In the event of a nasty solar storm, the brits can just pile up the acronyms generated by their preparedness program and hide behind those. The bureaucratic verbiage should be dense enough to stop all but neutrinos.
"BIS", the service likely shoving email for consumer blackberries, is RIM-hosted, authenticated through a carrier-branded RIM system, and can be compromised with RIM's cooperation, since they run it. This is the one that various governments(India, UAE, probably a whole lot of others who are quieter about it) are leaning on RIM to have hosted domestically, so they can keep an eye on their little consumers and ensure that they aren't getting up to mischief.
"BES", the historically pricey and complex enterprise offering, is hosted by the customer and wraps its tentacles around the customer's mailserver. Communications between the BES and the company blackberries do rely on RIM's network(which is how their mistakes can cause outages); but are encrypted at the BES with a secret that is generated at the customer site and not known to RIM.
Newer blackberries, at least, can also manage at least rudimentary email connectivity on their own(POP/IMAP) which is yet a third distinct situation in security terms.
The people who emphasize blackberry security usually mean 'in a BES context', along with the relative absence of trojans and bugged apps and whatnot, rather than the consumer version, which is about as secure from RIM as your webmail is from whoever is providing it...
There is an unfortunate additional quirk in this case: Since, naturally, the 'cyberwarriors' don't want to be stuck purely in the tedious and thankless job of playing defense, there is a demand for 'offensive capabilities'. This creates a perverse incentive: If a flaw is disclosed and patched or mitigated, it is no longer of offensive utility, so now the market for zero-days and exploit payloads isn't just black hats, scammers, and criminals; but 'respectable' defense industry types.
VUPEN is the crass, attention-whoring, bad-boy of the industry; but practically the entire who's-who of staid, tight-lipped, defense contractors has a division peddling bugs somewhere in the business.
Even if we were 100% warm and fuzzy about the use these exploits are being put to by these firms customers(Only the good guys, pinkie swear!), this situation is insane from the perspective of actual 'security'. Whose economies, financial systems, and infrastructure depend most heavily on complex IT systems? Ummm, mostly wealthy developed countries. Whose citizens are most vulnerable to electronic compromise of financial information and such? Countries with high rates of internet penetration and lots of computers. Who has the capability to deploy electronic attacks against unpatched vulnerabilities? Virtually everyone.
In addition to the usual grab for rights and money, this 'cybersecurity' industry begets insecurity, because of the demand for 'offensive capabilities', despite the fact that we are the ones with the most to lose in an insecure environment. At least classic corporate welfare military R&D is merely expensive, and once you hand over the money, Raytheon or whoever goes off to build some impractical toy that is largely useless; but at least largely harmless....
According to TFA, this 'social loot' nonsense requires some sort of affiliate ID baked in(presumably to the usual bit of gibberish at the end of the URL) for tracking the spamming performance of their little minions.
With any luck, this should allow automated recognition of people who are astroturfing for these guys and it's always good to have a new way of identifying awful people. At a service level, the astroturf can then be removed, downranked by search engines, etc. At a personal level, we can each do our part by reminding those culprits we know that spammers are abhuman scum who go to the special hell, and deserve it.
Intel makes some decent, and damn cheap, atom boards(This model is among the cheapest). Unfortunately, those are a bit on the large side: normal ATX and P4 power connectors, and a full PCI slot are nice for compatibility; but they strongly suggest that the board hasn't been tightened up nearly as much as it could be. For whatever reason, Intel has left the yet-smaller niches to VIA, who still manage to charge north of $200 for their ultracompact x86 boards, despite tepid performance. I'm not quite sure why Intel hasn't spun an Atom board of the same size and priced it $50 lower; but they don't seem to have done so.
Then there are the good, old fashioned, PC/104 boards. They tend not to be the cheapest, and a fair amount of the I/O might be on oddball headers; but they are available in a fair assortment of x86 flavors...
Plan 9's explicitly multi-node design would be very handy in such a situation. It wouldn't actually be a terribly good interface; but who could resist having every device in their house at their beck and call in/dev where they belong?
I don't know exactly how the marketshare numbers break down; but x86 compatibility is big for 'embedded' projects that are relatively short production run but not terribly power constrained(like digital signage). In smallish quantities, the development time savings often dominate the hardware cost difference.
Also, customers are likely to expect to be able to do things like 'yeah, just put this powerpoint I made up on the screen' or 'Here's the flash intro our brand consultant put together'... Much easier when you can just slap a copy of Windows Embedded whatever and a hardware watchdog to reset it if needed than to whip up something that can digest crap like that into some more neutral format...
It will actually be interesting to see who cracks it, if any.
Apple is way better at imposing their vision on somewhat fragmented and mediocre ecosystems and, by brutally culling the unfit even among its own, establishing itself as the center of the new ecosystem. However, there is a certain amount of built-in 'does not play well with others' required by this strategy.
