In a word, yes. I have a WinRadio 1550e, which allows monitoring within the waterhole (~1.4GHz) which is where most amateur seti astronomers look.
I have a dish, which had to get signoff from the secretary of state before I could install it:-) The picture shows the width of the house, with the dish being approx 4m across...
Making an interferometer poses major problems with time resolution though - to merge all these amateur radio telescopes together would (a) take a huge chunk of bandwidth for each telescope (ADSL ain't enough...), (b) need excellent synchronisation between the telescopes, which almost all of us don't have, and (c) need the dishes to be steerable, which most of them aren't...
There is however a project argus doing the same thing with lots of individual telescopes. As soon as I'm happy with the s/w running on mine, I'll be a member of the group:-)
"The US, between immigration and a rebounding birth-rate, will outstrip the population of Europe before 2050, if current trends hold."
You have got to be fucking joking. And for once I'm using the word "fucking" in the correct context.
Europe's population is approx 730 million people. The USA is approx 250 million. If you're telling me that the USA is going to TRIPLE in population in 50 years, then you're fucking mental! (ok, a bit of a stretch)
Um, unless you're using geolocation to detect those sites, using things like looking for '.com' and '.net' is highly unreliable. Almost every European company will try to get the.com as well as the local country code domain, and.org/.net as well for that matter...
Perhaps the super-computing problems are approached in the way they are, because of the limitations on bandwidth to the CPU(s).
Most of the super-computing problems are simulations, and I would have thought that being able to simulate more of the environemnt (therefore, more data to crunch) would be an advantage.
It's called an interferometer. There are major advantages to it in terms of resolution, but it still doesn't compare to a large dish in terms of sensitivity (most of the radio waves hit the ground, between the dishes)
The brochure doesn't actually say how they're going to do it - it puts forward some suggestions but none of these seem definite yet. A single one million square-kilometre dish would be quite something!
The difference between a single dish and a synthetic aperture of two dishes (as far as I understand it anyway) is that increasing the baseline to get the synthetic aperture will increase the resolving power of the telescope, but since you've only doubled your collecting area's size, it won't increase the amount of signal that you get. So yes, higher resolution, much the same sensitivity.
There is also a critical dimension to an interferometer (ie: a synthetic aperture radio telescope) which determines what such a telescope can look at. The angular presentation of the object you're studying must be small in comparison to twice (IIRC) the distance between the two dishes you're using for the interferometer, because you're using the phase difference in the signal arriving at both dishes to infer the resolution you gain. You can't look at objects which extend beyond this limit because the phase starts to overlap. Or so I think, anyway. You might want to ask your ex:-)
One other point is that you can do the same interferometry trick between any two nodes in the array. In fact you can do it to (all nodes)factorial and get as much data as possible. The superposition of all of this would take a fair amount of CPU though...
The reason radio telescopes have to be so much larger than their optical counterparts is due to the wavelengths they are looking for. For a given observation aperture, there's a simple rule-of-thumb which goes:
Voltage gain ~= circumference / wavelength.
... with the power gain (the "magnification") being the voltage gain squared.
Given that the wavelength of 'visible' light is approximately half a million times shorter than radio wave wavelengths, the collecting area has to be much larger to get the same antennae gain.
An interesting corollary of this is that the naked eye is (very roughly) as powerful (at visible light wavelengths) as Arecibo is (at radio wavelengths). See the The seti league pages for more info...
Yes. "The planet" in question being Earth. If a nuclear-powered device explodes on launch, or in low orbit, it's "not a good thing". At the very least you'll get radioactive debris spread over a wide area.
You typically composite and re-composite layer after layer to create decent effects, it's not a one-shot thing. Certainly professional video runs at ~48bit for film work.
I think the internet should remain global. Absolutely and unequivocably.
But to do the subject some justice:-) With the US becoming more and more isolationist over time, it's hardly surprising others are reacting in the same way. There are *more* people in the EU than the US. There are ~1/5 the population of the US in the UK! Why should't they demand more representation ?
