It is also probable that the magnitude of the additional fee required for Intel to offer this service without losing money would vary sharply between processor families.
The EE and K-series stuff is, shall we say, 'priced for the price insensitive'. Nothing wrong with that, voluntary on both sides, everybody knows that you can get 80-90% of the bang for less than half the buck by stepping back a few notches; but those parts are crazy overpriced. By contrast, their low end parts(especially in areas where they are going directly against AMD largely on basis of performance/$) aren't sold at a loss; but don't have nearly as much profit built in.
If they wanted to offer abuse insurance on value SKUs, and not lose money, the price would likely be a fair percentage of the OEM price of the CPU(very little margin on those parts, and only crazed overclockers would buy the insurance, so a high-risk pool and parts whose cost to intel is not so very different from their cost in store). Offering abuse insurance on the 'because we can' SKUs could be done at a much lower percentage of the OEM price of the CPU, because the cost to intel of that part is much lower than its price, and the entire market for those is crazed overclockers, so the additional riskiness implied by actually buying such insurance is not as great...
I'm not disputing that it, in fact, does. I have no reason to doubt it, and my understanding is that does so.
My point was just that, especially with 'media processors', where the customer demands of rather anti-user players enter the picture, the silicon isn't necessarily a good guide.
Because these are all teeny little ARM(occassionally MIPS) cores under serious power constraints, implementing media decode on the CPU is unlikely to be useful beyond a few audio codecs, which leaves you at the mercy of the special-purpose decoder unit, and its firmware, which is typically manufacturer-controlled even when the kernel and userspace running on the general purpose CPU are not.
It's in no way specific to this project(indeed, the places where it has cropped up most, to date, seem to be in the groups dedicated to hacking the various network set top boxes that Western Digital and others have released); but it is an important difference from the PC scene, where a monstrously powerful CPU(strong enough to brute force most common media tasks in adequate time, if not exactly efficient about it) and an increasingly versatile GPU are the rule.
"Media" SoCs in general(Broadcom certainly no exception) tend to combine a reasonably normal, open, well-understood ARM or MIPs general-purpose CPU with a GPU and/or hardware video decode unit. If you are lucky, these will be supported in some way(I think the BCM part here has a ~15mb blob of mystery powering the graphics bit); but they tend to be excitingly locked down because the manufacturers want to be able to sell them as set-top boxes and other areas where team DRM holds sway.
Even on devices that are "open" in the sense that the boatloader doesn't cryptographically lock out unsigned or self-signed kernels, it may well be the case that the media-related peripherals will lock out unsigned firmware blobs, which allows specific features of the peripheral to be locked or unlocked by the manufacturer without the expense of respinning the die(eg. for devices that are or are not H264 patent-paid.)
I'm guessing that the Gumstix guys might not be too happy, though. Their stuff is super nifty; but ends up being about a factor of ten more, once you get enough boards connected to do some I/O...
Aside from the SoC procurement issues, I suspect that the demand for 3rd party spins might be fairly small(not zero; but fairly small): The Raspberry Pi people seem to be shooting for lowest cost, so there isn't a lot of pure margin for a 3rd party to cut into and manufacturing cheaper than the guys who are already attempting to manufacture cheap is going to be tricky.
With something like the Arduino, there was definitely room for a bunch of 3rd-party spins: the original was pretty expensive, included a lot of arguably optional parts; but was simple enough that redesigns were doable: hence the versions that could be constructed single-layer, through-hole only, the size-minimized 'shieldless' versions, the versions without USB-serial or RS232-TTL converters, etc.
It's a pity, really; because there are definitely some niche-focused spins of the board that I would like to see; but it will have to get pretty enormously popular before that is likely, and pure clones would be a cutthroat business to get into...
Prisons also have staff, visitors, and(depending on the local jurisdiction's distinction between prison and jail, if any), suspects awaiting trial but not convicted of anything.
I reserve my humanity for people who act humanely. Is that bad?
Inconveniently(but as is generally true with situations where a given disease has a popular association with some flavor of moral failing), infectious diseases are generally quite happy to fester among the undesirables and then start leaking into the general population by whatever intersections between us and them exist but are politely not thought about.
Even if every inmate deserved it, anybody who would rather not have prison guards and their families, inmates released after comparatively short sentences, etc. continually bringing hardcore prison TB into the general population has a strong, pragmatic interest in making sure that they don't get it...