Microsoft, by contrast, has never been known for their nimble elegance; but their bread and butter, corporate and home, is cobbling things together, with a strong emphasis on not stepping on legacy customer toes, accommodating all kinds of shit from the hardware OEMs, and collaborating with those who they cannot crush, rather than just ignoring them.
If 'home automation' ends up being a relatively small number of devices per home, with limited grasp into the uglier and less-often-replaced corners, Apple should handily out-shiny Microsoft. If, on the other hand, it involves playing a long grind, licensing from existing vendors, putting up with 'smart' devices that ship with nothing but binary drivers, and customers crying if the support cycle isn't 15 years long, Microsoft might well be able to outfight them. MS may not like it; but they certainly know how to coddle legacy customers and systems. Apple can barely contain their desire to crush them and move on to the new hotness.
You could ask your eardrums how well moving things with sound works in a less dense fluid...
I suspect that the mathematical trickery required to get sound waves to push an object in a concerted way, rather than just bouncing around chaotically, was a fair trick; but (normally trivial) moving things is what makes sound sound like sound.
Cute concept, really: a gumstix board with a chunk of flash running reduced complexity variants of certain network-interactive processes that the host computer was running(bittorrent and AIM in the mockup). When the host went to sleep, the gumstix transparently hijacked the host's MAC and continued doing things like providing an AIM status and dumping incoming bittorrent data to its flash card. If something happened that required the host PC's attention(ie. flash card filled up), the gumstix would wake the host, copy any data from when the host was asleep that it would need to know about, and kicked control of network activity back to the host.
At present, I don't think Microsoft has done anything with the idea, despite it being a fairly logical extension of Windows Sideshow, which appears to have been left to wither and die, instead.
It's funny, actually: in this case, MS Research hacked something together with a linux device because linux devices are a great platform for hacking stuff together; but this is probably an area where an in-house-technologies implementation might actually have been a great deal more elegant: had the little appendage-processor been running a CLR VM,.NET applications on the host computer could be presented with an interface by which they could nominate bits of themselves to be executed on the coprocessor's CLR while the main computer is shut down, rather than having an entirely different program, running on a different architecture and OS, bundled in for the purpose. Not Microsoft specific, of course, Java could be used to similar ends, and some of the 'smart dust' and mesh guys probably have things designed, rather than kludged, for the purpose; but still...
I'd be fairly surprised if this exact product ever sees the light of retail, intel kicks out prototypes of stuff all the time; but it would entirely fail to surprise me if some substantial percentage of boring-business 'ultra-small-form-factor' systems and 'compact' consumer PCs are running variously OEMed trivial variants of the design soon enough...
If the price is on par with raspberry or just above it this is going to be awesome. 293479x times the power.
Not happening, unfortunately. At retail(newegg.com used, prices for CPUs tend to be pretty similar across the board at a given time) the cheapest LGA1155 CPU is ~$40. 1.6GHz, single core, desktop binned part(unfortunately, low-end mobile CPUs don't seem to be as available in the retail channel, so I couldn't find a number for something in the mobile TDP range). At least it comes with a fan. Now, even such a puny device will brutalize a 700mhz ARM SoC designed to run from whatever battery is slim enough to fit in a contemporary cellphone; but if the CPU alone costs $5 more than the entire rpi, CPU+motherboard is going to run at least double, and RAM and boot volume still haven't been taken care of.
An overwhelmingly more powerful platform, certainly, as one would expect in a PC vs. basically-a-cellphone matchup; but the price delta is about what one would expect as well...
Yo Dog, I herd you like mac minis, so I spent the past seven years hitting a mini with the ugly stick and then released it as some sort of revolutionary device... Seriously.
Just for giggles, I then compared it to an entirely different device based around a smartphone processor and in an entirely different price bracket. This makes total sense, just trust me.
Now, purely in itself, a standardized teeny-ATX motherboard would be nice(especially if we'll someday be able to get mini-PCIe cards that aren't NICs in any quantity... If Intel is planning one, that seems like a good thing all around: the world is already cluttered with various proprietary teeny-motherboard things, and it'd be nice to have a bit of unification in that area.
However, I'm just not seeing the novelty here: The x86/embedded/industrial market has been rotten with teeny motherboards for almost as long as there has been an embedded x86 market, most laptops are built around small x86 motherboards by necessity, and some comparatively niche players, along with Apple, have released desktop products of not dissimilar size already. Historically, they've been fairly expensive, since minaturization isn't free, and Intel has no reason to cut margins on their silicon if they can avoid it. If Chipzilla has decided to drop the hammer and specify where teeny motherboards Shall put their screw holes, great; but that would be about the only new aspect of all this...