The US legal system (which is where a *lot* of the problems are coming from) is very much a big-business-friendly institution; since most of the congressmen are funded by big business as well, it's hardly surprising that the internet is being mauled with the same fangs that savage the "common person" in the US. There is also much more of a "who do I sue" attitude within the US than just about anywhere else.
Still, it's clearly a nonsense to advocate separation, and it's not clear to me that other countries are overall any better. The term "swings and roundabouts" comes to mind.
Moore's law is NOT a law, at best it's an observation that has so far been consistent.
The cell is a highly parallel chip, it is outside the bounds of Moore's "law" because it doesn't follow the same design methodology. If I designed an FPGA today that had 1000 FPU's, and a simple CPU to control them, I could easily best a P4 in FLOPS. Trivial. Sony has done/will do in hardware what I have suggested, and given that they've been working on it for a couple of years, I think there may be more than just a couple of extra FPU's.
Not by end users. I suspect that the equipment to reprogram them costs more than an xbox
Actually FPGA's are normally programmed using EPROM's. Most FPGA's these days are actually static RAM cells, which are programmed at power up by reading data from an EPROM. EPROM burners are pretty cheap...
But in any case, FPGA hardware is ridiculously cheap. Go to fpgacpu.com and see for yourself - a 300,000-gate FPGA environment complete with programmer and s/w for ~$170 US. If you want a cheaper one, you can get 150k-gate ones for ~$120 US. Considering that a 32-bit CPU is ~20k-gates, that's pretty good:-)
Sort of, yes, but in a store-and-forward manner rather than a single-point => multiple destinations. At least as I understand TCP multicast, there are no relays (where chunks of the data can be stored for later, independent access). Perhaps I should read up on the MBONE again...
One thing I've not seen posited in the P2P arena is a technique analogous to the spread-spectrum military comms. Howabout splitting the [bigbadfile] into smaller binary chunks based on frequency of access. Sites host [bi], [igb], [gbad], [dfi], [ile].
Assume the chunks can be split arbitrarily, and the P2P s/w can do binary patches of overlapping data, surely this would prevent anyone claiming you were serving anything they had an interest in. A chunk of binary data from one site would be pretty much akin to a chunk of binary data from another. It's only in the entireity that copyright is held anyway, no? I can't believe someone has a copyright on middle-C...
Additional benefits come from the robustness of the distributed data. If you use a one-way hash of the text-to-search-on to the item-description and index by hash, then even the hosts wouldn't know what it was they were hosting, and splitting the files up into small chunks would abrogate their responsibility further.
Indexing it becomes more of a problem, but this is simply the same problem as we currently have, with the addition of a byte-range. The client simply has to fetch sufficient data to create the entire range.
You'd want hosts to merge chunks that were adjacent for efficiency, to within a host-defined fraction of the total size - you'd not want to be suddenly hosting an entire object or you're open to DOS'ing.
Use gzip/ssh to tunnel all comms (data and metadata) so it's hard to intercept. Add XML-RPC or similar so it can be tunnelled through a firewall webserver port. If you're really paranoid, add a steganographic mode where requests and data are embedded in mpeg/jpeg files in the lsb. You'd have to have rewritten compressors for steganography anyway, so preserving the high-frequency bits for data would be possible.
I've thought about doing something like this before. The fundamental flaw is that an open-source client can be modified to inject rubbish into the network. You can compensate for this with a slashdot-style moderation system. Operating on the principle that there are more IP addresses for us than for them, and letting every IP address have a '+' and a '-' vote per day per hosting server, bad files could be marked as such efficiently by having a 'Click here if it doesn't work' in the client GUI. If your IP's moderation is inconsistent with the majority then your moderation rights for that server are suspended for a while. I think that would work.
I don't know about you, but the instant someone gives me a percentage increase figure, I'm instantly curious why they don't give me an absolute figure...
