It wouldn't have to be quite so bad. One possibility is to sense where the viewer is and only send those pixels. Another is to send enough 2D layers such that the 3D scene can be recomputed properly for a wide variety of viewing positions (re: image-based rendering). That is still assuming stereoscopic display, though.
Your idea of using 1920x1080x1080 only permits you to look into a TV-size cube of space. What is more desirable is if the TV represents a "window" into another world of infinite dimension. This calls for a lightfield display, where each pixel sends out different light rays (of varying color & intensity) in different directions. Essentially, each pixel delivers a whole image, directionally dispersed. The resolution of each "pixel image" depends upon the range of viewing angles you wish to support, as well as the fidelity of angular resolution you want, of course. But if you assume an HD "image" for each pixel in an HD image, then you're talking 4 trillion total "pixels" (light rays), 60 times per second.
The drive controller has a clock that it uses to control the width (or length, actually) of the bits that it writes. However, due to temperature variation (quartz oscillators are temperature sensitive) and other factors, the clock that is used when the bits are read may be slightly different than the clock that was used when the bits were written. A PLL (phase lock loop) can keep the clocks aligned, but it needs a certain minimum number of transitions per unit time to take place to keep them aligned.
In truth, reading & writing bits is actually more complicated than that, since drive manufacturers have pushed the technology past the point of simply reading distinct 1's & 0's from the platter. The signal is analog, and when you try to squeeze the bits closer & closer together, it no longer resembles just "high" & "low" levels (or positive and negative, given that it's sensing the alignment of a magnetic field). Instead, you get a signal that looks more like noise, and it requires some fancy signal processing to figure out from what is read what the original pattern written was. The technique is called "PRML" or partial response, maximum likelihood.
To understand sync, you need to understand that, at a certain level, the signal being read from the drive head is a stream of 1's & 0's. To make sense of this stream, you need to know the boundaries of things. At one level, you need to know where each byte begins and ends. At a higher level, you need to know where each sector begins and ends. That's what sync is all about. To "do" sync, you use a sequence of 1's & 0's that you'd never use for actual data. When you see that sequence, you know, "ah, this is the start of something". Once you see the start, then you'll know where you are as long as you keep track of things and certain rules are followed. For instance, one rule is that you can't have too many 1's or 0's consecutively, or else you'll lose your place, since you can't actually see the "start" of a 1 or 0, but you can only see the changes from a 1 to a 0 or vice-versa.
To further understand the need for sectors, you must also think about how you write data to the disk. First, the disk has to read the data off the disk while the platter is spinning in order to find the start of the right sector. Then it has to switch the head into write mode and write out the new data. It must stop writing data at the end of this sector in order to not overwrite the sync header for the next sector. Due to several sources of variations (platter spin speed, temperature, mechanical play, clock frequency, etc.), there needs to be a bit of "dead zone" (ie, inter-sector gap) between sectors to make sure that the sync headers are never overwritten.
By going to larger sectors, you have fewer of them, and therefore need fewer inter-sector gaps.
Computer people who've been around may remember that the Commodore Amiga could write 880KB on floppy disks when PCs could only write 720KB. This was because the Amiga didn't use multiple sectors per track, but rather always just read & wrote an entire track's worth of data for each access. Similar idea going on here.
To become filthy rich, you need a few things: 1) The vision to first see what's available now, and second see the market opportunity in the future. 2) Some connections to smart people who can take you to that second step. Connections to money help, too. 3) The motivation to get there & keep going, without letting anyone slow you down. A lack of certain ethics may help with #3.
Lots of people may have one or two of these. Not so many have all three.
I don't think that patent holders should be allowed to escape licensing by claiming that they will bring a product to market themselves. Perhaps they can be given a time window in which to do so. If they are not satisfied with this, then they always have the option of not patenting.
As I wrote in another article, patents & copyrights should serve 2 purposes: -encourage innovation & creativity by providing a limited period to profit from it. -add to the public domain after said period expires.
The second part of the deal is generally being forgotten about in this day of "I do what's best for me" (vs. I do what's best for all).
> Also, you are proposing regulated pricing, AKA socialism, and bucking the free market.
No, not at all. The formula could very well be market-based. In fact, you may have hit upon the solution: a market (like the stock market) for patent licenses. Of course, this brings up a host of new issues to solve, but then, what's life for?
> achieve the same thing like by implementing features that are not supported in other platforms.
I think you hit upon it exactly. There's usually loopholes around legal issues. It works in MS's favor to have everyone embrace their technology; then, they can make their own platform special by extending the technology in ways that others cannot.
