Well, indeed you can upgrade the GPU, but the GeForce 8200 is hardly inadequate for a TVPC, its fully blu ray capable, in fact its relatively hard to find any kind of *downloaded* video file that it can't play and offload from the CPU. Its really only unoptimized stuff (like Adobe flash) where the CPU goes to work.
I'm from Neuros and as others have suggested, we'd LOVE to use an atom solution. The issue, as fuzzyfuzzyfungus suggested is that we can't anticipate what users will want to use the machine for, and sadly much of that is not optimized. flash 10.1 did not completely solve this problem (and its not available in 64bit yet anyway). Further, flash isn't the only inefficient application out there, and the entire point of our box is flexibility, and that you can run virtually any application you want without hassle. I would love to do what you are doing and go with an atom solution, it would save us money and hassle, and be more efficient. Its very close, but just not quiet there for what we're trying to do just yet. No one will be happier when that day comes.
I'm from Neuros (to get that out of the way)
You shouldn't lump ball bearings in with fluid bearings. Fluid bearings combined the long life of ball bearings are are practically silent. But you are right about going big and slow. That's why the product uses a 120mm fan that's speed controlled, in typical use its under 1000 rpm and pretty much dead silent.
That's been my experience too, that the dB specs are meaningless because they are so filled with lies. This device is about as quiet as any fanned system gets and in any normal ambient environment you can't tell if its off or on from 3 feet away. You'd all but need to be in a sound proof room to hear it from 6 feet away.
Also, if we're talking high end processors, we aren't talking about this particular device
A 2.8GHz single core is a pretty careful choice. It's a pretty good balance that supports a wide range of content, remember not everything supports multiple cores well (or hardware acceleration for example). This processor does everything up to 1080p24 (what you see on apple.com for example) and also supports flash video, etc. On one hand, there's a great deal of discussion of ION or other graphics centric solutions, which are great when that hardware matches *exactly* what you want to playback, but then try something not optimized (flash for example) and you are very limited. On the other side, there are more powerful CPUs, but the expense (and cooling requirements-noise go up) and in most applications you won't see one lick of improvement, I know because we tested a lot of them before settling on this one.
yep, sure have. It's funny just today in the mail I got a big envelope from a law firm I didn't know, and I put it aside just figuring it was another letter from a patent troll demanding royalties. That's just a routine part of being an electronics manufacturer nowadays. It turned out it was something else for once, but generally big envelope from a law firm you don't know mean patent troll nowadays.
The parent post does a very good job of posing the problem that the NeurosLink is trying to solve (with the exception of composite output which we've left for dead). The traditional embedded devices are more plug and play and a PC is vastly more flexible, how do you combine the benefits of both?
The problems with a HTPC being plug and play are not at the high level. As many posts in this discussion demonstrate, many people have an easy enough time installing XBMC on Linux and the machine works well at the basic level. Its the details where a big amount of effort is needed. Make sure audio over HDMI works. Make sure the remote works and all the buttons are mapped correctly. Make sure you can configure the system entirely with the remote (at a minimum get it on the network). Make sure all videos play flawlessly out of the box, not just downloaded MKV files, but Hulu HD and all the other flash video sites. These are the kinds of little details that its really helpful to have a manufacturer behind. Someone has got to hammer away at the "final 5%" that's required for it to be competitive with any dedicated device.
The good news is that once you get those details taken care of you left with a really powerful, robust flexible system. Obviously many talented folks have been hammering away at the various components for years (XBMC, MythTV, etc) so its really glue that's left to be done.
Gamma is a term we coined to refer to production white box hardware with early software (alpha level integration in this case). See the Neuros Gamma Wiki Page
I have a Xbox running XBMC and it's amazing for less than $100. Saying "XBMC is better" then going on to conclude that the $300 Link is "an exceptional value" doesn't make sense. His conclusion does not match his observations.
Well, this combines a HW and SW comparison. Wrt to HW, a modded second hand X Box is an unbeatable value for standard def, period. It was subsidized HW and can't be beat (particuliarly at $100). If you can get past the modding headaches and SD limitations, you won't find a better value. In fact, if you need component and composite connectors, the LINK doesn't support that at all.