The really ugly problem with 'Home Automation' doesn't seem to be purely technological(ie. PIRs, contact switches, reed switchs, and relays are antedeluvian, adequate-speed interconnect over assorted wires also ancient, over RF also pretty old, and computers capable of crunching rulesets based on some combination of sensor outputs and RTCs are 80s stuff); but in the fact that the low-hanging fruit is too boring to be worth the trouble and the higher-hanging fruit would be massively complex and require a continual morass of several-industries-wide cooperation...
Boring case, you can drop a relay on each of your light switches, get your thermostat under control, and put some PIR presence detectors in place. If security is a concern, put up some cameras and stash a DVR somewhere. All easy; but not as inexpensive as one might like(especially in jurisdictions where touching mains current means bringing in electricians and permits and stuff), and you really have to want your bitchin' home theatre system to automatically close the drapes and dim the lights, or be very prone to leaving the stove on to fork over the additional money for the ability to remote control those things(and we are all familiar with the office comedy that is frantically waving your arms to make the 'smart' lights turn back on...)
Now, interesting case is where you expose every detail of all the devices in the home, quite possibly adding more sensors depending on the device category, and start having them cooperate intelligently to achieve various objectives. However, this is where the complexity gets ugly. Consider the history of ACPI: They wanted a way to allow computers to be more intelligent about power use, peripheral idle states, and environmental monitoring. Despite being hammered out in a near-duopoly environment(with MS on the software side and Intel on hardware), ACPI was a screaming pile of shit that barely worked properly even on mainstream OEM wintels until comparatively recently(and, even today, there are vendor-specific quirk packages, messy hacks, and peripherals that don't play quite right), never mind the poor bastards who dabbled in DIY or alternate OSes.
Now, you want the Home Of The Future? See to it that your utility meters, consumer appliances, home entertainment electronics, computers, water heaters, plumbing fixtures, climate control and thermostat systems, entry detection and security systems, and who knows what else all expose their sensors and capabilities in a standardized way. Don't let the fact that most of the items on that list have one or more industry consortia squabbling about the details of how their own little fiefdoms will be semi-standardized within themselves, much less on a broader basis, worry you.
Once you have the data to look at and the buttons and knobs to fiddle with, you just need some rulesets that make the devices collaborate intelligently and a set of interfaces that expose the power, but hide most of the complexity, to a degree sufficient that the fancy new hotness is actually worth the trouble.
Basically, you've got a problem whose complexity is fairly similar in scale to the sort of thing that smallish networking/datacenter entities would have an SNMP jocky on hand for, except that none of the hardware actually has management support yet, and the end result has to be easy enough for Joe Consumer to use. Also, it should ideally not be a dystopian surveillance nightmare or a script-kiddie playground. Not an easy problem....
It is my understanding that Microsoft Research does more or less whatever the hell they want. Very occasionally, their stuff ends up in actual Microsoft products, at which point it relentlessly marches in lockstep with the needs of Windows and Office once again. Heck, some time ago we were discussing an MS research project that used a gumstix module, running linux(not even the WinCE port that is available!) for some sort of low-power quasi-persistent network connection scheme...
It is my understanding that making a free trade agreement with the US is just the diplomatic equivalent of kissing the Don's ring.
It is certainly possible that somebody on your side will receive assistance with some problems that they may have been having; but it's a show of fealty, not a contract of equals...
At least for the present generation, I'm pretty sure that all the Tesla boards are 1 GPU per card. Nvidia supports, and sellers offer, arrangements with a fair number of cards per node(including external enclosures for expansion of systems that can't accommodate all those cards internally, 4 cards in a 1U connected by a special PCIe cable seems to have replaced the previous toaster-shaped 'deskside' chassis, with 'deskside' now handled by motherboards with loads of PCIe x16 slots); but the two-GPUs/one-card arrangements only seem to crop up on the gaming side.
I don't know why exactly this is, whether routing traces for 6 GB of RAM, per GPU, is bad enough with only one on the board, whether it is a thermal thing, or whether customer demand is such that it is cheaper to produce only single-GPU cards, and multi-slot chassis for the heavy users, rather than multiple flavors of card with various chip populations...
According to the PCI-SIG(or at least their press flacks, actual standards are members only) revision 3 is compatible with both 2 and 1, and 2 is compatible with 1(excepting one minor hiccup where 2.1 increased the allowable PCIe x16 slot power draw compared to 2.0 and earlier, so there are 2.1 cards in the wild that are logically compatible with 2.0 and 1.0; but which will only function with auxiliary power).
Internet anecdote suggests that this glorious vision may or may not actually be 100% realized, your BIOS isn't our problem, go cry to your vendor, etc; with graphics cards being the most common offenders(probably both because they are the most common user-installed PCIe peripheral, and because most motherboards won't POST properly if they can't find a working video device, while all but the really dysfunctional ones can ignore missing or confused NICs and such)...
I don't have enough personal experience to give any odds, but the SIG says yes and some people say 'not in my case'....
