I used Linux on the desktop for years myself, but eventually I decided to ditch it and just stick to Windows exactly for the reasons you mention: there was always something breaking and I could never get all the hardware-features working that worked fine under Windows. The desktop-environments themselves were also a big, sodding mess and for anything more than typing text in gedit or similar or browsing the web you had to drop to CLI, and it's still that way even to this day.
I'm definitely very happy with Linux as a server-OS on both my NAS and my router, but on the desktop? No, not happening; on the desktop I want pretty GUI-applications I can just go clicky-click around when I want something happening, I want all my hardware working, I want to play games, I want to be able to consume online-content, and I don't want an update to GRUB fucking up my boot so I have to dig out systemrescuecd yet again.
This is what I was coming to say. I know bashing Microsoft/Windows is trendy and all that, especially in this den of rabid, mouth-breathing troglodytes, but these Atom CPUs have a GPU made by Imagination Tech., not Intel, and it seems that is the culprit here. Acer did say they are working with Microsoft to try to get this fixed, too, so this should just be temporary.
I can't help but wonder if/how much it affects performance, though. There are plenty of cases where hyper-threading is a very welcome feature and some people might be upset, if there was a hit to performance. I don't own a Skylake - or Kaby Lake - product, but I am just curious of real-world implications.
The RPi isn't the only SBC with a GPIO-header with user-accessible I2C and SPI. There are about a billion different ARM-based SBCs like that around by now, and plenty of x86- and MIPS-based ones with available GPIO-headers, too. Of course it has "strength" over a board without such accessible interfaces, but that's just a stupid comparison to begin with, like comparing apples and airplanes, and it has no "strength" over any other board with those interfaces.
It's not a bug, it's a hardware-limitation. The SoC only provides a single USB-port and the only way of connecting a USB-Ethernet and providing 4 USB-ports for the end-user to use is through a USB-hub, which obviously will mean that they all share the 480Mbps bandwidth of the single USB-port the hub is connected to. The only way the RPi Foundation could fix that is... by switching to another SoC. They'd lose backwards-compatibility, though, so that's not going to happen anytime soon.
There is an open-source driver for the GPU nowadays, capable of doing OpenGL (not just OpenGL ES) and such, and there is work going on for developing an open-source blob for booting the boards, too. I haven't checked in a good while how the projects are faring, but the last time I tried the open-source GPU-driver, it worked surprisingly well already. These projects give hope for a pretty good long-term support as long as one doesn't need H/W-accelerated decoding/encoding (there is no open-source support for that atm., AFAIK) of video, and even that may still one day get an open-source implementation.
One of the problems with x86 SBCs is that they are pretty much solely based on Intel's offerings, which makes them quite a bit more expensive and therefore not as appealing to home-tinkerers and the likes. I suppose that is the primary reason for their apparent unpopularity. That said, I do like the promise of the x86-boards myself, what with proper QuickSync for all sorts video-needs, like e.g. realtime transcoding, and full OpenGL instead of OpenGL ES for any graphical applications and quite mature software-landscape in the big picture. They're just lacking in community-support and e.g. all the easy-to-follow tutorials and such for home-tinkerers to interface with all sorts of sensors and stuff. One can always hope things will pick up; the fact that Intel dropped their line of SBCs isn't necessarily an indication of the market at large, as Intel has a tendency of doing this kind of a thing, and there are several other players who manufacture x86 SBCs.
You do realize that there are vendors who do x86 SBCs other than Intel? It's totally not pointless to bring up x86 SBCs in a discussion about, you know, SBCs.
How do the different Pis have a "strength in communicating with an Arduino?" The x86- and MIPS-boards can do that just as well, there is nothing stopping one from communicating with an Arduino over I2C, SPI, serial or whatever even on the other platforms. And no, Pi definitely doesn't have a "strength" in Ethernet, either, considering it's just 10/100 and it's actually a USB-device and thus eats bandwidth from the USB-ports, all of which are internally connected to a single USB-hub on the PCB.
"has a good Privacy Policy, and doesn't have a history of breaches or allowing snooping." -- Runbox fits all of those. Norwegians have very strong laws regarding privacy, which should please you, and the company doesn't do any advertising or crawling through your emails for tracking or anything like that. It's not a free service nor is it the cheapest one available, but I've been their customer for several years and I would at least recommend one to take a look at their offerings.