People's understanding of percentages is really poor:
A 131% increase in sales results in 231% of the previous amount A 31% increase in sales results in 131% of the previous amount Sales of 131% of the previous amount is a 31% increase.
This seems to confuse just about everyone. Not to mention the magic (greater than 100%) figure, and the fact that 131% of 10 is ~13 (so 10 => 23), while 1% of 10,000 is 100 (so 10,000 => 10,100). Would you rather have the 1% figure or the 131% figure ?
Cynicism *is* in my nature (hey, I'm British:-) but I *really* prefer to see numbers than percentages.
Microsoft don't think so - they're (Balmer) admitting that Linux TCO is less than windows, and now they need better arguments than "We're cheaper".
Plus, big government contracts are only sometimes for general-purpose bloke-in-a-box solutions, and hell, the UK will just go for MS at the moment with those ones. Fair enough.
The really lucrative contracts are the defence and systems contracts though. There's a lot of space there for Linux to make its' own....
I am, in fact:-) Probably on www.gornall.net when I get it all together. Nothing there yet, but the PS2 site is slowly coming together (ps2.gornall.net).
I'm planning on using a PC to do the data analysis (and my background is in signal processing, so I'm quite looking forward to it), but another idea is to use a PS2 to do the work - my PC is worth ~1.5 GFlops (Athlon 1700), but the PS2 can do ~6 GFlops, and the vector units are native-floating point devices. At that rate I could look into doing chirp analysis (compensating for time-distortion due to the motion of source and the earth) as well as plain old fourier analysis.
It's still reasonably expensive to get hold of all the equipment though (I know, I'm in the final stages of doing it - the antenna arrives next week:-)
The dish (3m diameter minimum) will set you back ~150 quid, the radio receiver (1.4GHz typically) is ~300 quid for the WinRadio 1550e (PC-attachable high-frequency radio receiver), and the LNA (low-noise amplifier), feed-horn, cable and fittings came to another 300 quid. So a total of ~750 quid.
All of this is in the UK, and apart from the dish, I bought new kit. I'm sure that (a) outside the UK it's easier to get large dishes, and (b) if you bought 2nd hand, you'd be able to get the price of the hardware down. You can even make your own components if you have the knowledge.
A good source of info is the setileague website for the mini-manual or the UK site run by Jenny Bailey, although it was a bit out of date last time I looked.
Anyway, this time next week, I should be searching for aliens from my back-yard too:-)
The PS2's geometry engine is a 4x4 arrangement (16 pixel pipelines in total), so the fastest possible render is irrespective of 4x4 or 0x0. Given that a 4x4 triangle at least possesses triangular nature, I'm surprised that they would go for 0x0.
The PS2 also doesn't use co-ordinate space of 0,0 to be anything special - the hardware has automatic scaling from an abstract 4096x4096 space into whatever resolution you happen to be working in. Typically (at 640x480), the co-ordinate (0,0) is at (1728,1808). Why then does it matter where you render the triangle ?
Whether you draw the buffer to the screen or not is also not relevant - you can't "draw" to local memory - it's drawn to RAM with a 2048-bit bus on the same chip as the video processor. There's no reason why displaying the screen would slow that down, so why open yourself to criticism if that were the case. Odd behaviour to say the least...
In short, I think you're wrong.
There are detailed figures for different types of draw operation in the GS users manual. The 75 million/sec is for no-texture, flat-shade, no-anti-alias. It drops down to 16 million/sec for textured, gourard shaded, fogged, anti-aliased triangles.
I've had my PS2 (running linux of course) on since I bought it about 2 months ago. Works great.
I think you'll find the reason it's not yet popular to leave the PS2 on is that there's no reason to. As soon as there is a reason to, it will happen. In my case, it's so I can ssh into it from work:-)
Slashdot Mirror
In a word, yes. I have a WinRadio 1550e, which allows monitoring within the waterhole (~1.4GHz) which is where most amateur seti astronomers look.
:-) The picture shows the width of the house, with the dish being approx 4m across...
