I'll hazard a guess that clock speed excels in one particular case: tight-loop iteration. You can't do that with parallelism (ignoring some fancy pipelining to get part-way there). The fastest way to get the 1 millionth result in a no-shortcut iterative sequence is to get the loop processing at the highest frequency possible.
I was about to post something similar to this parent (saved my typing!), so I'll just add a 'diff': $5 won't stop someone trying to make 10,000 times that in a scam.
There's loads of money to be had selling a device has closed content delivery - the iPad is good at that. However, it is not really a device for being creative, which is what a computer excels at, and it's harder for a hardware manufacturer to make money out of that, unless you feed all output into a closed content delivery system...
I'm not sure what the author wants to make, in real terms. Is it the user base, the third-party integration, the hardware infrastructure? Yes, all of the above, as they're all necessary, but that requires a new build, with different policies. However the viability reduces to the reality of money: money to build it, and money to sustain it. Facebook is now starting to explain the cost of Free.
Here's a question: Can any diffusely-owned project or data be trusted? Does it require that all members of the project or support infrastructure are also trusted, or must there be a certificate-based identity/trust system to unlock the data on various levels?
A subtractive cell (stacked CMYK layers, to filter out R,G,B,All respectively) would let more light through than separate R, G, B windows. The article alludes to using a primary subtractive cell (Y) to help one combination, but it would only be 100% brighter for saturated yellows (not whites); CMYK would be about 250% brighter for all colours (not just yellow) with very good blacks.
Then he would sue the manufacturers of his garden gate for not stopping vandals, and they would counter-sue for not evaluating fit-for-purpose before purchase. The judge would counter-sue for failing to provide information that the gate was not suitable for stopping vandals,...
Yes, this is all about anonymous postings, but surely anyone can make up an identity online? Law has a habit of applying judgements to other cases (in the same country), and encourages prosecutors to take a punt in other countries. In what other cases would this frustrate the everyday running of the web? ISPs failing to moderate comments from their customers? Allowing file sharing?
This model seems to follow 'genetic programming' principles, but is flawed in many ways: (a) It assumes that most people know everything relevant to the problem under consideration - they often don't.
(b) What the model is looking for is an expert among the crowd. On average, you can find an expert among 1024 people, to predict 10 coin tosses - this is with random data having no relation to specialized wisdom.
(c) Eurovision (mentioned above) is in the rare category of scenarios that can make use of 'crowdsourcing prediction', but only because the simulation correlates to the probable reality: it's effectively a poll, where the opinions of lots of people are used to model the opinions of lots of people.
(d) Can you really assume that if someone gets it right 10 times, the same person will get it right a further 10 times? It needs to be the same specialism.
(e) You'd need to iron out the randomness by running lots of trials that will be of no use to anyone. Can this operate commercially?
Although I've been marked as 'troll' (boo!), I'm pleased that some relevant points have been raised in reply.;o)
I think we're still waiting for a unified breakthrough in core, memory, code, and data design; I can't say what those breakthroughs are, of course.
I'm not complaining about CISC vs RISC, but that our current memory architecture does not serve our instruction sets well, our layered and dynamically abstracted OO code is awkward to implement without queues of indirection, and our compilers are slaves to all the above.
Try instead charging for software transactions, which is a useful measure of work done by the software. It's a bit like paying for fuel by the gallon, rather than charging different amounts depending on the size of engine in the car you drive.
x86 has been led into a blind alley. Time now for a redesign, to make an instruction set and execution model that doesn't waste 95% of its cycles waiting for memory.
Thanks for the considered reply. TBH, it's usually quite difficult to express every thought that is involved in theses discussions, and it's a minefield of misinterpretation - solved by good writing, of course:o) You might be surprised to see that I agree with everything you wrote.
My main point was that typical 'consumer-level' image processing applications use images that are a regular grid of data points, quite different from the sparse data points that this algorithm is suited to. Typical images contain a lot of noise, and distinguishing the signal from the noise in existing data is the problem that must be overcome, before this technique would be really useful. If an image was randomly reduced, and the algorithm applied, then the noise would contribute to the characteristic of the final image, but perhaps there could be some improvement from the analytical ideal result.