Regardless of whether it is an ethical problem, allowing dangerous diseases to become endemic within some distasteful underclass is foolish(but an excellent way to learn more about how that underclass connects to society at large...)
The chap in question mentioned "'the frequent failure of our space launches, which occur at a time when they are flying over the part of Earth not visible from Russia, where we do not see the spacecraft and do not receive telemetric information, are not clear to us,"
My question was, for a civilian scientific launch of an object so close by, Russia couldn't obtain reasonably good visual and telemetric data either by simply politely requesting that the US not interfere with part of a purely scientific operation, or by some degree of stealth. There should be any number of locations in Alaska, the Yukon, or the Northwest Territories, and nearby waters, where nobody would pay you the slightest attention(coast guard might pick up a boat on radar; but as long as there wasn't anything unusual about it, they have a lot of cold, icy water to cover) if you set up a modest visual observatory and RF communication system for a shortish period of time. Unless you really shot the job somehow, I'd suspect that it would be most likely for nobody to notice, with lesser possibilities of having your gear confiscated or your people brought up on some sort of border/immigration violations...
That would do absolutely nothing to stop anything being done; but it would resolve the curvature-of-earth-hides-the-object problem.
Also, if the unit is only 50 miles up, one could presumably have a friendly chat with it using comparatively small, low-power, hardware. The sort of hardware that you could either place covertly or (this is a civilian scientific mission, after all) just have some guys with beards and PhDs walk right in with...
HAARP is the pet villain of practically every flavor of fun conspiracy lore there is. Weather control? Check. Mind Rays? Check. Communications with the Greys? Check. Interfering with Orgone flows to ensure the success of the fluoridation conspiracy? Check. Guiding black helicopters back to their spawning grounds to mate and reproduce? Check.
If he thinks that he can just waltz in and grab some time out of HAARP's very busy schedule to have it sabotaging his spacecraft, he has another thing coming. He'll have to fight for HAARP time with practically every conspiracy theorist out there...
Unfortunately, it'll be a different half-assed build, with a different shit interface, and tragicomedic 'app store', on every single model...
The only thing they'll have in common is being cryptographically locked, so that the only thing that can be installed are the manufacturer firmware updates that never materialize.
Biometrics must be the 'security' concept that combines the worst features with the best wiz-bang sci-fi aesthetic appeal... I can only assume that it was invented during a sort of 'product blackjack', where a group of players competed to see who could come up with the most awful ideal that could still be successfully sold...
"Hey guys, I'm trying to build a truly awful security system. Can anybody think of something like a password, only absurdly hard to change voluntarily, occasionally changed traumatically by forces beyond the user's control, and preferably left in traces all over the place during the course of daily life? Drinks are on me if successfully compromising it for one institution renders it strongly likely that it will be compromised across a large number of unrelated ones simultaneously!"
This doesn't help if you need a gigabit link to your switch, or only have one PC,(or, most likely, if you just don't look good in a tinfoil hat); but constructing a passive tap for 10/100 ethernet is trivial and allows you to sample the ether between your system and the hostile world of the internet from a second host.
If you need gigabit, or want to be all classy about it, you'll need a switch with port mirroring; but this is the easy and cheap way to slip an almost-certainly-OK-because-it-was-just-booted-from-LiveCD system onto the wire to have a look at what a possibly compromised host is doing...
Symantec has a well-deserved reputation for being atrocious; but pretty much any AV mechanism that does on-access scanning(which is most of them by default, though it can generally be turned off somewhere, if you feel particularly lucky) is going to tank your apparent disk access speeds, since the AV process has to chew on the data before handing them over to the program requesting them. Unless you have an SSD or a fairly punchy RAID setup, lousy disk access speeds are one of the best ways to make a system feel miserably slow, especially now that abundant RAM and fast CPUs are so cheap...
In this case, his advice is probably correct for those running Windows at home, fluff about his decade-long record of having no viruses he has noticed aside. Security Essentials is 'free' as in 'bundled with your Windows license'; but if you've got a Windows license already, that makes it cheaper than anything that costs additional money and the products that do make a very, very, very, tepid case for why you should purchase them.