I did mention they are quite expensive, but they do have prices listed in the shop quite clearly. Maybe you missed the two links I gave above, but there's their online-shop behind the first link with a selection of the various Up^2 - boards at http://up-shop.org/4-up-boards . They even sell a UPS-like expansion board for it on the shop, too, so one doesn't have to hack one together themselves necessarily -- though, at $89 it may not be the world's best value. I suppose it depends on one's capability of cobbling one together themselves.
If one was going the DIY-way, I'd rather recommend Up^2. It's an actual x86-board, so it can run all the usual x86-stuff, there's a proper mPCI-E - slot for mSATA- and/or NVMe-drives or whatever mPCI-E card you may want to use, an M.2 2230 E-key for real, proper WiFi-cards, a SATA-connector, 3x USB3.0 (and a couple USB2.0-ports via a pin-header), a lot, lot more capable GPU than Raspberry Pi's one, built-in eMMC (the top-end model has a 128GB one) and so on.
The thing is, an RPi makes for a really crappy desktop-experience. The Up^2 is significantly more expensive, but it's also significantly more capable and much better suited for desktop-use.
I actually really liked the first Resident Evil - movie, the second was ok, but the rest of them I just yawned and rolled my eyes through and now afterwards I just can't recall those movies at all -- it's like I had never even seen them in the first place, which I find quite funny. Now, I have not played the games, I do not have any connection to the lore or anything, so I only watched those movies as exactly that: movies about some mutating viral zombie-thingamabob. Seeing as how exceedingly quickly the movies hit the rock bottom quality-wise I have a hard time imagining any reboots can get any worse; either they'll stay as bad, or get better.
A standard router is better optimized H/W-wise for these tasks, just get one that is supported by OpenWRT and/or LEDE. OpenWRT/LEDE gives you SSH and everything else just as well, if you want that, plus it's actually all optimized for router-use.
I have access to a Puma6-based device and sure, the dual-core Atom is fast enough to do a lot of stuff, but the single ARM-core is excruciatingly slow. And guess what? All the cable-management stuff is relegated to the ARM-core, the web-UI runs on the ARM-core, nearly everything runs on it and the x86-cores, in the meantime, just sit idle -- they are only used for NAS-functionality, streaming DVB-C content and Google Music. It's ridiculous how stupid the whole thing is. The box is also ridiculously easy to cause to crash, it's really easy to break into, even without physical access to the device in the first place and so on.
Indeed, this isn't anything new and all sorts of conductive inks have been around for a good while. You can even modify an inkjet-printer to print with conductive ink, or you can use conductive filament on a 3D-printer. The problem has always been the high resistance and I do not see anything in the article indicating that they'd have solved this.
Staring at the score a single reviewer gives is, IMHO, stupid, but taking a look at what thousands of people think gives a much better image of what it's actually like. On Rotten Tomatoes, IMDB and the likes I always look at the user-score, not the critic-score, and pretty often the user-score more-or-less matches with my opinion.
No. HTTPS protects the traffic between the server and your machine, but it doesn't hide the URL you've requested, as you can only establish an HTTPS-connection in the first place after you've contacted the URL.
They meant "mouthing the password," it's just poorly worded. There's e.g. the excerpt of "Third, lip passwords don't rely on speech recognition, meaning they can be used in noisy environments." in the article, which obviously wouldn't work if you had to actually say the password out loud -- the background noise would just drown you out. The system just relies on lip shape and mouth movement, not actually hearing anything.
The official Orange Pi - images are total fucking crapshoot, being so bad they make even your mum's cameltoe look appealing in comparison! It's not the availability of apps that is the problem, it's the support for the boards and all of their features, including Mali-drivers, or the closed, undocumented WiFi-chips, and so on that is the problem! Xunlong ain't doing shit to help get drivers mainlined in the kernel, they just produce a shitty image that barely boots and then hope the community will do all the hard work for them.
So, what's the point with these "snaps?" How do they make the situation any better? Oh, they don't? Weeellll...