For my understanding of the workings of the algorithm: I do appreciate the 'projection' and 'change of basis' aspects of FT when applied to higher dimensions, and how different power spectra can fit into a set of data points. FT theory tells us that complete time-domain (or space-domain) and complete frequency-domain signals contain the same amount of information, but here we exploit the fact that in the frequency domain, we can use a small amount of data to represent the important perceived aspects of the signal that can span the whole of the time (or space) domain.
In CS, the small amount of space-domain data is used to infer the best minimal frequency-domain spectrum, which is then applied back to the space-domain with appropriate limits.
I see the reverse technique being useful, for an image compression algorithm to choose minimal data points (or their FTs) that best represent the perceived characteristics of an image.
Without the purpose of the original question, we're all left running around like headless chickens.
I can only assume that the unrevealed purpose is kept hidden for a reason. I'll make a punt and suggest that anything legitimate could be done by implementing a file system along with diagnostic tools, or even just with flat files.
My response would be that the target technology is all proprietary, and the whole chain is carefully-engineered to convert the material analogues into digitals. There's even a fair chance that the near-physical stuff abstracts away the bit-level materials, so you'd have to re-engineer right down to the platter. You'd need a very expensive and shiny lab to do that.
From the referenced reports, it looks like people might get the wrong idea about the possible applications. This algorithm starts with discrete data points with gaps in-between, and works out the remaining arbitrary data points in a pleasing way, as if it were a continuous field (represented as a fourier transform, for example).
In other words, it works with data where the signal is already separated from the noise. My last sentence is crucial for an understanding of the possible applications: it will not infer elements that are absent in the measured signal, but will instead repeat elements that are already present. I expect this story will be mis-reported in future, by reporters who do not understand how it really works (and I might count myself in that, as I've only glanced at a couple of the arxiv papers).
Formally-schooled programmers tend to think 'bigger-picture' and code for wider issues, whereas self-taught take pleasure in the engineering, will adopt minimalist solutions and tend to re-invent the wheel in fragments (as opposed to methods from a large library). As such, the self-taught solutions are likely to be more 'magical' and innovative - not necessarily good for deployment, but could be useful for prototyping and proof-of-concept.
The 'Apple way' for media gadgets is that you buy their hardware (no hardware cloning), and buy everything for it through them, so they have part of the revenue pie. It is not in their interests to open up their architecture. As such, the argument is not about choice of functionality, but of customers being wowed into buying the product, and then finding themselves OK (or not) with the exclusive media channels, which limits the functionality of their limited-rights purchases. There's one thing that has not yet been locked down: ordinary software for the computers - is that next?
Several points here...
Clearly establish any conditions that go with accepting the cash. Are they buying any part of ownership, or are they just providing a bonus or reward to help you along? Is there any expectation that your new computer will have 100% up-time, or have a stated life in years?
The common sense approach would be to say that if you have been doing any personal stuff on a company computer, and it has been in your own time, without affecting your work time or your work, then it's OK, and they can only object if what you're doing reduces the life of the computer.
Given that it's a 'personal' computer, not a company computer, then wear and tear would be OK, and we eliminate that objection. They can only object if your personal activities are affecting your work productivity.
Remember, there has to be a reasonable suspicion of wrongdoing for them to capture the computer or data on it. Keep your nose clean, and all should be well.
Best of all, in case your computer needs to be inspected for maintenance (probably only applicable to company assets), or put on a company network, keep any personal stuff separable, in known folders, backed up regularly, and if shared then only securely shared(!).
Agree: it's theft, AND a very inefficient way of generating electricity. We should be moving away from biofuel generation (and this is yet more inefficiency).
Slashdot Mirror
I'll hazard a guess that clock speed excels in one particular case: tight-loop iteration. You can't do that with parallelism (ignoring some fancy pipelining to get part-way there). The fastest way to get the 1 millionth result in a no-shortcut iterative sequence is to get the loop processing at the highest frequency possible.