That being said, the LINK is many years newer hardware. The processor is 2.8 GHz 64 bit v Xbox's 733 MHz (IIRC) etc. and the LINK has HDMI, HD GPU, etc.
Wrt to SW, the LINK is evolving from its starting point of basically stock Ubuntu with a collection of apps (including XBMC and Boxee) to a more seamless, integrated experience. The app software is all GPL, so parts of many apps will be integrated over time, there is still lots of experimentation ongoing (and getting community input during this period is precisely why we launched to users early)
A box like this has to compete with AppleTV/Xbox on the lowend and
mini pc's like the Mac Mini on the highend.
exactly right, and I believe there's a place in the middle. A device with comprehensive playback capabilities (both downloaded content and web video) that operates like a piece of electronics. That's the vision, as many, including the reviewer point out, we're not there yet. We started from the PC side and are evolving to be more electronics like, both software and eventually more stripped down hardware. Our focus since release (and until production- remember this is a gamma product) will be on enhancing the UI for couch use. And, yes it will be a period of experimentation, XBMC, Boxee, Miro all work on the box and are in use by various members of the community.
We (neuros) feel this is the sweet spot. No, it's not as cheap (or small) as a straight AppleTV (or the like), and the web interface isn't as couch friendly in the navigation. but its more open, expandable, free standing and comprehensive in terms of access to content and we are working on the shortcomings.
Plus there are ample
possibilities for home built machines with integrated chipsets that
support full h264 acceleration.
At this point, to a certain extent, you can really view this as a home built effort. It's not home built in the sense you have complete access to all the components, but honestly if there's something you want to customize about it, you can literally buy the parts yourself, or we'll sell you a sub assembly and we've listed all the components line by line on the wiki.
The point is that there are economies of scale of a community working in collaboration directly with the manufacturer on a focused set of hardware. A good example of this is ATI, as mentioned elsewhere, we are working directly with ATI as a customer to solve the issues with the system. I think it stands to reason that we have more influence with ATI as a group than as individual hackers, and in fact you will find ATI/AMD engineers participating directly in our community.
As anyone that's put together one of these systems knows, there are tons of minor details that need to be worked out, and it's vastly more efficient if we join forces on a focused application on a defined hardware platform. That's why we've made sure that engineers at Boxee, XBMC, Miro, etc all have sample hardware.
As mentioned elsewhere the hardware does have direct h264 acceleration and is 1080p capable, athough this has not been full implemented in Linux yet.
Unless it can play bluray rips, $300 isn't a very compelling
price for the box. It's more expensive than other CD options and
not as capable of the more expensive HD capable alternatives.
ATI is simply the wrong direction to go for a box like this.
the ATI hardware is capable, and believe me, I share the frustration wrt ATI. But understand that we came, like most here, biased against ATI at the outset (and we have ulterior motive or connection with ATI). Despite that, they were able to convince us, both with the offers of support for the project as well as current features that they were the right vendor for this device. This is something we continue to evaluate, but if you do the detailed analysis, there are issues with Linux support for both ATI and nVidia, and either vendor must evolve their Linux performance. The good news is that the playing field for Linux, unlike the desktop, is *vastly * more level on the set-top box side than it is on the desktop. More accurately, slanted to Linux's favor, so I believe this will drive a lot of effort. Obviously the success or failure of this project and those like it will be important too.
Is the ATI card in it one that has ATI's vdvpu equivlent? then i'd be much more inclined to belive that it will work.
the ATI interface is evolving, we (Neuros) is working with them directly to evolve the Linux drivers (both proprietary and open). Today it doesn't yet have the vdpau type interface.
However, ATI specs that this is a full bluray capable card, so it's in the software interfaces. Without that we're currently at the cusp of 1080p (24 fps like the apple.com trailers) but there's clearly improvements coming
I realize that ATI doesn't have the best FOSS reputation, but when we looked at nVidia, ATI had some advantages, particularly on the HDMI side (which nVidia doesn't well support yet) as well as a lot of support for this project, which is obviously important.