:-)
:-)
I have a dish, which had to get signoff from the secretary of state before I could install it
Making an interferometer poses major problems with time resolution though - to merge all these amateur radio telescopes together would (a) take a huge chunk of bandwidth for each telescope (ADSL ain't enough...), (b) need excellent synchronisation between the telescopes, which almost all of us don't have, and (c) need the dishes to be steerable, which most of them aren't...
There is however a project argus doing the same thing with lots of individual telescopes. As soon as I'm happy with the s/w running on mine, I'll be a member of the group
And no, no aliens yet
"The US, between immigration and a rebounding birth-rate, will outstrip the population of Europe before 2050, if current trends hold."
You have got to be fucking joking. And for once I'm using the word "fucking" in the correct context.
Europe's population is approx 730 million people. The USA is approx 250 million. If you're telling me that the USA is going to TRIPLE in population in 50 years, then you're fucking mental! (ok, a bit of a stretch)
Simon
USA is about that, but the EU is about 750 million, not 350 million.
Simon
Is this meant to be a joke? Maybe I'm missing the humour, or perhaps it's just incredibly arrogant.
Perhaps we know about NAT as well....
Simon
Um, unless you're using geolocation to detect those sites, using things like looking for '.com' and '.net' is highly unreliable. Almost every European company will try to get the .com as well as the local country code domain, and .org/.net as well for that matter...
Simon
Perhaps the super-computing problems are approached in the way they are, because of the limitations on bandwidth to the CPU(s).
Most of the super-computing problems are simulations, and I would have thought that being able to simulate more of the environemnt (therefore, more data to crunch) would be an advantage.
Simon
Out of all the posts here, that one made me laugh out loud:-))
Simon.
It's called an interferometer. There are major advantages to it in terms of resolution, but it still doesn't compare to a large dish in terms of sensitivity (most of the radio waves hit the ground, between the dishes)
Simon
The brochure doesn't actually say how they're going to do it - it puts forward some suggestions but none of these seem definite yet. A single one million square-kilometre dish would be quite something!
:-)
The difference between a single dish and a synthetic aperture of two dishes (as far as I understand it anyway) is that increasing the baseline to get the synthetic aperture will increase the resolving power of the telescope, but since you've only doubled your collecting area's size, it won't increase the amount of signal that you get. So yes, higher resolution, much the same sensitivity.
There is also a critical dimension to an interferometer (ie: a synthetic aperture radio telescope) which determines what such a telescope can look at. The angular presentation of the object you're studying must be small in comparison to twice (IIRC) the distance between the two dishes you're using for the interferometer, because you're using the phase difference in the signal arriving at both dishes to infer the resolution you gain. You can't look at objects which extend beyond this limit because the phase starts to overlap. Or so I think, anyway. You might want to ask your ex
One other point is that you can do the same interferometry trick between any two nodes in the array. In fact you can do it to (all nodes)factorial and get as much data as possible. The superposition of all of this would take a fair amount of CPU though...
Simon
I'd like to see you get the material for a square kilometre of telescope into space. Maybe once we actually get that space elevator, but not until.
:-) does matter
It's important for a radio telescope to be large (see my other post). Size (at least here
Simon
Given that the wavelength of 'visible' light is approximately half a million times shorter than radio wave wavelengths, the collecting area has to be much larger to get the same antennae gain.
An interesting corollary of this is that the naked eye is (very roughly) as powerful (at visible light wavelengths) as Arecibo is (at radio wavelengths). See the The seti league pages for more info...
Simon.
Yes. "The planet" in question being Earth. If a nuclear-powered device explodes on launch, or in low orbit, it's "not a good thing". At the very least you'll get radioactive debris spread over a wide area.
Simon.
The BBC are reporting that they've found the probe orbiting the sun... No comets then ...
Simon
Once, definitely. Twice, probably. Thrice, perhaps.
You typically composite and re-composite layer after layer to create decent effects, it's not a one-shot thing. Certainly professional video runs at ~48bit for film work.
Simon
I think the internet should remain global. Absolutely and unequivocably.