Who's taking odds as to the possibility that these fancy little toys will end up inflicting enough casualties to cover a front-page photo spread through either accident or malice, in a situation where just leaving them out of the picture would have gone fine?
PCIe 3.0 is, at least in magical specifications-actually-working-as-planned land, backwards compatible with 2.0 and 1.0, though obviously only at the highest mutually available speed between the two devices.
Is this optimistic theory a horrible pack of lies in general, are Nvidia products specifically broken in this respect, or do the newer ones make assumptions about bus speed that cause them to underperform on PCIe 2.0 boards?
The last company to get all "multiple core happy" and "SLI On A Board" happy was 3dfx. Who NVidia bought out when they... oh yeah, crash and burned.
Whoops.
I'm pretty sure that both ATI and Nvidia(or one of their OEM partners at the time) have kicked out a 'logically speaking, this is two cards in SLI/Crossfire; but on one card!!!' product for basically every generation since the introduction of the respective GPU-linking features.
The hard part is the fancy tricks that make cooperation between two separate GPUs work at all. Once the vendors decided that they did, in fact, want that to work, the rest is constrained largely by the fact that people willing to pay $1k for a graphics card aren't all that common(especially now that motherboards with 4 PCIe 16x slots are quite reasonably available, so even one's excessive desires can be satisfied with cheaper, more common, single chip cards).
That was my intended point: Allow(and, where necessary, legally facilitate) free parallel importation('free' as in 'subject to the same import standards as anything else, regardless of whether the manufacturer likes it or not, not as in 'all laws abolished') and watch arbitrage annihilate any excessive price differentials in short order.
Creating, and then attempting to restrain, monopoly entities is a very, very, tricky problem. On occasion, it cannot be avoided(electrical transmission systems, water piping, etc.); but parallel imports are a case where it would be trivial to allow anybody who so wishes to import goods legally purchased elsewhere, subject to no restriction beyond the import duties and standards that apply to all. Especially as Australia is an English language market, that should crush most attempts to price-discriminate pretty robustly.
Instead of this ad-hoc 'inquiry' nonsense, which is necessarily reactive and highly liable to regulatory capture, why don't we just adopt some of that 'free trade' stuff that assorted Respectable People tell us is so salubrious when the chaps who produce the products in question are shopping around for the cheapest inputs?
Absent legal barriers, arbitrage in software should cost next to nothing, especially now that much of it doesn't even come on shiny disks anymore. See to it that Australian customers can legally import goods from the location of their choice, and that middlemen can import goods from the location of their choice for domestic sale, and the price difference should collapse in a loud puff of nebulous whining about 'intellectual property'...
The whole notion behind the term 'grey market' is pernicious. It Should Not Matter whether the manufacturer/seller of a good is pleased by the ultimate destination of the goods they are selling. Yes, we would all like to enjoy perfect price discrimination. No, that isn't a good argument for letting us do so. In the absence of absurd restrictions on arbitrage, various pricing shenanigans, release-date bullshit, and other nonsense simply collapse.
Such restrictions would be one thing if they were applied evenhandedly, if the producers weren't already shopping all over the world for the lowest prices, laxest laws, and sweetest tax breaks; but they are not. You want cozy protectionism for your retail prices? Well, perhaps you shouldn't expect to enjoy worldwide free trade on your input prices... You want worldwide free trade for the things you buy? Well, that's nice, you deserve no less than worldwide free trade in the things you sell.
It is a rather curious idea, though I suspect that, in practice, people will end up cheating by throwing 'coprocessors' into the mix...
Think of, say, virtually all of the media set top box SoCs of the world. Usually ARM or MIPs(nominally); but that is just a weedy little core that draws a few UI elements and does OS housekeeping. The meat is inevitably some DSP, GPU, or video decode unit that may or may not even have a publicly known instruction set beyond the basic 'shove compressed video in like so, obtain frames and audio stream over here'.
Even the venerable x86 isn't so x86 anymore the moment you start shoving the 3D stuff reasonably seriously, at which point things kick over to a GPU of some flavor that is very much a different beast.
It also seems unlikely that they'll manage to eliminate the 'Microcontroller lurking somewhere in there to handle a bunch of very low speed GPIO with minimal expenditure of pins on the main CPU' phenomenon. 8051s and PICs and stuff turn up in the oddest places, sometimes visible, sometimes an IP core embedded in some peripheral package.
Unless the non-techies have a nigh-religious zeal for one ISA to rule them all, and rules-lawyer the process very tightly, all they'll really insure(above and beyond what the market would have delivered anyway) is that there is only one ISA, rather than 3-ish, running the 'main' OS and passing data between as many bits of hardware speaking all sorts of curious languages as the design happens to require...