To really cut the price, you'd need to go with some non-x86 embedded setup. Trouble is, that would increase the software development costs and (much worse) would mean that you'd need to hassle a whole lot of "content providers" to get things working with their services. x86+flash, by contrast, is pretty broadly supported without special agreements.
you got it exactly right, and in fact, traditionally we've done all embedded HW, and we could produce a proprietary solution in quantity of 100 if we wanted. It's all about the benefit of the above.
And let me expand upon the above and explain why x86 makes fundamental sense for a device like this (at least for the forseeable future). What you outline above portays it from our (the mfg's perspective) but let's look at it from the user's perspective, because that's where it really comes alive.
When you say "special agreements," let's think about the ramifications for you as a user. Let's also recognize that it's not just agreements, it's technical integration effort too. So all of a sudden, a new Hulu launches tomorrow, and it's virtually guaranteed that it will, right. Well, first off no embedded device yet supports flash 10, so you are out there. Eventually these device will support flash 10, but how well, flash 10 that includes HD video, because if not, well again, you might be left in the cold. What if its some other plug-in (which often is it) well, then, in many cases, the integration hasn't even started.
The net effect is that you find yourself in the current situation. It's great that there's a $100 box that plays youtube and netflix, but what about nbc.com? what about youtube HD? what about stumbleupon video or dig video, etc etc. Sure, in theory they can all ultimately support all these content provides but a) that doesn't do you much good today and b)how many new content providers will have appeared by then?
Until there is broad agreement on a standard for video (and not just playback but browsing and navigation, etc) then x86 as flawed and expensive as it is is only choice for comprehensive playback.
Another note, at this point, it's not enough to just have x86, you need a powerful CPU. Again, this limitation will surely go away eventually, but for the time being, there is sadly lots of software that runs only on the CPU. Boxee did a great job of porting to the AppleTV, but ask them about the limitations of the ~1 GHz x86 CPU, which prevented Hulu HD from playing, etc.
Don't get me wrong, this stuff will change eventually, and Neuros will be delighted to go back to embedded to embrace it will smaller, cheaper hardware, but for now, the lack of a set of standards forced us to a PC architecture to provide comprehensive access to content.
I'm from Neuros for those that don't already know, I'll answer the x86 v embedded elsewhere.
Regarding economies of scale, let's look at it a different way. As many of you know, Neuros history is in embedded systems from scratch. This is our first x86 project as well as the first using off the shelf components.
We actually did it because of the economies of scale advantage. The PC industry has such an established supply chain, with such huge volumes that its creates a huge advantage. You have to live with x86 plusses and minuses, which, again I'll address separately. But if you favor x86, which obviously we do for this currently. Sticking with off the shelf makes sense (at least to start).
As posted elsewhere, this MB *at retail* is $80 that's really quiet a value for that piece of hardware. Look at the GPU, and the host of peripherals and expansion. Firewire, tons of USB, s/pdif, 7.1 audio, HDMI, DVI, VGA, PCI, SATA, etc. Sure, much of that is worthless for a mainstream wal-mart product, but not for a Gamma launch to folks that want to experiment and play around. Likewise with the case, it's nice having size for a internal 3.5" drive, and/or an optical drive. The case could be smaller certainly, but today, it's the size of a TiVo, so it's not an unreasonable compromise IMHO.
Of course, as the product matures, we'll cut out many peripherals, and put it in a smaller case, certainly there will be cost savings, but some not immaterial tradeoffs. It wouldn't surprise me if we continued to sell this unit along side it. I suspect many would continue to prefer it.
Best guess is that the savings will be on the order of $50 for some of the customization above. Important no doubt, but not everyone would want those tradeoffs, particularly at a stage where we're specifically targeting hackers.
don't forget you've got a wifi card, wireless keyboard (and you want one with an integrated trackball of touchpad) flash drive and, as you mention, case.