I think patent law should be modified to disallow patent trolling. Patent holders should be required to license patents for reasonable fees, or else lose the patents. Of course, the key issue is determining what a "reasonable" fee should be, but I imagine that someone smart enough can come up with a reasonable formula.
I imagine that the original purpose of both copyrights & patents was two-fold: 1. Encourage creativity & innovation by providing a limited time period to profit on new works & ideas. 2. Grow the public domain of ideas after said limited time expires.
These days, however, that second part has been forgotten, because the word "profit" only appears in the first part, and big companies will do all that they can to keep profiting from something.
I tend to visualize commerce as a flowing river or stream. You make something people want to buy, and money flows in your direction when you sell. Big companies want to maximize profits by finding existing streams, claiming ownership over them, and redirecting the flow into themselves. Then they do everything that they can to prevent that stream from drying up or otherwise escaping their control.
In essence, assuming the two purposes stated above are true, big companies are stealing from the public by claiming longer & longer holds on creative works & inventions. They have the laws changed to favor themselves, and Joe Public doesn't fight back.
The main accomplishment of Fastra II is that they put 7 GPU cards into the machine. Fastra I already had 4 GPU cards, which is what the FOH machines top out at. The GPU cards are the same cards all around (Nvidia GTX 295). The linked article & video point out the difficulties they faced & overcame in trying to make 7 GPU cards work simultaneously in the same box.
Oh, and by the way, I'm wondering quite the opposite: why do we still see so many over-sized full ATX size cases being offered, when microATX motherboards have everything we (most of us) need? Indeed, even mini-ITX motherboards are often adequate for so many needs, and yet mini-ITX cases still seem to command a premium because they are relatively rare. It's easy (and boring) to design a big rectangular ATX box. It's an engineering challenge to make a good-looking small box that does everything you need and is still practical to work with.
It's known as "market forces". In case you haven't noticed, the computing needs of most people can be crammed into something the size of a paperback book or so. Larger computing devices are available, but the bigger you go, the smaller the market, and thus the larger the price. If you want something big, you might take look at a computer named "Jaguar". It has a big price, too.
As far as personal computers go, they tend to be designed around CPU strengths & limitations. Intel and AMD have figured out that the most efficient way to increase computing power is to put more and more processing power into a single chip, and have systems designed around a single CPU chip, as opposed to systems that put multiple CPU chips on the motherboard. Because of this approach, it became unnecessary to build systems larger than your typical ATX desktop.
If you needed more computing power than that, your best bet was to get multiple machines. Indeed, you can fill refrigerator-sized racks full of ATX (or other form factor) motherboards. For instance, check out: http://www.cse.illinois.edu/turing/Images/FrontView.html
Only recently have GPUs become recognized as an efficient way of adding lots of computing power to a desktop machine. As evidenced by the motherboard that made Fastra II possible, hardware is slowly becoming available to embrace this new computing paradigm. Perhaps in a few years, you'll get your 12-double-wide-slot motherboard and you'll be able to populate it with GeForce 28000 boards. But more than likely, it still won't be cheap, since few people seem to need this kind of performance.
There were several difficulties. The most obvious is that they fit 7 double-wide cards into 7 single-wide slots. The next was that the motherboard BIOS crashes when more than 5? boards are installed. The next was that in order to allocate enough I/O space, all unnecessary devices had to be disabled, and even still the Linux kernel needed to be hacked to reduce the space allocated to various resources. After all that, it was a piece of cake.
Apparently, the regular BIOS can't boot with more than 5? graphics cards installed due to the amount of resources (memory & I/O space) that each one requires. So the researchers asked ASUS to make a special BIOS for them which doesn't set up the graphics card resources. However, the BIOS still needs to initialize at least one video card, so they agreed that the boot video card would be the one with only a single GPU. Presumably, they could have also chosen a dual GPU card that happened to be different from the others in some way.
It's not silly: (1) this is a research project, not production medical equipment, meaning that the funds to buy Tesla cards were probably not available, and they aren't particularly worried about occasional bit errors. (2) Their particular application doesn't need much inter-GPU communication, if any, so that bandwidth is not an issue. They just need for each GPU to load datasets, chew on them, and spit out the results.
How much does your proposed GPU supercomputer cost for 13 GPUs?
"After a one-month checkout period, WISE will spend six months mapping the whole sky. It will then begin a second scan to uncover even more objects and to look for any changes in the sky that might have occurred since the first survey, according to NASA. This second partial sky survey will end about three months later when the spacecraft's frozen-hydrogen cryogen runs out."