In corporate use, it isn't as clear; because ForeFront sure as hell isn't free, or necessarily superior to competing products(no matter how cynical you attempt to be, it is shocking how much more awful AV software is when aimed at intimidating some poor end user who got 90 days 'free' with their best buy box, rather than it is aimed at IT and therefore mostly keeps its mouth shut on the client side, so even some of the vendors that you wouldn't touch with somebody else's 10-foot pole at home can at least produce unobtrusive software for corporate.)
All that they would really have to do(even ignoring various 'probably horribly illegal; but do you have 10 years and a good lawyer, punk?' strategies), is have a list and issue a little certificate, and make operating without the little certificate the offense. The list would be secret; but could be used to check suspected forged certificates, and DefinitelynottheCIA LLC and friends could simply hold their certificates privately, making assembling a list of certificate holders very difficult.
It would then be trivial to prove guilt without revealing anything about the set of licensed parties: "Sir, regulations require that you possess a drone license to operate a drone. If yours is lost or stolen, you can simply contact us for a temporary drone license and a replacement in 5-8 business days. We have no record of your drone license, or any report of loss; but we do have record of you operating a drone. Please present your drone license."
Just what I've always wanted in a durable metal tool: a delicate electronic attachment highly likely to break or become obsolete well before the rest of the thing. Can we think of a way to make it rely on some obscure teeny batteries, with a chintzy plastic door that falls off if you look at it funny? That's the only thing I like even better...
Just get a damned K-bar and intimidate the bits at your destination into the correct pattern.
If the, um, situation with the "GMA500" is anything to go by, people will be begging for an intel GPU(or the sweet embrace of death, which might actually move more polygons per second...)
That was the previous generation attempt at a very low power Atom platform(also used in the embedded-only CE1400 chips, for set-top boxes), and was also a PowerVR licensed core(Intel's in-house at the time was the GMA950, an entirely different animal). Now, it may have been nice and cool-running; but the driver situation was a Lovecraftian clusterfuck of epic dimensions. The GMA950 was slow as a dog, and driver support for actual features relentlessly delayed; but it was fairly mature and well-behaved, so long as not asked to actually exert itself. The GMA500 was an opaque nightmare on the One True Kernel/xorg Combination Of Antioch, and simply unsupported elsewhere.
Unless Intel has kidnapped the children of selected PowerVR executives this time around, I'm not sure I want to know how much fun the upcoming SoCs are going to be...
History suggests that 'voluntary'('voluntary' in the sense that would render a contractual relationship valid according to the usual principles of contract law, meeting of minds, not of adhesion, no force or fraud, etc.) relationships between unequal actors are vanishingly rarely a stable situation.
The chap who wants the compound tested almost always has substantially more data(from preliminary pre-human study) about safety than does the chap who will potentially be testing it. He does not have an incentive to be forthcoming about those data unless he thinks that it will make recruiting test subjects easier and/or cheaper... What, as they say, could possibly go wrong?
Most likely, they lied twice: The final product was tested on customers and the ingredients were almost certainly tested, on animals, prior to general availability, just not tested in this particular combination... You can put your "cruelty free" sticker on the box without reference to your supply chain.
Some mice are more expensive than others. Your basic boring brown ones are pretty damn cheap, as are common research variants.
A bit of poking around on the expensive side of the menu though, and you can end up paying north of $200/mouse, plus any additional costs for special requests.
Of course, since this sensor widget is designed to be used in tissue cultures, you'll end up paying extra for exotic genomes whether in goo form or in mouse form(on the other hand, the instruments/diagnostics/dissection/whatever tests done on the mice also aren't necessarily cheap, depending on what you are testing for).
Why do people(TFS and TFA notably not excluded) insist on talking about the part in terms of its GPU performance?
Let's see here... Intel is throwing their hat into the ARM-level power arena... we could discuss how fast their processor is, or we could do a bunch of irrelevant jabbering about how fast the SGX540 that virtually everybody licenses from PowerVR is... Hmm. Hey, let's focus on the part that everybody already knows about and make it even more fascinating by not discussing power for GPU operations; but encode and decode of some (unspecified; but quite possibly a restricted baseline of H.264) 'HD Video' format, and the maximum output resolution!
It's actually a pretty impressive way to natter on about the product without the slightest mention of what may or may not make it interesting. In other news, it is probably made of silicon, and in some sort of density-optimized epoxy package!
It is also probable that the magnitude of the additional fee required for Intel to offer this service without losing money would vary sharply between processor families.