I have a kind of unique perspective on this, we're actually building a discless Ubuntu TVPC box for mass production and its clear that getting it to "just work" at least in all situations does take some work. We use a new chipset (AMD 780g) because it supports audio over HDMI (note that many of the comments on here quietly note that they are using DVI with analog audio of some kind) nVidia's drivers still don't support this. We're actually working directly with ATI to make sure that the graphics chipset resumes properly from sleep, that it auto-senses the display properly, etc. For many on here, those kind of hiccups are no big deal, but when it comes to the WAF or making a real mainstream product, there is a lot of little details that need fixing. Take my word, there is a lot of tweaking needed to truly have something "just work" and bear in mind that we're dealing with production runs of perfectly identical hardware, so the problem goes up dramatically if you are piecing together a variety of components in a DIY way, although I suppose you have more options of tweaking during install than we have for a product sold to consumers at retail.
Boxee is basically three things:
1. XBMC-strong playback of downloaded content, rich media browsing, automatic population of cover art, etc
2. "remote controlled" internet tv- it adds plug-ins that XBMC didn't have (some have apparently since been backported) for viewing hulu in an "up down left right" interface that eliminates the need for a mouse.
3. social features- see what your friends are watching, twitter what you are watching, share, etc.
I don't know plex, but neither XBMC nor Boxee is DVR software, for that you want mythtv or something like that. Myth has a programming guide built (have to signup for the service though) works with the HD homerun, etc.
No reason you can't run both on one box, that's what we do on the Neuros LINK (which is essentially a discless TVPC). It's not as seamless as a single app that does everything, but its a good solution today since there isn't a single app that does all that well yet.
The Neuros LINK is essentially a $299 thin TVPC,discless (runs off flash storage) designed for playback (including running Boxee) comes with ubuntu 8.10, wifi, already tweaked and running with wireless controller, expandable with PCI, SATA, etc.
http://www.neurostechnology.com/neuros-link
The GPL is being satisfied, but what's significant about this announcement is that Neuros and M2X are going a step beyond to make sure that code is being contributed back upstream, which very few device manufacturers do. Typically, modifications are made available (through some means as parent sites) but no effort is made to bring those patches back to the central repo, so many of the patches are effectively lost.
All the video codec's are implemented in dsp's, all the audio is done on the arm side. TI has, with a kind of subcontracting, libraries for every codec available.
They will only use the framework & rtsp/etc implementations. The hard stuff will be done in their own lib's (for which they pay the license fees) That's right, and furthermore, the bulk of the licensing issues is about patents, not copyrights, so generally, you pay the patent holders and use code from whatever source you want, and they don't care.
Luckily VLC decided to drop wxWidgets entirely (which they say was causing a lot of issues) and rewrite the UI in Qt4 for the upcoming version. It's not perfect, but it's already a big step up. The Neuros device will use its own UI based on Qt4 FWIW. Remember the device will be a TV + remote control living room type device, so it's expected to be full screen and remote navigable, so the UI will be quite different from the PC version.
\Maybe TI money will provide them with the developer resources to actually implement a ASS/SSA renderer. Neuros and TI are good candidates for funding this kind of stuff since Neuros and TI make money on every device sold, regardless of the type of software. It's to TI's benefit to have VLC working well on their silicon, since enlarging the body of open software that runs on their silicon makes it easier to get TI based devices to market. It's good news for any potential customer trying to get a device to market, and that's good for TI.
Well, it was a deliberate dig, and perhaps not totally appropriate, but I think the real point of the article is about a mainstream perception, and to that end, I think it is related to the topic. The point is that the subtleties known to the people reading this discussion are often lost on the average user, but that doesn't mean that they don't care at all. More often what happens is that a lot of the details get aggregated into a more understandable general concept. That's why the term, "openness" is really a more useful term because it describes a host of user level attributes: allowing third party applications, using standard file formats, encouraging alternative uses of the device on a variety of levels. Slashdotters don't like this because it's too vague and allows marketing people to wreak havock with terms that should seemingly be precise, but I think that ignores the fact that even a fairly involved, but non technical user simply doesn't have the appetite for distinctions that seem like splitting hairs to them. To win the war or words, you need sound bites and easily digestible concepts. The Neuros OSD is flexible and allows different applications, consider that an artifact of it being "open." Many other products are closed, and as a result can't realistically do anything not intended by the manufacturer, and it's often not interoperable. Sure, drill a level down and there's a precise definition as to what's open and what's closed and what's open source or just an open API or just open Standards or just DRM free media, but don't hit the average user over the head with those details with the elevator pitch so to speak.