You are focusing on the specifics of this particular case. I have made no comments about whether Apple's actions are "good" or "bad," but rather that the whole situation is kind of ridiculous.
I'm trying to see beyond this particular case and envision what set of conditions would allow for companies in general to focus on "progress" (whatever that means) instead of in-fighting.
This particular case revolves around patents. These days, patents are kind of like weapons. Companies try to generate or otherwise acquire as many patents as possible as kind of an arms race. Once a company has achieved a critical mass of patents, they generally try to work out non-aggression pacts (cross-licensing agreements) with their competitors to avoid patent wars. But the landscape is full of pirates (in the "Aaarh!" sense) who will gather up some patents to go on offensives just to score some quick booty.
Anyone who's given the matter some thought will realize that the present patent system is broken and needs revamping. However, companies are kind of like drug addicts with regard to patents. They can see the damage that they can do, but they really don't want to give them up.
Regarding patents, I think it's time for rehab. However, patents are just one of the many ways that corporate culture is messed up. Trying to fix everything would be a much bigger job than trying to overhaul the US's health-care system. And, just like that problem, if the majority of people don't see that there's anything wrong, nothing will ever change.
I think you're kind of missing the point. We shouldn't be behaving like cave-men in the first place.
True, the corporate environment that all companies currently exist in tends to encourage this behavior. But the nature of this corporate environment is hurting everyone (except the lawyers), so how long will it be before we can all evolve and pursue better things?
Reminds me of an anecdote concerning the game Mule. I played this game with my brothers, and we always competed against each other. When I mentioned the game to a friend of mine, she said "That's a great game. What was the richest your colony ever got?" I responded with some score I remembered at the time. She said something to the effect of "Why so low? We usually got 10x that!" It had not even occurred to me that one might try to play the game cooperatively, thus benefiting the colony as a whole, rather than the individual player.
Somehow, mankind as a whole needs to make a similar observation.
Slashdot Mirror
It wouldn't have to be quite so bad. One possibility is to sense where the viewer is and only send those pixels. Another is to send enough 2D layers such that the 3D scene can be recomputed properly for a wide variety of viewing positions (re: image-based rendering). That is still assuming stereoscopic display, though.
Your idea of using 1920x1080x1080 only permits you to look into a TV-size cube of space. What is more desirable is if the TV represents a "window" into another world of infinite dimension. This calls for a lightfield display, where each pixel sends out different light rays (of varying color & intensity) in different directions. Essentially, each pixel delivers a whole image, directionally dispersed. The resolution of each "pixel image" depends upon the range of viewing angles you wish to support, as well as the fidelity of angular resolution you want, of course. But if you assume an HD "image" for each pixel in an HD image, then you're talking 4 trillion total "pixels" (light rays), 60 times per second.
Seems like $700-$800/month is more likely:
http://www.nypost.com/p/news/local/manhattan/small_price_to_pay_yk3QVpgdWYWTBnEhJ0QPXN
The drive controller has a clock that it uses to control the width (or length, actually) of the bits that it writes. However, due to temperature variation (quartz oscillators are temperature sensitive) and other factors, the clock that is used when the bits are read may be slightly different than the clock that was used when the bits were written. A PLL (phase lock loop) can keep the clocks aligned, but it needs a certain minimum number of transitions per unit time to take place to keep them aligned.
In truth, reading & writing bits is actually more complicated than that, since drive manufacturers have pushed the technology past the point of simply reading distinct 1's & 0's from the platter. The signal is analog, and when you try to squeeze the bits closer & closer together, it no longer resembles just "high" & "low" levels (or positive and negative, given that it's sensing the alignment of a magnetic field). Instead, you get a signal that looks more like noise, and it requires some fancy signal processing to figure out from what is read what the original pattern written was. The technique is called "PRML" or partial response, maximum likelihood.
To understand sync, you need to understand that, at a certain level, the signal being read from the drive head is a stream of 1's & 0's. To make sense of this stream, you need to know the boundaries of things. At one level, you need to know where each byte begins and ends. At a higher level, you need to know where each sector begins and ends. That's what sync is all about. To "do" sync, you use a sequence of 1's & 0's that you'd never use for actual data. When you see that sequence, you know, "ah, this is the start of something". Once you see the start, then you'll know where you are as long as you keep track of things and certain rules are followed. For instance, one rule is that you can't have too many 1's or 0's consecutively, or else you'll lose your place, since you can't actually see the "start" of a 1 or 0, but you can only see the changes from a 1 to a 0 or vice-versa.