Slashdot Mirror
In the event of a nasty solar storm, the brits can just pile up the acronyms generated by their preparedness program and hide behind those. The bureaucratic verbiage should be dense enough to stop all but neutrinos.
"BIS", the service likely shoving email for consumer blackberries, is RIM-hosted, authenticated through a carrier-branded RIM system, and can be compromised with RIM's cooperation, since they run it. This is the one that various governments(India, UAE, probably a whole lot of others who are quieter about it) are leaning on RIM to have hosted domestically, so they can keep an eye on their little consumers and ensure that they aren't getting up to mischief.
"BES", the historically pricey and complex enterprise offering, is hosted by the customer and wraps its tentacles around the customer's mailserver. Communications between the BES and the company blackberries do rely on RIM's network(which is how their mistakes can cause outages); but are encrypted at the BES with a secret that is generated at the customer site and not known to RIM.
Newer blackberries, at least, can also manage at least rudimentary email connectivity on their own(POP/IMAP) which is yet a third distinct situation in security terms.
The people who emphasize blackberry security usually mean 'in a BES context', along with the relative absence of trojans and bugged apps and whatnot, rather than the consumer version, which is about as secure from RIM as your webmail is from whoever is providing it...
There is an unfortunate additional quirk in this case: Since, naturally, the 'cyberwarriors' don't want to be stuck purely in the tedious and thankless job of playing defense, there is a demand for 'offensive capabilities'. This creates a perverse incentive: If a flaw is disclosed and patched or mitigated, it is no longer of offensive utility, so now the market for zero-days and exploit payloads isn't just black hats, scammers, and criminals; but 'respectable' defense industry types.
This is not a merely theoretical problem.
VUPEN is the crass, attention-whoring, bad-boy of the industry; but practically the entire who's-who of staid, tight-lipped, defense contractors has a division peddling bugs somewhere in the business.
Even if we were 100% warm and fuzzy about the use these exploits are being put to by these firms customers(Only the good guys, pinkie swear!), this situation is insane from the perspective of actual 'security'. Whose economies, financial systems, and infrastructure depend most heavily on complex IT systems? Ummm, mostly wealthy developed countries. Whose citizens are most vulnerable to electronic compromise of financial information and such? Countries with high rates of internet penetration and lots of computers. Who has the capability to deploy electronic attacks against unpatched vulnerabilities? Virtually everyone.
In addition to the usual grab for rights and money, this 'cybersecurity' industry begets insecurity, because of the demand for 'offensive capabilities', despite the fact that we are the ones with the most to lose in an insecure environment. At least classic corporate welfare military R&D is merely expensive, and once you hand over the money, Raytheon or whoever goes off to build some impractical toy that is largely useless; but at least largely harmless....
According to TFA, this 'social loot' nonsense requires some sort of affiliate ID baked in(presumably to the usual bit of gibberish at the end of the URL) for tracking the spamming performance of their little minions.
With any luck, this should allow automated recognition of people who are astroturfing for these guys and it's always good to have a new way of identifying awful people. At a service level, the astroturf can then be removed, downranked by search engines, etc. At a personal level, we can each do our part by reminding those culprits we know that spammers are abhuman scum who go to the special hell, and deserve it.
The PS/2 port only has one use; but it is Good.
IBM Model M.
Intel makes some decent, and damn cheap, atom boards(This model is among the cheapest). Unfortunately, those are a bit on the large side: normal ATX and P4 power connectors, and a full PCI slot are nice for compatibility; but they strongly suggest that the board hasn't been tightened up nearly as much as it could be. For whatever reason, Intel has left the yet-smaller niches to VIA, who still manage to charge north of $200 for their ultracompact x86 boards, despite tepid performance. I'm not quite sure why Intel hasn't spun an Atom board of the same size and priced it $50 lower; but they don't seem to have done so.
Then there are the good, old fashioned, PC/104 boards. They tend not to be the cheapest, and a fair amount of the I/O might be on oddball headers; but they are available in a fair assortment of x86 flavors...
Plan 9's explicitly multi-node design would be very handy in such a situation. It wouldn't actually be a terribly good interface; but who could resist having every device in their house at their beck and call in /dev where they belong?
I don't know exactly how the marketshare numbers break down; but x86 compatibility is big for 'embedded' projects that are relatively short production run but not terribly power constrained(like digital signage). In smallish quantities, the development time savings often dominate the hardware cost difference.
Also, customers are likely to expect to be able to do things like 'yeah, just put this powerpoint I made up on the screen' or 'Here's the flash intro our brand consultant put together'... Much easier when you can just slap a copy of Windows Embedded whatever and a hardware watchdog to reset it if needed than to whip up something that can digest crap like that into some more neutral format...
It will actually be interesting to see who cracks it, if any.
Apple is way better at imposing their vision on somewhat fragmented and mediocre ecosystems and, by brutally culling the unfit even among its own, establishing itself as the center of the new ecosystem. However, there is a certain amount of built-in 'does not play well with others' required by this strategy.