Slashdot Mirror
Well, indeed you can upgrade the GPU, but the GeForce 8200 is hardly inadequate for a TVPC, its fully blu ray capable, in fact its relatively hard to find any kind of *downloaded* video file that it can't play and offload from the CPU. Its really only unoptimized stuff (like Adobe flash) where the CPU goes to work.
I'm from Neuros and as others have suggested, we'd LOVE to use an atom solution. The issue, as fuzzyfuzzyfungus suggested is that we can't anticipate what users will want to use the machine for, and sadly much of that is not optimized. flash 10.1 did not completely solve this problem (and its not available in 64bit yet anyway). Further, flash isn't the only inefficient application out there, and the entire point of our box is flexibility, and that you can run virtually any application you want without hassle. I would love to do what you are doing and go with an atom solution, it would save us money and hassle, and be more efficient. Its very close, but just not quiet there for what we're trying to do just yet. No one will be happier when that day comes.
I'm from Neuros (to get that out of the way) You shouldn't lump ball bearings in with fluid bearings. Fluid bearings combined the long life of ball bearings are are practically silent. But you are right about going big and slow. That's why the product uses a 120mm fan that's speed controlled, in typical use its under 1000 rpm and pretty much dead silent.
Agreed, may flash die the miserable death that is so long overdue.
That's been my experience too, that the dB specs are meaningless because they are so filled with lies. This device is about as quiet as any fanned system gets and in any normal ambient environment you can't tell if its off or on from 3 feet away. You'd all but need to be in a sound proof room to hear it from 6 feet away.
Here's a review of the device with pictures, etc: http://www.geardiary.com/2010/04/10/review-neuros-link/
Also, if we're talking high end processors, we aren't talking about this particular device
A 2.8GHz single core is a pretty careful choice. It's a pretty good balance that supports a wide range of content, remember not everything supports multiple cores well (or hardware acceleration for example). This processor does everything up to 1080p24 (what you see on apple.com for example) and also supports flash video, etc. On one hand, there's a great deal of discussion of ION or other graphics centric solutions, which are great when that hardware matches *exactly* what you want to playback, but then try something not optimized (flash for example) and you are very limited. On the other side, there are more powerful CPUs, but the expense (and cooling requirements-noise go up) and in most applications you won't see one lick of improvement, I know because we tested a lot of them before settling on this one.
yep, sure have. It's funny just today in the mail I got a big envelope from a law firm I didn't know, and I put it aside just figuring it was another letter from a patent troll demanding royalties. That's just a routine part of being an electronics manufacturer nowadays. It turned out it was something else for once, but generally big envelope from a law firm you don't know mean patent troll nowadays.
The parent post does a very good job of posing the problem that the NeurosLink is trying to solve (with the exception of composite output which we've left for dead). The traditional embedded devices are more plug and play and a PC is vastly more flexible, how do you combine the benefits of both?
The problems with a HTPC being plug and play are not at the high level. As many posts in this discussion demonstrate, many people have an easy enough time installing XBMC on Linux and the machine works well at the basic level. Its the details where a big amount of effort is needed. Make sure audio over HDMI works. Make sure the remote works and all the buttons are mapped correctly. Make sure you can configure the system entirely with the remote (at a minimum get it on the network). Make sure all videos play flawlessly out of the box, not just downloaded MKV files, but Hulu HD and all the other flash video sites. These are the kinds of little details that its really helpful to have a manufacturer behind. Someone has got to hammer away at the "final 5%" that's required for it to be competitive with any dedicated device.