Try Joules (in context as a total), or watts (as a measure per unit time).
I was about to post something similar to this parent (saved my typing!), so I'll just add a 'diff': $5 won't stop someone trying to make 10,000 times that in a scam.
There's loads of money to be had selling a device has closed content delivery - the iPad is good at that. However, it is not really a device for being creative, which is what a computer excels at, and it's harder for a hardware manufacturer to make money out of that, unless you feed all output into a closed content delivery system...
There needs to be a commercial argument for keeping non-commercial solutions.
I'm not sure what the author wants to make, in real terms. Is it the user base, the third-party integration, the hardware infrastructure? Yes, all of the above, as they're all necessary, but that requires a new build, with different policies. However the viability reduces to the reality of money: money to build it, and money to sustain it. Facebook is now starting to explain the cost of Free.
Here's a question: Can any diffusely-owned project or data be trusted? Does it require that all members of the project or support infrastructure are also trusted, or must there be a certificate-based identity/trust system to unlock the data on various levels?
A subtractive cell (stacked CMYK layers, to filter out R,G,B,All respectively) would let more light through than separate R, G, B windows. The article alludes to using a primary subtractive cell (Y) to help one combination, but it would only be 100% brighter for saturated yellows (not whites); CMYK would be about 250% brighter for all colours (not just yellow) with very good blacks.
Then he would sue the manufacturers of his garden gate for not stopping vandals, and they would counter-sue for not evaluating fit-for-purpose before purchase. The judge would counter-sue for failing to provide information that the gate was not suitable for stopping vandals, ...
Yes, this is all about anonymous postings, but surely anyone can make up an identity online? Law has a habit of applying judgements to other cases (in the same country), and encourages prosecutors to take a punt in other countries. In what other cases would this frustrate the everyday running of the web? ISPs failing to moderate comments from their customers? Allowing file sharing?
This model seems to follow 'genetic programming' principles, but is flawed in many ways: (a) It assumes that most people know everything relevant to the problem under consideration - they often don't.
(b) What the model is looking for is an expert among the crowd. On average, you can find an expert among 1024 people, to predict 10 coin tosses - this is with random data having no relation to specialized wisdom.
(c) Eurovision (mentioned above) is in the rare category of scenarios that can make use of 'crowdsourcing prediction', but only because the simulation correlates to the probable reality: it's effectively a poll, where the opinions of lots of people are used to model the opinions of lots of people.
(d) Can you really assume that if someone gets it right 10 times, the same person will get it right a further 10 times? It needs to be the same specialism.
(e) You'd need to iron out the randomness by running lots of trials that will be of no use to anyone. Can this operate commercially?
cache:http://paul-m-jones.com/?cat=27 into Google search (the original link). With any luck, the old content being referred-to might be there.
Although I've been marked as 'troll' (boo!), I'm pleased that some relevant points have been raised in reply. ;o)
I think we're still waiting for a unified breakthrough in core, memory, code, and data design; I can't say what those breakthroughs are, of course.
I'm not complaining about CISC vs RISC, but that our current memory architecture does not serve our instruction sets well, our layered and dynamically abstracted OO code is awkward to implement without queues of indirection, and our compilers are slaves to all the above.
</untroll>
Try instead charging for software transactions, which is a useful measure of work done by the software. It's a bit like paying for fuel by the gallon, rather than charging different amounts depending on the size of engine in the car you drive.
x86 has been led into a blind alley. Time now for a redesign, to make an instruction set and execution model that doesn't waste 95% of its cycles waiting for memory.
Thanks for the considered reply. TBH, it's usually quite difficult to express every thought that is involved in theses discussions, and it's a minefield of misinterpretation - solved by good writing, of course :o) You might be surprised to see that I agree with everything you wrote.
My main point was that typical 'consumer-level' image processing applications use images that are a regular grid of data points, quite different from the sparse data points that this algorithm is suited to. Typical images contain a lot of noise, and distinguishing the signal from the noise in existing data is the problem that must be overcome, before this technique would be really useful. If an image was randomly reduced, and the algorithm applied, then the noise would contribute to the characteristic of the final image, but perhaps there could be some improvement from the analytical ideal result.
