individually much more likely to pay more money than Windows users.
The weekly trends on Channel 9 and MSDN blogs don't dictate company policy. They aren't deprecating C#. In fact, the situation for C# will be much better in Windows 8 than it is right now: Metro is all managed, and you'll be able to access WinRT directly (no more P/Invoke hell like with Win32 API.)
There are lots of reasons to use C++ on Windows: it's faster, supports intrinsics, offers a superior energy efficiency, does not have an embarrassing pause, trivially offers convenient access to the native APIs, and has a compile-time Turing complete purely functional metaprogramming language. And, yes, it also has namespaces and string objects with bounds-checking. None of these features are new.
You're doing your career a huge disservice by limiting yourself to one language.
Slashdot Mirror
Half of these sound extremely fake. Most of these look like it's VERY unlikely they would get themselves on this list if it didn't gain them money ...
Not that I tell myself these guys collectively contributed even 1% of those amounts ... very strange names here. Were the pressured into signing this ?
Yes, basically.
The teamsters would have signed up because they get work from the motion picture industry. Union solidarity pulls in AFL-CIO, electrical workers, fire fighters, construction workers, and law enforcement.
The rest sound fake because they are. Mostly large businesses and rich people who fund private associations to commit their sins, so they don't get literally executed when the revolution comes.
Everything except the Riemann hypothesis. Fuck complex analysis.
"MIT is overrated because I can't get into MIT."
Yeah. Keep telling yourself that.
My house has too many rats. I need to throw in some more so they have competition.
The object files will contain redundancies, such as symbols and template instantiations. Without good header management the linker (and profiler, since this is done per translation unit) will need vastly more memory.
Thank you for posting this. Filling the virtual address space of the linker probably does indicate some problems with the Firefox source code - crazy big translation units, for example - but it doesn't imply anything about the size or quality of the resulting binary.
I thought people on this site were supposed to know something about computers.
Regular 15 minute breaks.
Hey you businesses! Any of you want to pay a decent wage for all those vocational/technical jobs you're screaming for?
(crickets)
(Hmmm...I wonder if there's a correlation there...)
Seriously.
If there's a shortage of qualified people in a field, the answer isn't to "encourage" (read: throw money at) the schools teaching in the field. The answer is for employers to man up, quit whining to the government, and pay the clearing wage.
EA, Activision/Blizzard,... anybody else who has a stick up their butt about Steam.
I use one game service: Steam. I have a lot of games on it. I'm also not huge into games any more, so almost everything I download is an impulse buy (usually during Steam sales.) There is absolutely zero chance that EA or its subsidiaries will ever make an Arab hunting simulation good enough that I would actually go out of my way to sign up for their service.
Steam allows it to persist because they want to serve the interests of their customers and their shareholders.
Steam is only a good service if it carries the games that people want to play. Most people want games from big publishers, and the big publishers won't sell to a store that won't let them conduct business as usual. They just aren't smart enough.
Has anyone?
Be warned, the following is only hearsay:
The CC info was encrypted in the database, and Sony used a separate internal-facing server to handle credit card transactions. The problem is, the transaction server wasn't configured properly; unencrypted credit card numbers and billing information were being recorded in Apache logs.
No. Changes, e.g. in transistor and wafer size, require new equipment almost everywhere. Changes in transistor size (or technology) require higher accuracy and sometimes totally new lithography technologies. Changes in wafer size require new cleaning, dicing and packaging machines. These machines are extremely precise and purpose-built. They are not the kind of thing you can strip down and re-purpose.
Yes, seconding this. People just don't seem to realize the completely insane amount of money it takes. It's also not a one-time payment; any time you change technologies you literally need to build an entire new facility.
There are a ton of American semiconductor manufacturers which cannot reasonably afford to run their own fabs. Qualcomm, Broadcom, Conexant, Marvell, NVIDIA and Apple are all fabless. Even AMD and Intel outsource some work to TSMC. When Intel, the world's largest semiconductor company, balks at the idea of building enough manufacturing capacity, it should give you an idea about how much it would cost.
That's certainly possible, or two 256-bit SIMD units (one add/div/sqrt, one mult) at double precision. That means a lot of overhead is showing up in the supercomputer benchmark, though. Usually they aren't so honest.
If politicians understood the reason for scientific and cultural pursuits, they wouldn't need to be politicians. It's almost tautological that a country would have bad laws regarding such things.
"Why Alpha?" indeed.
According to my back-of-the-napkin calculations, and their claimed performance, each CPU in this supercomputer is comitting 6.5 floating point instructions per clock. I don't know the specific engineering details of this new CPU, but the one it's based on is four-issue superscalar with two floating point units, one for multiplication and the other for all other operations. Either they made extensive modifications, in which case I would question why they needed to license the design in the first place, or their performance measurements are fabricated. I'm leaning toward the latter. Either way, their floating point performance is a complete joke compared to even low-end GPUs.
One thing to note is that FLOPS is completely useless as a measure for performance. Basically it is up to the supercomputer vendor to choose between measuring only the fastest operations (adds,) or to use some arbitrary convention for weighting the other operations (multiply, divide, sqrt.) They also eliminate all overhead; they run benchmarks with all operands in registers, eliminate read-after-write dependencies, and only measure after the branch predictor and I-cache/trace cache are in steady state. Completely unrealistic. The entire supercomputer industry is like this, by the way. (Even if they really did make their CPU 7- or 8-way superscalar, any realistic workload is going to cut the instruction throughput by more than half.)