:-) With the US becoming more and more isolationist over time, it's hardly surprising others are reacting in the same way. There are *more* people in the EU than the US. There are ~1/5 the population of the US in the UK! Why should't they demand more representation ?
But to do the subject some justice
The US legal system (which is where a *lot* of the problems are coming from) is very much a big-business-friendly institution; since most of the congressmen are funded by big business as well, it's hardly surprising that the internet is being mauled with the same fangs that savage the "common person" in the US. There is also much more of a "who do I sue" attitude within the US than just about anywhere else.
Still, it's clearly a nonsense to advocate separation, and it's not clear to me that other countries are overall any better. The term "swings and roundabouts" comes to mind.
Simon.
The cell is a highly parallel chip, it is outside the bounds of Moore's "law" because it doesn't follow the same design methodology. If I designed an FPGA today that had 1000 FPU's, and a simple CPU to control them, I could easily best a P4 in FLOPS. Trivial. Sony has done/will do in hardware what I have suggested, and given that they've been working on it for a couple of years, I think there may be more than just a couple of extra FPU's.
All it takes is a little thought....
Simon
Actually FPGA's are normally programmed using EPROM's. Most FPGA's these days are actually static RAM cells, which are programmed at power up by reading data from an EPROM. EPROM burners are pretty cheap...
But in any case, FPGA hardware is ridiculously cheap. Go to fpgacpu.com and see for yourself - a 300,000-gate FPGA environment complete with programmer and s/w for ~$170 US. If you want a cheaper one, you can get 150k-gate ones for ~$120 US. Considering that a 32-bit CPU is ~20k-gates, that's pretty good :-)
Simon
Sort of, yes, but in a store-and-forward manner rather than a single-point => multiple destinations. At least as I understand TCP multicast, there are no relays (where chunks of the data can be stored for later, independent access). Perhaps I should read up on the MBONE again...
Unless you mean multicasting in a non-TCP sense ?
Simon
One thing I've not seen posited in the P2P arena is a technique analogous to the spread-spectrum military comms. Howabout splitting the [bigbadfile] into smaller binary chunks based on frequency of access. Sites host [bi], [igb], [gbad], [dfi], [ile].
Assume the chunks can be split arbitrarily, and the P2P s/w can do binary patches of overlapping data, surely this would prevent anyone claiming you were serving anything they had an interest in. A chunk of binary data from one site would be pretty much akin to a chunk of binary data from another. It's only in the entireity that copyright is held anyway, no? I can't believe someone has a copyright on middle-C...
Additional benefits come from the robustness of the distributed data. If you use a one-way hash of the text-to-search-on to the item-description and index by hash, then even the hosts wouldn't know what it was they were hosting, and splitting the files up into small chunks would abrogate their responsibility further.
Indexing it becomes more of a problem, but this is simply the same problem as we currently have, with the addition of a byte-range. The client simply has to fetch sufficient data to create the entire range.
You'd want hosts to merge chunks that were adjacent for efficiency, to within a host-defined fraction of the total size - you'd not want to be suddenly hosting an entire object or you're open to DOS'ing.
Use gzip/ssh to tunnel all comms (data and metadata) so it's hard to intercept. Add XML-RPC or similar so it can be tunnelled through a firewall webserver port. If you're really paranoid, add a steganographic mode where requests and data are embedded in mpeg/jpeg files in the lsb. You'd have to have rewritten compressors for steganography anyway, so preserving the high-frequency bits for data would be possible.
I've thought about doing something like this before. The fundamental flaw is that an open-source client can be modified to inject rubbish into the network. You can compensate for this with a slashdot-style moderation system. Operating on the principle that there are more IP addresses for us than for them, and letting every IP address have a '+' and a '-' vote per day per hosting server, bad files could be marked as such efficiently by having a 'Click here if it doesn't work' in the client GUI. If your IP's moderation is inconsistent with the majority then your moderation rights for that server are suspended for a while. I think that would work.