The good news is that once you get those details taken care of you left with a really powerful, robust flexible system. Obviously many talented folks have been hammering away at the various components for years (XBMC, MythTV, etc) so its really glue that's left to be done.
Gamma is a term we coined to refer to production white box hardware with early software (alpha level integration in this case). See the Neuros Gamma Wiki Page
I have a Xbox running XBMC and it's amazing for less than $100. Saying "XBMC is better" then going on to conclude that the $300 Link is "an exceptional value" doesn't make sense. His conclusion does not match his observations.
Well, this combines a HW and SW comparison. Wrt to HW, a modded second hand X Box is an unbeatable value for standard def, period. It was subsidized HW and can't be beat (particuliarly at $100). If you can get past the modding headaches and SD limitations, you won't find a better value. In fact, if you need component and composite connectors, the LINK doesn't support that at all.
That being said, the LINK is many years newer hardware. The processor is 2.8 GHz 64 bit v Xbox's 733 MHz (IIRC) etc. and the LINK has HDMI, HD GPU, etc.
Wrt to SW, the LINK is evolving from its starting point of basically stock Ubuntu with a collection of apps (including XBMC and Boxee) to a more seamless, integrated experience. The app software is all GPL, so parts of many apps will be integrated over time, there is still lots of experimentation ongoing (and getting community input during this period is precisely why we launched to users early)
Joe (from Neuros)
A box like this has to compete with AppleTV/Xbox on the lowend and mini pc's like the Mac Mini on the highend.
exactly right, and I believe there's a place in the middle. A device with comprehensive playback capabilities (both downloaded content and web video) that operates like a piece of electronics. That's the vision, as many, including the reviewer point out, we're not there yet. We started from the PC side and are evolving to be more electronics like, both software and eventually more stripped down hardware. Our focus since release (and until production- remember this is a gamma product) will be on enhancing the UI for couch use. And, yes it will be a period of experimentation, XBMC, Boxee, Miro all work on the box and are in use by various members of the community. We (neuros) feel this is the sweet spot. No, it's not as cheap (or small) as a straight AppleTV (or the like), and the web interface isn't as couch friendly in the navigation. but its more open, expandable, free standing and comprehensive in terms of access to content and we are working on the shortcomings.
Plus there are ample possibilities for home built machines with integrated chipsets that support full h264 acceleration.
At this point, to a certain extent, you can really view this as a home built effort. It's not home built in the sense you have complete access to all the components, but honestly if there's something you want to customize about it, you can literally buy the parts yourself, or we'll sell you a sub assembly and we've listed all the components line by line on the wiki.
The point is that there are economies of scale of a community working in collaboration directly with the manufacturer on a focused set of hardware. A good example of this is ATI, as mentioned elsewhere, we are working directly with ATI as a customer to solve the issues with the system. I think it stands to reason that we have more influence with ATI as a group than as individual hackers, and in fact you will find ATI/AMD engineers participating directly in our community.
As anyone that's put together one of these systems knows, there are tons of minor details that need to be worked out, and it's vastly more efficient if we join forces on a focused application on a defined hardware platform. That's why we've made sure that engineers at Boxee, XBMC, Miro, etc all have sample hardware.
As mentioned elsewhere the hardware does have direct h264 acceleration and is 1080p capable, athough this has not been full implemented in Linux yet.
Unless it can play bluray rips, $300 isn't a very compelling price for the box. It's more expensive than other CD options and not as capable of the more expensive HD capable alternatives.
ATI is simply the wrong direction to go for a box like this.
the ATI hardware is capable, and believe me, I share the frustration wrt ATI. But understand that we came, like most here, biased against ATI at the outset (and we have ulterior motive or connection with ATI). Despite that, they were able to convince us, both with the offers of support for the project as well as current features that they were the right vendor for this device. This is something we continue to evaluate, but if you do the detailed analysis, there are issues with Linux support for both ATI and nVidia, and either vendor must evolve their Linux performance. The good news is that the playing field for Linux, unlike the desktop, is *vastly * more level on the set-top box side than it is on the desktop. More accurately, slanted to Linux's favor, so I believe this will drive a lot of effort. Obviously the success or failure of this project and those like it will be important too.