The EE and K-series stuff is, shall we say, 'priced for the price insensitive'. Nothing wrong with that, voluntary on both sides, everybody knows that you can get 80-90% of the bang for less than half the buck by stepping back a few notches; but those parts are crazy overpriced. By contrast, their low end parts(especially in areas where they are going directly against AMD largely on basis of performance/$) aren't sold at a loss; but don't have nearly as much profit built in.
If they wanted to offer abuse insurance on value SKUs, and not lose money, the price would likely be a fair percentage of the OEM price of the CPU(very little margin on those parts, and only crazed overclockers would buy the insurance, so a high-risk pool and parts whose cost to intel is not so very different from their cost in store). Offering abuse insurance on the 'because we can' SKUs could be done at a much lower percentage of the OEM price of the CPU, because the cost to intel of that part is much lower than its price, and the entire market for those is crazed overclockers, so the additional riskiness implied by actually buying such insurance is not as great...
I'm not disputing that it, in fact, does. I have no reason to doubt it, and my understanding is that does so.
My point was just that, especially with 'media processors', where the customer demands of rather anti-user players enter the picture, the silicon isn't necessarily a good guide.
Because these are all teeny little ARM(occassionally MIPS) cores under serious power constraints, implementing media decode on the CPU is unlikely to be useful beyond a few audio codecs, which leaves you at the mercy of the special-purpose decoder unit, and its firmware, which is typically manufacturer-controlled even when the kernel and userspace running on the general purpose CPU are not.
It's in no way specific to this project(indeed, the places where it has cropped up most, to date, seem to be in the groups dedicated to hacking the various network set top boxes that Western Digital and others have released); but it is an important difference from the PC scene, where a monstrously powerful CPU(strong enough to brute force most common media tasks in adequate time, if not exactly efficient about it) and an increasingly versatile GPU are the rule.
Assuming that particular bit is flipped...
"Media" SoCs in general(Broadcom certainly no exception) tend to combine a reasonably normal, open, well-understood ARM or MIPs general-purpose CPU with a GPU and/or hardware video decode unit. If you are lucky, these will be supported in some way(I think the BCM part here has a ~15mb blob of mystery powering the graphics bit); but they tend to be excitingly locked down because the manufacturers want to be able to sell them as set-top boxes and other areas where team DRM holds sway.
Even on devices that are "open" in the sense that the boatloader doesn't cryptographically lock out unsigned or self-signed kernels, it may well be the case that the media-related peripherals will lock out unsigned firmware blobs, which allows specific features of the peripheral to be locked or unlocked by the manufacturer without the expense of respinning the die(eg. for devices that are or are not H264 patent-paid.)
I'm guessing that the Gumstix guys might not be too happy, though. Their stuff is super nifty; but ends up being about a factor of ten more, once you get enough boards connected to do some I/O...
Aside from the SoC procurement issues, I suspect that the demand for 3rd party spins might be fairly small(not zero; but fairly small): The Raspberry Pi people seem to be shooting for lowest cost, so there isn't a lot of pure margin for a 3rd party to cut into and manufacturing cheaper than the guys who are already attempting to manufacture cheap is going to be tricky.
With something like the Arduino, there was definitely room for a bunch of 3rd-party spins: the original was pretty expensive, included a lot of arguably optional parts; but was simple enough that redesigns were doable: hence the versions that could be constructed single-layer, through-hole only, the size-minimized 'shieldless' versions, the versions without USB-serial or RS232-TTL converters, etc.
It's a pity, really; because there are definitely some niche-focused spins of the board that I would like to see; but it will have to get pretty enormously popular before that is likely, and pure clones would be a cutthroat business to get into...
Countess Bathory, MD, PhD, postumously awarded the Nobel Prize in Medicine?
Prisons also have staff, visitors, and(depending on the local jurisdiction's distinction between prison and jail, if any), suspects awaiting trial but not convicted of anything.
I reserve my humanity for people who act humanely. Is that bad?
Inconveniently(but as is generally true with situations where a given disease has a popular association with some flavor of moral failing), infectious diseases are generally quite happy to fester among the undesirables and then start leaking into the general population by whatever intersections between us and them exist but are politely not thought about.
Even if every inmate deserved it, anybody who would rather not have prison guards and their families, inmates released after comparatively short sentences, etc. continually bringing hardcore prison TB into the general population has a strong, pragmatic interest in making sure that they don't get it...