To further understand the need for sectors, you must also think about how you write data to the disk. First, the disk has to read the data off the disk while the platter is spinning in order to find the start of the right sector. Then it has to switch the head into write mode and write out the new data. It must stop writing data at the end of this sector in order to not overwrite the sync header for the next sector. Due to several sources of variations (platter spin speed, temperature, mechanical play, clock frequency, etc.), there needs to be a bit of "dead zone" (ie, inter-sector gap) between sectors to make sure that the sync headers are never overwritten.
By going to larger sectors, you have fewer of them, and therefore need fewer inter-sector gaps.
Computer people who've been around may remember that the Commodore Amiga could write 880KB on floppy disks when PCs could only write 720KB. This was because the Amiga didn't use multiple sectors per track, but rather always just read & wrote an entire track's worth of data for each access. Similar idea going on here.
The correct technical term for "breakage" is "internal fragmentation".
To become filthy rich, you need a few things:
1) The vision to first see what's available now, and second see the market opportunity in the future.
2) Some connections to smart people who can take you to that second step. Connections to money help, too.
3) The motivation to get there & keep going, without letting anyone slow you down.
A lack of certain ethics may help with #3.
Lots of people may have one or two of these. Not so many have all three.
I don't think that patent holders should be allowed to escape licensing by claiming that they will bring a product to market themselves. Perhaps they can be given a time window in which to do so. If they are not satisfied with this, then they always have the option of not patenting.
As I wrote in another article, patents & copyrights should serve 2 purposes:
-encourage innovation & creativity by providing a limited period to profit from it.
-add to the public domain after said period expires.
The second part of the deal is generally being forgotten about in this day of "I do what's best for me" (vs. I do what's best for all).
> Also, you are proposing regulated pricing, AKA socialism, and bucking the free market.
No, not at all. The formula could very well be market-based. In fact, you may have hit upon the solution: a market (like the stock market) for patent licenses. Of course, this brings up a host of new issues to solve, but then, what's life for?
> achieve the same thing like by implementing features that are not supported in other platforms.
I think you hit upon it exactly. There's usually loopholes around legal issues. It works in MS's favor to have everyone embrace their technology; then, they can make their own platform special by extending the technology in ways that others cannot.
I think patent law should be modified to disallow patent trolling. Patent holders should be required to license patents for reasonable fees, or else lose the patents.
Of course, the key issue is determining what a "reasonable" fee should be, but I imagine that someone smart enough can come up with a reasonable formula.
I imagine that the original purpose of both copyrights & patents was two-fold:
1. Encourage creativity & innovation by providing a limited time period to profit on new works & ideas.
2. Grow the public domain of ideas after said limited time expires.
These days, however, that second part has been forgotten, because the word "profit" only appears in the first part, and big companies will do all that they can to keep profiting from something.
I tend to visualize commerce as a flowing river or stream. You make something people want to buy, and money flows in your direction when you sell. Big companies want to maximize profits by finding existing streams, claiming ownership over them, and redirecting the flow into themselves. Then they do everything that they can to prevent that stream from drying up or otherwise escaping their control.
In essence, assuming the two purposes stated above are true, big companies are stealing from the public by claiming longer & longer holds on creative works & inventions. They have the laws changed to favor themselves, and Joe Public doesn't fight back.
The main accomplishment of Fastra II is that they put 7 GPU cards into the machine. Fastra I already had 4 GPU cards, which is what the FOH machines top out at. The GPU cards are the same cards all around (Nvidia GTX 295). The linked article & video point out the difficulties they faced & overcame in trying to make 7 GPU cards work simultaneously in the same box.
until we change our mind.
Really, what's to prevent them from waiting until the tech is firmly embraced, then changing the deal?
Oh, and here's your mini-fridge size case with 10 slots:
http://www.mountainmods.com/computer-cases-c-21.html
Oh, and by the way, I'm wondering quite the opposite: why do we still see so many over-sized full ATX size cases being offered, when microATX motherboards have everything we (most of us) need? Indeed, even mini-ITX motherboards are often adequate for so many needs, and yet mini-ITX cases still seem to command a premium because they are relatively rare. It's easy (and boring) to design a big rectangular ATX box. It's an engineering challenge to make a good-looking small box that does everything you need and is still practical to work with.
It's known as "market forces". In case you haven't noticed, the computing needs of most people can be crammed into something the size of a paperback book or so. Larger computing devices are available, but the bigger you go, the smaller the market, and thus the larger the price. If you want something big, you might take look at a computer named "Jaguar". It has a big price, too.