Is the ATI card in it one that has ATI's vdvpu equivlent? then i'd be much more inclined to belive that it will work.
the ATI interface is evolving, we (Neuros) is working with them directly to evolve the Linux drivers (both proprietary and open). Today it doesn't yet have the vdpau type interface.
However, ATI specs that this is a full bluray capable card, so it's in the software interfaces. Without that we're currently at the cusp of 1080p (24 fps like the apple.com trailers) but there's clearly improvements coming
I realize that ATI doesn't have the best FOSS reputation, but when we looked at nVidia, ATI had some advantages, particularly on the HDMI side (which nVidia doesn't well support yet) as well as a lot of support for this project, which is obviously important.
To really cut the price, you'd need to go with some non-x86 embedded setup. Trouble is, that would increase the software development costs and (much worse) would mean that you'd need to hassle a whole lot of "content providers" to get things working with their services. x86+flash, by contrast, is pretty broadly supported without special agreements.
you got it exactly right, and in fact, traditionally we've done all embedded HW, and we could produce a proprietary solution in quantity of 100 if we wanted. It's all about the benefit of the above.
And let me expand upon the above and explain why x86 makes fundamental sense for a device like this (at least for the forseeable future). What you outline above portays it from our (the mfg's perspective) but let's look at it from the user's perspective, because that's where it really comes alive.
When you say "special agreements," let's think about the ramifications for you as a user. Let's also recognize that it's not just agreements, it's technical integration effort too. So all of a sudden, a new Hulu launches tomorrow, and it's virtually guaranteed that it will, right. Well, first off no embedded device yet supports flash 10, so you are out there. Eventually these device will support flash 10, but how well, flash 10 that includes HD video, because if not, well again, you might be left in the cold. What if its some other plug-in (which often is it) well, then, in many cases, the integration hasn't even started.
The net effect is that you find yourself in the current situation. It's great that there's a $100 box that plays youtube and netflix, but what about nbc.com? what about youtube HD? what about stumbleupon video or dig video, etc etc. Sure, in theory they can all ultimately support all these content provides but a) that doesn't do you much good today and b)how many new content providers will have appeared by then?
Until there is broad agreement on a standard for video (and not just playback but browsing and navigation, etc) then x86 as flawed and expensive as it is is only choice for comprehensive playback.
Another note, at this point, it's not enough to just have x86, you need a powerful CPU. Again, this limitation will surely go away eventually, but for the time being, there is sadly lots of software that runs only on the CPU. Boxee did a great job of porting to the AppleTV, but ask them about the limitations of the ~1 GHz x86 CPU, which prevented Hulu HD from playing, etc.
Don't get me wrong, this stuff will change eventually, and Neuros will be delighted to go back to embedded to embrace it will smaller, cheaper hardware, but for now, the lack of a set of standards forced us to a PC architecture to provide comprehensive access to content.
I'm from Neuros for those that don't already know, I'll answer the x86 v embedded elsewhere. Regarding economies of scale, let's look at it a different way. As many of you know, Neuros history is in embedded systems from scratch. This is our first x86 project as well as the first using off the shelf components. We actually did it because of the economies of scale advantage. The PC industry has such an established supply chain, with such huge volumes that its creates a huge advantage. You have to live with x86 plusses and minuses, which, again I'll address separately. But if you favor x86, which obviously we do for this currently. Sticking with off the shelf makes sense (at least to start). As posted elsewhere, this MB *at retail* is $80 that's really quiet a value for that piece of hardware. Look at the GPU, and the host of peripherals and expansion. Firewire, tons of USB, s/pdif, 7.1 audio, HDMI, DVI, VGA, PCI, SATA, etc. Sure, much of that is worthless for a mainstream wal-mart product, but not for a Gamma launch to folks that want to experiment and play around. Likewise with the case, it's nice having size for a internal 3.5" drive, and/or an optical drive. The case could be smaller certainly, but today, it's the size of a TiVo, so it's not an unreasonable compromise IMHO. Of course, as the product matures, we'll cut out many peripherals, and put it in a smaller case, certainly there will be cost savings, but some not immaterial tradeoffs. It wouldn't surprise me if we continued to sell this unit along side it. I suspect many would continue to prefer it. Best guess is that the savings will be on the order of $50 for some of the customization above. Important no doubt, but not everyone would want those tradeoffs, particularly at a stage where we're specifically targeting hackers.
don't forget you've got a wifi card, wireless keyboard (and you want one with an integrated trackball of touchpad) flash drive and, as you mention, case.