Microsoft, by contrast, has never been known for their nimble elegance; but their bread and butter, corporate and home, is cobbling things together, with a strong emphasis on not stepping on legacy customer toes, accommodating all kinds of shit from the hardware OEMs, and collaborating with those who they cannot crush, rather than just ignoring them.
If 'home automation' ends up being a relatively small number of devices per home, with limited grasp into the uglier and less-often-replaced corners, Apple should handily out-shiny Microsoft. If, on the other hand, it involves playing a long grind, licensing from existing vendors, putting up with 'smart' devices that ship with nothing but binary drivers, and customers crying if the support cycle isn't 15 years long, Microsoft might well be able to outfight them. MS may not like it; but they certainly know how to coddle legacy customers and systems. Apple can barely contain their desire to crush them and move on to the new hotness.
You could ask your eardrums how well moving things with sound works in a less dense fluid...
I suspect that the mathematical trickery required to get sound waves to push an object in a concerted way, rather than just bouncing around chaotically, was a fair trick; but (normally trivial) moving things is what makes sound sound like sound.
Here it is, back in 2009...
.NET applications on the host computer could be presented with an interface by which they could nominate bits of themselves to be executed on the coprocessor's CLR while the main computer is shut down, rather than having an entirely different program, running on a different architecture and OS, bundled in for the purpose. Not Microsoft specific, of course, Java could be used to similar ends, and some of the 'smart dust' and mesh guys probably have things designed, rather than kludged, for the purpose; but still...
Cute concept, really: a gumstix board with a chunk of flash running reduced complexity variants of certain network-interactive processes that the host computer was running(bittorrent and AIM in the mockup). When the host went to sleep, the gumstix transparently hijacked the host's MAC and continued doing things like providing an AIM status and dumping incoming bittorrent data to its flash card. If something happened that required the host PC's attention(ie. flash card filled up), the gumstix would wake the host, copy any data from when the host was asleep that it would need to know about, and kicked control of network activity back to the host.
At present, I don't think Microsoft has done anything with the idea, despite it being a fairly logical extension of Windows Sideshow, which appears to have been left to wither and die, instead.
It's funny, actually: in this case, MS Research hacked something together with a linux device because linux devices are a great platform for hacking stuff together; but this is probably an area where an in-house-technologies implementation might actually have been a great deal more elegant: had the little appendage-processor been running a CLR VM,
I'd be fairly surprised if this exact product ever sees the light of retail, intel kicks out prototypes of stuff all the time; but it would entirely fail to surprise me if some substantial percentage of boring-business 'ultra-small-form-factor' systems and 'compact' consumer PCs are running variously OEMed trivial variants of the design soon enough...
If the price is on par with raspberry or just above it this is going to be awesome. 293479x times the power.
Not happening, unfortunately. At retail(newegg.com used, prices for CPUs tend to be pretty similar across the board at a given time) the cheapest LGA1155 CPU is ~$40. 1.6GHz, single core, desktop binned part(unfortunately, low-end mobile CPUs don't seem to be as available in the retail channel, so I couldn't find a number for something in the mobile TDP range). At least it comes with a fan. Now, even such a puny device will brutalize a 700mhz ARM SoC designed to run from whatever battery is slim enough to fit in a contemporary cellphone; but if the CPU alone costs $5 more than the entire rpi, CPU+motherboard is going to run at least double, and RAM and boot volume still haven't been taken care of.
An overwhelmingly more powerful platform, certainly, as one would expect in a PC vs. basically-a-cellphone matchup; but the price delta is about what one would expect as well...
Yo Dog, I herd you like mac minis, so I spent the past seven years hitting a mini with the ugly stick and then released it as some sort of revolutionary device... Seriously.
Just for giggles, I then compared it to an entirely different device based around a smartphone processor and in an entirely different price bracket. This makes total sense, just trust me.
Now, purely in itself, a standardized teeny-ATX motherboard would be nice(especially if we'll someday be able to get mini-PCIe cards that aren't NICs in any quantity... If Intel is planning one, that seems like a good thing all around: the world is already cluttered with various proprietary teeny-motherboard things, and it'd be nice to have a bit of unification in that area.
However, I'm just not seeing the novelty here: The x86/embedded/industrial market has been rotten with teeny motherboards for almost as long as there has been an embedded x86 market, most laptops are built around small x86 motherboards by necessity, and some comparatively niche players, along with Apple, have released desktop products of not dissimilar size already. Historically, they've been fairly expensive, since minaturization isn't free, and Intel has no reason to cut margins on their silicon if they can avoid it. If Chipzilla has decided to drop the hammer and specify where teeny motherboards Shall put their screw holes, great; but that would be about the only new aspect of all this...