For my understanding of the workings of the algorithm: I do appreciate the 'projection' and 'change of basis' aspects of FT when applied to higher dimensions, and how different power spectra can fit into a set of data points. FT theory tells us that complete time-domain (or space-domain) and complete frequency-domain signals contain the same amount of information, but here we exploit the fact that in the frequency domain, we can use a small amount of data to represent the important perceived aspects of the signal that can span the whole of the time (or space) domain.
In CS, the small amount of space-domain data is used to infer the best minimal frequency-domain spectrum, which is then applied back to the space-domain with appropriate limits.
I see the reverse technique being useful, for an image compression algorithm to choose minimal data points (or their FTs) that best represent the perceived characteristics of an image.
Without the purpose of the original question, we're all left running around like headless chickens.
I can only assume that the unrevealed purpose is kept hidden for a reason. I'll make a punt and suggest that anything legitimate could be done by implementing a file system along with diagnostic tools, or even just with flat files.
My response would be that the target technology is all proprietary, and the whole chain is carefully-engineered to convert the material analogues into digitals. There's even a fair chance that the near-physical stuff abstracts away the bit-level materials, so you'd have to re-engineer right down to the platter. You'd need a very expensive and shiny lab to do that.
From the referenced reports, it looks like people might get the wrong idea about the possible applications. This algorithm starts with discrete data points with gaps in-between, and works out the remaining arbitrary data points in a pleasing way, as if it were a continuous field (represented as a fourier transform, for example).
In other words, it works with data where the signal is already separated from the noise. My last sentence is crucial for an understanding of the possible applications: it will not infer elements that are absent in the measured signal, but will instead repeat elements that are already present. I expect this story will be mis-reported in future, by reporters who do not understand how it really works (and I might count myself in that, as I've only glanced at a couple of the arxiv papers).
Formally-schooled programmers tend to think 'bigger-picture' and code for wider issues, whereas self-taught take pleasure in the engineering, will adopt minimalist solutions and tend to re-invent the wheel in fragments (as opposed to methods from a large library). As such, the self-taught solutions are likely to be more 'magical' and innovative - not necessarily good for deployment, but could be useful for prototyping and proof-of-concept.
The 'Apple way' for media gadgets is that you buy their hardware (no hardware cloning), and buy everything for it through them, so they have part of the revenue pie. It is not in their interests to open up their architecture. As such, the argument is not about choice of functionality, but of customers being wowed into buying the product, and then finding themselves OK (or not) with the exclusive media channels, which limits the functionality of their limited-rights purchases. There's one thing that has not yet been locked down: ordinary software for the computers - is that next?
Several points here... Clearly establish any conditions that go with accepting the cash. Are they buying any part of ownership, or are they just providing a bonus or reward to help you along? Is there any expectation that your new computer will have 100% up-time, or have a stated life in years? The common sense approach would be to say that if you have been doing any personal stuff on a company computer, and it has been in your own time, without affecting your work time or your work, then it's OK, and they can only object if what you're doing reduces the life of the computer. Given that it's a 'personal' computer, not a company computer, then wear and tear would be OK, and we eliminate that objection. They can only object if your personal activities are affecting your work productivity. Remember, there has to be a reasonable suspicion of wrongdoing for them to capture the computer or data on it. Keep your nose clean, and all should be well. Best of all, in case your computer needs to be inspected for maintenance (probably only applicable to company assets), or put on a company network, keep any personal stuff separable, in known folders, backed up regularly, and if shared then only securely shared(!).
Most code gets assembled into GOTOs (jumps) anyway. I'm sure some competent programmers can manage their own GOTOs just like a compiler can.
Imagine the chaos that a cat could cause if one-click ordering is enabled.
Put a dymnamo on a pair of lifts on a multi-storey car park, for the cars GOING DOWN :o)
Agree: it's theft, AND a very inefficient way of generating electricity. We should be moving away from biofuel generation (and this is yet more inefficiency).