Regardless of whether it is an ethical problem, allowing dangerous diseases to become endemic within some distasteful underclass is foolish(but an excellent way to learn more about how that underclass connects to society at large...)
The chap in question mentioned "'the frequent failure of our space launches, which occur at a time when they are flying over the part of Earth not visible from Russia, where we do not see the spacecraft and do not receive telemetric information, are not clear to us,"
My question was, for a civilian scientific launch of an object so close by, Russia couldn't obtain reasonably good visual and telemetric data either by simply politely requesting that the US not interfere with part of a purely scientific operation, or by some degree of stealth. There should be any number of locations in Alaska, the Yukon, or the Northwest Territories, and nearby waters, where nobody would pay you the slightest attention(coast guard might pick up a boat on radar; but as long as there wasn't anything unusual about it, they have a lot of cold, icy water to cover) if you set up a modest visual observatory and RF communication system for a shortish period of time. Unless you really shot the job somehow, I'd suspect that it would be most likely for nobody to notice, with lesser possibilities of having your gear confiscated or your people brought up on some sort of border/immigration violations...
That would do absolutely nothing to stop anything being done; but it would resolve the curvature-of-earth-hides-the-object problem.
Also, if the unit is only 50 miles up, one could presumably have a friendly chat with it using comparatively small, low-power, hardware. The sort of hardware that you could either place covertly or (this is a civilian scientific mission, after all) just have some guys with beards and PhDs walk right in with...
I wonder if he knows what he is getting into.
HAARP is the pet villain of practically every flavor of fun conspiracy lore there is. Weather control? Check. Mind Rays? Check. Communications with the Greys? Check. Interfering with Orgone flows to ensure the success of the fluoridation conspiracy? Check. Guiding black helicopters back to their spawning grounds to mate and reproduce? Check.
If he thinks that he can just waltz in and grab some time out of HAARP's very busy schedule to have it sabotaging his spacecraft, he has another thing coming. He'll have to fight for HAARP time with practically every conspiracy theorist out there...
Unfortunately, it'll be a different half-assed build, with a different shit interface, and tragicomedic 'app store', on every single model...
The only thing they'll have in common is being cryptographically locked, so that the only thing that can be installed are the manufacturer firmware updates that never materialize.
Biometrics must be the 'security' concept that combines the worst features with the best wiz-bang sci-fi aesthetic appeal... I can only assume that it was invented during a sort of 'product blackjack', where a group of players competed to see who could come up with the most awful ideal that could still be successfully sold...
"Hey guys, I'm trying to build a truly awful security system. Can anybody think of something like a password, only absurdly hard to change voluntarily, occasionally changed traumatically by forces beyond the user's control, and preferably left in traces all over the place during the course of daily life? Drinks are on me if successfully compromising it for one institution renders it strongly likely that it will be compromised across a large number of unrelated ones simultaneously!"
This doesn't help if you need a gigabit link to your switch, or only have one PC,(or, most likely, if you just don't look good in a tinfoil hat); but constructing a passive tap for 10/100 ethernet is trivial and allows you to sample the ether between your system and the hostile world of the internet from a second host.
If you need gigabit, or want to be all classy about it, you'll need a switch with port mirroring; but this is the easy and cheap way to slip an almost-certainly-OK-because-it-was-just-booted-from-LiveCD system onto the wire to have a look at what a possibly compromised host is doing...
Symantec has a well-deserved reputation for being atrocious; but pretty much any AV mechanism that does on-access scanning(which is most of them by default, though it can generally be turned off somewhere, if you feel particularly lucky) is going to tank your apparent disk access speeds, since the AV process has to chew on the data before handing them over to the program requesting them. Unless you have an SSD or a fairly punchy RAID setup, lousy disk access speeds are one of the best ways to make a system feel miserably slow, especially now that abundant RAM and fast CPUs are so cheap...
In this case, his advice is probably correct for those running Windows at home, fluff about his decade-long record of having no viruses he has noticed aside. Security Essentials is 'free' as in 'bundled with your Windows license'; but if you've got a Windows license already, that makes it cheaper than anything that costs additional money and the products that do make a very, very, very, tepid case for why you should purchase them.