The really ugly problem with 'Home Automation' doesn't seem to be purely technological(ie. PIRs, contact switches, reed switchs, and relays are antedeluvian, adequate-speed interconnect over assorted wires also ancient, over RF also pretty old, and computers capable of crunching rulesets based on some combination of sensor outputs and RTCs are 80s stuff); but in the fact that the low-hanging fruit is too boring to be worth the trouble and the higher-hanging fruit would be massively complex and require a continual morass of several-industries-wide cooperation...
Boring case, you can drop a relay on each of your light switches, get your thermostat under control, and put some PIR presence detectors in place. If security is a concern, put up some cameras and stash a DVR somewhere. All easy; but not as inexpensive as one might like(especially in jurisdictions where touching mains current means bringing in electricians and permits and stuff), and you really have to want your bitchin' home theatre system to automatically close the drapes and dim the lights, or be very prone to leaving the stove on to fork over the additional money for the ability to remote control those things(and we are all familiar with the office comedy that is frantically waving your arms to make the 'smart' lights turn back on...)
Now, interesting case is where you expose every detail of all the devices in the home, quite possibly adding more sensors depending on the device category, and start having them cooperate intelligently to achieve various objectives. However, this is where the complexity gets ugly. Consider the history of ACPI: They wanted a way to allow computers to be more intelligent about power use, peripheral idle states, and environmental monitoring. Despite being hammered out in a near-duopoly environment(with MS on the software side and Intel on hardware), ACPI was a screaming pile of shit that barely worked properly even on mainstream OEM wintels until comparatively recently(and, even today, there are vendor-specific quirk packages, messy hacks, and peripherals that don't play quite right), never mind the poor bastards who dabbled in DIY or alternate OSes.
Now, you want the Home Of The Future? See to it that your utility meters, consumer appliances, home entertainment electronics, computers, water heaters, plumbing fixtures, climate control and thermostat systems, entry detection and security systems, and who knows what else all expose their sensors and capabilities in a standardized way. Don't let the fact that most of the items on that list have one or more industry consortia squabbling about the details of how their own little fiefdoms will be semi-standardized within themselves, much less on a broader basis, worry you.
Once you have the data to look at and the buttons and knobs to fiddle with, you just need some rulesets that make the devices collaborate intelligently and a set of interfaces that expose the power, but hide most of the complexity, to a degree sufficient that the fancy new hotness is actually worth the trouble.
Basically, you've got a problem whose complexity is fairly similar in scale to the sort of thing that smallish networking/datacenter entities would have an SNMP jocky on hand for, except that none of the hardware actually has management support yet, and the end result has to be easy enough for Joe Consumer to use. Also, it should ideally not be a dystopian surveillance nightmare or a script-kiddie playground. Not an easy problem....
It is my understanding that Microsoft Research does more or less whatever the hell they want. Very occasionally, their stuff ends up in actual Microsoft products, at which point it relentlessly marches in lockstep with the needs of Windows and Office once again. Heck, some time ago we were discussing an MS research project that used a gumstix module, running linux(not even the WinCE port that is available!) for some sort of low-power quasi-persistent network connection scheme...
It is my understanding that making a free trade agreement with the US is just the diplomatic equivalent of kissing the Don's ring.
It is certainly possible that somebody on your side will receive assistance with some problems that they may have been having; but it's a show of fealty, not a contract of equals...
At least for the present generation, I'm pretty sure that all the Tesla boards are 1 GPU per card. Nvidia supports, and sellers offer, arrangements with a fair number of cards per node(including external enclosures for expansion of systems that can't accommodate all those cards internally, 4 cards in a 1U connected by a special PCIe cable seems to have replaced the previous toaster-shaped 'deskside' chassis, with 'deskside' now handled by motherboards with loads of PCIe x16 slots); but the two-GPUs/one-card arrangements only seem to crop up on the gaming side.
I don't know why exactly this is, whether routing traces for 6 GB of RAM, per GPU, is bad enough with only one on the board, whether it is a thermal thing, or whether customer demand is such that it is cheaper to produce only single-GPU cards, and multi-slot chassis for the heavy users, rather than multiple flavors of card with various chip populations...
According to the PCI-SIG(or at least their press flacks, actual standards are members only) revision 3 is compatible with both 2 and 1, and 2 is compatible with 1(excepting one minor hiccup where 2.1 increased the allowable PCIe x16 slot power draw compared to 2.0 and earlier, so there are 2.1 cards in the wild that are logically compatible with 2.0 and 1.0; but which will only function with auxiliary power).
Internet anecdote suggests that this glorious vision may or may not actually be 100% realized, your BIOS isn't our problem, go cry to your vendor, etc; with graphics cards being the most common offenders(probably both because they are the most common user-installed PCIe peripheral, and because most motherboards won't POST properly if they can't find a working video device, while all but the really dysfunctional ones can ignore missing or confused NICs and such)...