As far as personal computers go, they tend to be designed around CPU strengths & limitations. Intel and AMD have figured out that the most efficient way to increase computing power is to put more and more processing power into a single chip, and have systems designed around a single CPU chip, as opposed to systems that put multiple CPU chips on the motherboard. Because of this approach, it became unnecessary to build systems larger than your typical ATX desktop.
If you needed more computing power than that, your best bet was to get multiple machines. Indeed, you can fill refrigerator-sized racks full of ATX (or other form factor) motherboards. For instance, check out: http://www.cse.illinois.edu/turing/Images/FrontView.html
Only recently have GPUs become recognized as an efficient way of adding lots of computing power to a desktop machine. As evidenced by the motherboard that made Fastra II possible, hardware is slowly becoming available to embrace this new computing paradigm. Perhaps in a few years, you'll get your 12-double-wide-slot motherboard and you'll be able to populate it with GeForce 28000 boards. But more than likely, it still won't be cheap, since few people seem to need this kind of performance.
There were several difficulties. The most obvious is that they fit 7 double-wide cards into 7 single-wide slots. The next was that the motherboard BIOS crashes when more than 5? boards are installed. The next was that in order to allocate enough I/O space, all unnecessary devices had to be disabled, and even still the Linux kernel needed to be hacked to reduce the space allocated to various resources. After all that, it was a piece of cake.
Apparently, the regular BIOS can't boot with more than 5? graphics cards installed due to the amount of resources (memory & I/O space) that each one requires. So the researchers asked ASUS to make a special BIOS for them which doesn't set up the graphics card resources. However, the BIOS still needs to initialize at least one video card, so they agreed that the boot video card would be the one with only a single GPU. Presumably, they could have also chosen a dual GPU card that happened to be different from the others in some way.
It makes perfect sense, given appropriate units, such as (1/watt)'s. Okay, maybe not.
It's not silly: (1) this is a research project, not production medical equipment, meaning that the funds to buy Tesla cards were probably not available, and they aren't particularly worried about occasional bit errors. (2) Their particular application doesn't need much inter-GPU communication, if any, so that bandwidth is not an issue. They just need for each GPU to load datasets, chew on them, and spit out the results.
How much does your proposed GPU supercomputer cost for 13 GPUs?
Did more what, exactly? None of the Folding setups listed have more than 4 GPU cards per motherboard.
Ah, I'm wrong. From TFA:
"After a one-month checkout period, WISE will spend six months mapping the whole sky. It will then begin a second scan to uncover even more objects and to look for any changes in the sky that might have occurred since the first survey, according to NASA. This second partial sky survey will end about three months later when the spacecraft's frozen-hydrogen cryogen runs out."
It may start repeating sections that it has done before it finishes the whole thing. (I didn't RTFA yet.)
We are arguing different points.
You are focusing on the specifics of this particular case. I have made no comments about whether Apple's actions are "good" or "bad," but rather that the whole situation is kind of ridiculous.
I'm trying to see beyond this particular case and envision what set of conditions would allow for companies in general to focus on "progress" (whatever that means) instead of in-fighting.
This particular case revolves around patents. These days, patents are kind of like weapons. Companies try to generate or otherwise acquire as many patents as possible as kind of an arms race. Once a company has achieved a critical mass of patents, they generally try to work out non-aggression pacts (cross-licensing agreements) with their competitors to avoid patent wars. But the landscape is full of pirates (in the "Aaarh!" sense) who will gather up some patents to go on offensives just to score some quick booty.
Anyone who's given the matter some thought will realize that the present patent system is broken and needs revamping. However, companies are kind of like drug addicts with regard to patents. They can see the damage that they can do, but they really don't want to give them up.
Regarding patents, I think it's time for rehab. However, patents are just one of the many ways that corporate culture is messed up. Trying to fix everything would be a much bigger job than trying to overhaul the US's health-care system. And, just like that problem, if the majority of people don't see that there's anything wrong, nothing will ever change.
> When someone punches you, you punch back.
I think you're kind of missing the point. We shouldn't be behaving like cave-men in the first place.
True, the corporate environment that all companies currently exist in tends to encourage this behavior. But the nature of this corporate environment is hurting everyone (except the lawyers), so how long will it be before we can all evolve and pursue better things?
Reminds me of an anecdote concerning the game Mule. I played this game with my brothers, and we always competed against each other. When I mentioned the game to a friend of mine, she said "That's a great game. What was the richest your colony ever got?" I responded with some score I remembered at the time. She said something to the effect of "Why so low? We usually got 10x that!" It had not even occurred to me that one might try to play the game cooperatively, thus benefiting the colony as a whole, rather than the individual player.
Somehow, mankind as a whole needs to make a similar observation.