In corporate use, it isn't as clear; because ForeFront sure as hell isn't free, or necessarily superior to competing products(no matter how cynical you attempt to be, it is shocking how much more awful AV software is when aimed at intimidating some poor end user who got 90 days 'free' with their best buy box, rather than it is aimed at IT and therefore mostly keeps its mouth shut on the client side, so even some of the vendors that you wouldn't touch with somebody else's 10-foot pole at home can at least produce unobtrusive software for corporate.)
They are merely respectable businessmen, offering you their protection...
All that they would really have to do(even ignoring various 'probably horribly illegal; but do you have 10 years and a good lawyer, punk?' strategies), is have a list and issue a little certificate, and make operating without the little certificate the offense. The list would be secret; but could be used to check suspected forged certificates, and DefinitelynottheCIA LLC and friends could simply hold their certificates privately, making assembling a list of certificate holders very difficult.
It would then be trivial to prove guilt without revealing anything about the set of licensed parties: "Sir, regulations require that you possess a drone license to operate a drone. If yours is lost or stolen, you can simply contact us for a temporary drone license and a replacement in 5-8 business days. We have no record of your drone license, or any report of loss; but we do have record of you operating a drone. Please present your drone license."
Just what I've always wanted in a durable metal tool: a delicate electronic attachment highly likely to break or become obsolete well before the rest of the thing. Can we think of a way to make it rely on some obscure teeny batteries, with a chintzy plastic door that falls off if you look at it funny? That's the only thing I like even better...
Just get a damned K-bar and intimidate the bits at your destination into the correct pattern.
If the, um, situation with the "GMA500" is anything to go by, people will be begging for an intel GPU(or the sweet embrace of death, which might actually move more polygons per second...)
That was the previous generation attempt at a very low power Atom platform(also used in the embedded-only CE1400 chips, for set-top boxes), and was also a PowerVR licensed core(Intel's in-house at the time was the GMA950, an entirely different animal). Now, it may have been nice and cool-running; but the driver situation was a Lovecraftian clusterfuck of epic dimensions. The GMA950 was slow as a dog, and driver support for actual features relentlessly delayed; but it was fairly mature and well-behaved, so long as not asked to actually exert itself. The GMA500 was an opaque nightmare on the One True Kernel/xorg Combination Of Antioch, and simply unsupported elsewhere.
Unless Intel has kidnapped the children of selected PowerVR executives this time around, I'm not sure I want to know how much fun the upcoming SoCs are going to be...
History suggests that 'voluntary'('voluntary' in the sense that would render a contractual relationship valid according to the usual principles of contract law, meeting of minds, not of adhesion, no force or fraud, etc.) relationships between unequal actors are vanishingly rarely a stable situation.
The chap who wants the compound tested almost always has substantially more data(from preliminary pre-human study) about safety than does the chap who will potentially be testing it. He does not have an incentive to be forthcoming about those data unless he thinks that it will make recruiting test subjects easier and/or cheaper... What, as they say, could possibly go wrong?
Most likely, they lied twice: The final product was tested on customers and the ingredients were almost certainly tested, on animals, prior to general availability, just not tested in this particular combination... You can put your "cruelty free" sticker on the box without reference to your supply chain.
Some mice are more expensive than others. Your basic boring brown ones are pretty damn cheap, as are common research variants.
A bit of poking around on the expensive side of the menu though, and you can end up paying north of $200/mouse, plus any additional costs for special requests.
Of course, since this sensor widget is designed to be used in tissue cultures, you'll end up paying extra for exotic genomes whether in goo form or in mouse form(on the other hand, the instruments/diagnostics/dissection/whatever tests done on the mice also aren't necessarily cheap, depending on what you are testing for).
Not to worry, you'll have a fine time proving that your name is on the list...
Why do people(TFS and TFA notably not excluded) insist on talking about the part in terms of its GPU performance?
Let's see here... Intel is throwing their hat into the ARM-level power arena... we could discuss how fast their processor is, or we could do a bunch of irrelevant jabbering about how fast the SGX540 that virtually everybody licenses from PowerVR is... Hmm. Hey, let's focus on the part that everybody already knows about and make it even more fascinating by not discussing power for GPU operations; but encode and decode of some (unspecified; but quite possibly a restricted baseline of H.264) 'HD Video' format, and the maximum output resolution!
It's actually a pretty impressive way to natter on about the product without the slightest mention of what may or may not make it interesting. In other news, it is probably made of silicon, and in some sort of density-optimized epoxy package!