I don't have enough personal experience to give any odds, but the SIG says yes and some people say 'not in my case'....
Who's taking odds as to the possibility that these fancy little toys will end up inflicting enough casualties to cover a front-page photo spread through either accident or malice, in a situation where just leaving them out of the picture would have gone fine?
PCIe 3.0 is, at least in magical specifications-actually-working-as-planned land, backwards compatible with 2.0 and 1.0, though obviously only at the highest mutually available speed between the two devices.
Is this optimistic theory a horrible pack of lies in general, are Nvidia products specifically broken in this respect, or do the newer ones make assumptions about bus speed that cause them to underperform on PCIe 2.0 boards?
The last company to get all "multiple core happy" and "SLI On A Board" happy was 3dfx. Who NVidia bought out when they... oh yeah, crash and burned.
Whoops.
I'm pretty sure that both ATI and Nvidia(or one of their OEM partners at the time) have kicked out a 'logically speaking, this is two cards in SLI/Crossfire; but on one card!!!' product for basically every generation since the introduction of the respective GPU-linking features.
The hard part is the fancy tricks that make cooperation between two separate GPUs work at all. Once the vendors decided that they did, in fact, want that to work, the rest is constrained largely by the fact that people willing to pay $1k for a graphics card aren't all that common(especially now that motherboards with 4 PCIe 16x slots are quite reasonably available, so even one's excessive desires can be satisfied with cheaper, more common, single chip cards).
That was my intended point: Allow(and, where necessary, legally facilitate) free parallel importation('free' as in 'subject to the same import standards as anything else, regardless of whether the manufacturer likes it or not, not as in 'all laws abolished') and watch arbitrage annihilate any excessive price differentials in short order.
Creating, and then attempting to restrain, monopoly entities is a very, very, tricky problem. On occasion, it cannot be avoided(electrical transmission systems, water piping, etc.); but parallel imports are a case where it would be trivial to allow anybody who so wishes to import goods legally purchased elsewhere, subject to no restriction beyond the import duties and standards that apply to all. Especially as Australia is an English language market, that should crush most attempts to price-discriminate pretty robustly.
Instead of this ad-hoc 'inquiry' nonsense, which is necessarily reactive and highly liable to regulatory capture, why don't we just adopt some of that 'free trade' stuff that assorted Respectable People tell us is so salubrious when the chaps who produce the products in question are shopping around for the cheapest inputs?
Absent legal barriers, arbitrage in software should cost next to nothing, especially now that much of it doesn't even come on shiny disks anymore. See to it that Australian customers can legally import goods from the location of their choice, and that middlemen can import goods from the location of their choice for domestic sale, and the price difference should collapse in a loud puff of nebulous whining about 'intellectual property'...
The whole notion behind the term 'grey market' is pernicious. It Should Not Matter whether the manufacturer/seller of a good is pleased by the ultimate destination of the goods they are selling. Yes, we would all like to enjoy perfect price discrimination. No, that isn't a good argument for letting us do so. In the absence of absurd restrictions on arbitrage, various pricing shenanigans, release-date bullshit, and other nonsense simply collapse.
Such restrictions would be one thing if they were applied evenhandedly, if the producers weren't already shopping all over the world for the lowest prices, laxest laws, and sweetest tax breaks; but they are not. You want cozy protectionism for your retail prices? Well, perhaps you shouldn't expect to enjoy worldwide free trade on your input prices... You want worldwide free trade for the things you buy? Well, that's nice, you deserve no less than worldwide free trade in the things you sell.
It is a rather curious idea, though I suspect that, in practice, people will end up cheating by throwing 'coprocessors' into the mix...
Think of, say, virtually all of the media set top box SoCs of the world. Usually ARM or MIPs(nominally); but that is just a weedy little core that draws a few UI elements and does OS housekeeping. The meat is inevitably some DSP, GPU, or video decode unit that may or may not even have a publicly known instruction set beyond the basic 'shove compressed video in like so, obtain frames and audio stream over here'.
Even the venerable x86 isn't so x86 anymore the moment you start shoving the 3D stuff reasonably seriously, at which point things kick over to a GPU of some flavor that is very much a different beast.
It also seems unlikely that they'll manage to eliminate the 'Microcontroller lurking somewhere in there to handle a bunch of very low speed GPIO with minimal expenditure of pins on the main CPU' phenomenon. 8051s and PICs and stuff turn up in the oddest places, sometimes visible, sometimes an IP core embedded in some peripheral package. Unless the non-techies have a nigh-religious zeal for one ISA to rule them all, and rules-lawyer the process very tightly, all they'll really insure(above and beyond what the market would have delivered anyway) is that there is only one ISA, rather than 3-ish, running the 'main' OS and passing data between as many bits of hardware speaking all sorts of curious languages as the design happens to require...