I have a kind of unique perspective on this, we're actually building a discless Ubuntu TVPC box for mass production and its clear that getting it to "just work" at least in all situations does take some work. We use a new chipset (AMD 780g) because it supports audio over HDMI (note that many of the comments on here quietly note that they are using DVI with analog audio of some kind) nVidia's drivers still don't support this. We're actually working directly with ATI to make sure that the graphics chipset resumes properly from sleep, that it auto-senses the display properly, etc. For many on here, those kind of hiccups are no big deal, but when it comes to the WAF or making a real mainstream product, there is a lot of little details that need fixing. Take my word, there is a lot of tweaking needed to truly have something "just work" and bear in mind that we're dealing with production runs of perfectly identical hardware, so the problem goes up dramatically if you are piecing together a variety of components in a DIY way, although I suppose you have more options of tweaking during install than we have for a product sold to consumers at retail.
Boxee is basically three things: 1. XBMC-strong playback of downloaded content, rich media browsing, automatic population of cover art, etc 2. "remote controlled" internet tv- it adds plug-ins that XBMC didn't have (some have apparently since been backported) for viewing hulu in an "up down left right" interface that eliminates the need for a mouse. 3. social features- see what your friends are watching, twitter what you are watching, share, etc. I don't know plex, but neither XBMC nor Boxee is DVR software, for that you want mythtv or something like that. Myth has a programming guide built (have to signup for the service though) works with the HD homerun, etc. No reason you can't run both on one box, that's what we do on the Neuros LINK (which is essentially a discless TVPC). It's not as seamless as a single app that does everything, but its a good solution today since there isn't a single app that does all that well yet.
The Neuros LINK is essentially a $299 thin TVPC,discless (runs off flash storage) designed for playback (including running Boxee) comes with ubuntu 8.10, wifi, already tweaked and running with wireless controller, expandable with PCI, SATA, etc. http://www.neurostechnology.com/neuros-link
The GPL is being satisfied, but what's significant about this announcement is that Neuros and M2X are going a step beyond to make sure that code is being contributed back upstream, which very few device manufacturers do. Typically, modifications are made available (through some means as parent sites) but no effort is made to bring those patches back to the central repo, so many of the patches are effectively lost.
Neuros and TI are putting a lot of money to fund this and other open source development. see http://bounties.neurostechnology.com/
Well, it was a deliberate dig, and perhaps not totally appropriate, but I think the real point of the article is about a mainstream perception, and to that end, I think it is related to the topic. The point is that the subtleties known to the people reading this discussion are often lost on the average user, but that doesn't mean that they don't care at all. More often what happens is that a lot of the details get aggregated into a more understandable general concept. That's why the term, "openness" is really a more useful term because it describes a host of user level attributes: allowing third party applications, using standard file formats, encouraging alternative uses of the device on a variety of levels. Slashdotters don't like this because it's too vague and allows marketing people to wreak havock with terms that should seemingly be precise, but I think that ignores the fact that even a fairly involved, but non technical user simply doesn't have the appetite for distinctions that seem like splitting hairs to them. To win the war or words, you need sound bites and easily digestible concepts. The Neuros OSD is flexible and allows different applications, consider that an artifact of it being "open." Many other products are closed, and as a result can't realistically do anything not intended by the manufacturer, and it's often not interoperable. Sure, drill a level down and there's a precise definition as to what's open and what's closed and what's open source or just an open API or just open Standards or just DRM free media, but don't hit the average user over the head with those details with the elevator pitch so to speak.