This, exactly.
The classical treatments argue that mathematics is discovered because there is some ephemeral "world" of all possible abstract thoughts, including all possible systems of mathematics for all possible choices of consistent axioms. It's a useful thought experiment, but that's about it. For example, this world would necessarily include all possible novels, because all novels of finite length can be uniquely represented by natural numbers. We can't "discover" a novel, because randomly choosing novels will never yield one that has any meaning. In order to choose a novel that has meaning, we need to invent it first. (If you don't quite understand what I mean, try reading the short story "The Library of Babel" by Jorge Luis Borges.)
Note that the classical interpretation differs from the contemporary interpretation. The modern opinion isn't based on an epistemological argument, it's based on the layperson's misconception of mathematics: that mathematics is some sort of transcendental model of the natural world. You can't blame people for thinking this way, since most people - even at the university level - never encounter mathematics outside of the rote.
The fact is, we choose axioms. We choose them all the time. Sometimes we choose them because they self-evidently reflect reality (e.g. Euclid's axioms) and sometimes we choose them because it is convenient (e.g. completeness, the axiom of choice.) There's nothing stopping you from devising your own totally original system of mathematics, especially now that you no longer have a teacher around to call you stupid for doing so.
Really.
For the past two versions of Ubuntu I have had issues with mouse focus and clicks on my desktop. It's due to a bug that was first reported and confirmed in 2006, but nobody has ever given enough of a shit to fix it.
If I'm stuck with patching and compiling xorg, why am I running Ubuntu? I might as well just run Gentoo. Better yet, I might as well run OSX or Windows, where I know I will never have a problem this stupid.
The median is, itself, a statement about the probability distribution for the data. You can't say anything intelligent about a median based on a probability distribution you assumed, because the act of assuming a probability distribution also assumes the median.
No, you can't estimate the median with the given information.
They haven't provided enough data to make that kind of conclusion. I wish they would release more, but if I ran the thing I definitely wouldn't. If the humble bundle people are intelligent at all they know how much of a competitive advantage this sort of market research can give them.
Some tl;dr about why this data can't be used to estimate potential revenues:
The data they've shown can only be used to make a make a conclusion if you can show the following:
1.) the people who bought the bundle are representative of potential customers as a whole,
2.) the charity aspect does not affect sales or prices, and
3.) it is possible to engage in perfect price discrimination, i.e. choose the highest price that all customers are willing to pay.
I don't think the first two are correct, but they might be reasonable assumptions. The third point is completely impossible. The reason this is a problem is because the mean is strongly influenced by outliers: if the mean price is being inflated by a very small number of extremely generous people, it will be much higher than the profit-maximizing price in general. If the sales prices are skewed left, you'll make a lot more money by lowering the price than you will by catering to a few rich people, but the total amount of money isn't going to match the estimate you see here. You'd need to follow rich people around all day, forcing them to pay $2000 for every game they buy, and that's not realistically possible.
Some more considerations: modern games can be developed relatively cheaply because they rely on middleware, but sometimes those packages require the purchase of additional licenses for other platforms. Each new platform also adds a lot of QA and support costs, especially for a platform like Linux which is so nebulous. (It's nice to say that the Linux community will support the game for you but you can't rely on volunteer tech support. You can't even endorse it, because that might be illegal - e.g. AOL CL class action judgement.)
Notch single-handedly raised the grand average purchase price by a cent. That's a pretty big difference for one person.
The mean is strongly influenced by outliers, which is why it's generally avoided in academic discussions about things like income. With the mean I can make a reasonable prediction about a *group* of buyers, but I can't make a reasonable prediction about a *single* buyer. For example, it's true that Linux users gave proportionally the most as a group, but you can't use the same evidence to say that Linux users are more generous in general.
Example: suppose you are given a mean sale of $9.20 spread across 70,000 buyers. You could achieve this average if each of the 70,000 paid $9.20. You could also achieve this average if 322 people spent $2000, and the remaining 69,678 people only spent $0.01. The averages are the same, but the people in the first group are much, much more generous than the people in the second.
The median is more constructive for this discussion. The median means that half of the people are above it, and half of the people are below it. If I know $9.20 is the median for Linux and $4.11 is the median for Windows, I *can* say that Linux users are generally more generous, because they are individually much more likely to pay more money than Windows users.
The weekly trends on Channel 9 and MSDN blogs don't dictate company policy. They aren't deprecating C#. In fact, the situation for C# will be much better in Windows 8 than it is right now: Metro is all managed, and you'll be able to access WinRT directly (no more P/Invoke hell like with Win32 API.)
There are lots of reasons to use C++ on Windows: it's faster, supports intrinsics, offers a superior energy efficiency, does not have an embarrassing pause, trivially offers convenient access to the native APIs, and has a compile-time Turing complete purely functional metaprogramming language. And, yes, it also has namespaces and string objects with bounds-checking. None of these features are new.
You're doing your career a huge disservice by limiting yourself to one language.
How to survive a plane crash:
Step 1: Fly a major, well-established airline.
Step 2: Leave a family wealthy enough to successfully sue the airline in the event of your death.