Just ideas...
Simon.
I don't know about you, but the instant someone gives me a percentage increase figure, I'm instantly curious why they don't give me an absolute figure...
:-) but I *really* prefer to see numbers than percentages.
People's understanding of percentages is really poor:
A 131% increase in sales results in 231% of the previous amount
A 31% increase in sales results in 131% of the previous amount
Sales of 131% of the previous amount is a 31% increase.
This seems to confuse just about everyone. Not to mention the magic (greater than 100%) figure, and the fact that 131% of 10 is ~13 (so 10 => 23), while 1% of 10,000 is 100 (so 10,000 => 10,100). Would you rather have the 1% figure or the 131% figure ?
Cynicism *is* in my nature (hey, I'm British
Simon.
Microsoft don't think so - they're (Balmer) admitting that Linux TCO is less than windows, and now they need better arguments than "We're cheaper".
Plus, big government contracts are only sometimes for general-purpose bloke-in-a-box solutions, and hell, the UK will just go for MS at the moment with those ones. Fair enough.
The really lucrative contracts are the defence and systems contracts though. There's a lot of space there for Linux to make its' own....
Simon.
I am, in fact :-) Probably on www.gornall.net when I get it all together. Nothing there yet, but the PS2 site is slowly coming together (ps2.gornall.net).
I'm planning on using a PC to do the data analysis (and my background is in signal processing, so I'm quite looking forward to it), but another idea is to use a PS2 to do the work - my PC is worth ~1.5 GFlops (Athlon 1700), but the PS2 can do ~6 GFlops, and the vector units are native-floating point devices. At that rate I could look into doing chirp analysis (compensating for time-distortion due to the motion of source and the earth) as well as plain old fourier analysis.
ATB,
Simon.
The dish (3m diameter minimum) will set you back ~150 quid, the radio receiver (1.4GHz typically) is ~300 quid for the WinRadio 1550e (PC-attachable high-frequency radio receiver), and the LNA (low-noise amplifier), feed-horn, cable and fittings came to another 300 quid. So a total of ~750 quid.
All of this is in the UK, and apart from the dish, I bought new kit. I'm sure that (a) outside the UK it's easier to get large dishes, and (b) if you bought 2nd hand, you'd be able to get the price of the hardware down. You can even make your own components if you have the knowledge.
A good source of info is the setileague website for the mini-manual or the UK site run by Jenny Bailey, although it was a bit out of date last time I looked.
Anyway, this time next week, I should be searching for aliens from my back-yard too :-)
Simon.
Hmm. That's very odd.
The PS2's geometry engine is a 4x4 arrangement (16 pixel pipelines in total), so the fastest possible render is irrespective of 4x4 or 0x0. Given that a 4x4 triangle at least possesses triangular nature, I'm surprised that they would go for 0x0.
The PS2 also doesn't use co-ordinate space of 0,0 to be anything special - the hardware has automatic scaling from an abstract 4096x4096 space into whatever resolution you happen to be working in. Typically (at 640x480), the co-ordinate (0,0) is at (1728,1808). Why then does it matter where you render the triangle ?
Whether you draw the buffer to the screen or not is also not relevant - you can't "draw" to local memory - it's drawn to RAM with a 2048-bit bus on the same chip as the video processor. There's no reason why displaying the screen would slow that down, so why open yourself to criticism if that were the case. Odd behaviour to say the least...
In short, I think you're wrong.
There are detailed figures for different types of draw operation in the GS users manual. The 75 million/sec is for no-texture, flat-shade, no-anti-alias. It drops down to 16 million/sec for textured, gourard shaded, fogged, anti-aliased triangles.
Simon
"Who leaves their playstaion on all the time?"
:-)
I do.
I've had my PS2 (running linux of course) on since I bought it about 2 months ago. Works great.
I think you'll find the reason it's not yet popular to leave the PS2 on is that there's no reason to. As soon as there is a reason to, it will happen. In my case, it's so I can ssh into it from work
Simon