Um, the schematics for the Orange Pi - boards have been available on their website from the beginning. I would know, considering I've fucking downloaded and read through them!
Slashdot Mirror
I used Linux on the desktop for years myself, but eventually I decided to ditch it and just stick to Windows exactly for the reasons you mention: there was always something breaking and I could never get all the hardware-features working that worked fine under Windows. The desktop-environments themselves were also a big, sodding mess and for anything more than typing text in gedit or similar or browsing the web you had to drop to CLI, and it's still that way even to this day.
I'm definitely very happy with Linux as a server-OS on both my NAS and my router, but on the desktop? No, not happening; on the desktop I want pretty GUI-applications I can just go clicky-click around when I want something happening, I want all my hardware working, I want to play games, I want to be able to consume online-content, and I don't want an update to GRUB fucking up my boot so I have to dig out systemrescuecd yet again.
This is what I was coming to say. I know bashing Microsoft/Windows is trendy and all that, especially in this den of rabid, mouth-breathing troglodytes, but these Atom CPUs have a GPU made by Imagination Tech., not Intel, and it seems that is the culprit here. Acer did say they are working with Microsoft to try to get this fixed, too, so this should just be temporary.
Yeah, no. I ain't going to just blindly believe your claim, unless you can actually back it up.
I can't help but wonder if/how much it affects performance, though. There are plenty of cases where hyper-threading is a very welcome feature and some people might be upset, if there was a hit to performance. I don't own a Skylake - or Kaby Lake - product, but I am just curious of real-world implications.
The RPi isn't the only SBC with a GPIO-header with user-accessible I2C and SPI. There are about a billion different ARM-based SBCs like that around by now, and plenty of x86- and MIPS-based ones with available GPIO-headers, too. Of course it has "strength" over a board without such accessible interfaces, but that's just a stupid comparison to begin with, like comparing apples and airplanes, and it has no "strength" over any other board with those interfaces.
It's not a bug, it's a hardware-limitation. The SoC only provides a single USB-port and the only way of connecting a USB-Ethernet and providing 4 USB-ports for the end-user to use is through a USB-hub, which obviously will mean that they all share the 480Mbps bandwidth of the single USB-port the hub is connected to. The only way the RPi Foundation could fix that is... by switching to another SoC. They'd lose backwards-compatibility, though, so that's not going to happen anytime soon.
There is an open-source driver for the GPU nowadays, capable of doing OpenGL (not just OpenGL ES) and such, and there is work going on for developing an open-source blob for booting the boards, too. I haven't checked in a good while how the projects are faring, but the last time I tried the open-source GPU-driver, it worked surprisingly well already. These projects give hope for a pretty good long-term support as long as one doesn't need H/W-accelerated decoding/encoding (there is no open-source support for that atm., AFAIK) of video, and even that may still one day get an open-source implementation.
One of the problems with x86 SBCs is that they are pretty much solely based on Intel's offerings, which makes them quite a bit more expensive and therefore not as appealing to home-tinkerers and the likes. I suppose that is the primary reason for their apparent unpopularity. That said, I do like the promise of the x86-boards myself, what with proper QuickSync for all sorts video-needs, like e.g. realtime transcoding, and full OpenGL instead of OpenGL ES for any graphical applications and quite mature software-landscape in the big picture. They're just lacking in community-support and e.g. all the easy-to-follow tutorials and such for home-tinkerers to interface with all sorts of sensors and stuff. One can always hope things will pick up; the fact that Intel dropped their line of SBCs isn't necessarily an indication of the market at large, as Intel has a tendency of doing this kind of a thing, and there are several other players who manufacture x86 SBCs.
You do realize that there are vendors who do x86 SBCs other than Intel? It's totally not pointless to bring up x86 SBCs in a discussion about, you know, SBCs.
How do the different Pis have a "strength in communicating with an Arduino?" The x86- and MIPS-boards can do that just as well, there is nothing stopping one from communicating with an Arduino over I2C, SPI, serial or whatever even on the other platforms. And no, Pi definitely doesn't have a "strength" in Ethernet, either, considering it's just 10/100 and it's actually a USB-device and thus eats bandwidth from the USB-ports, all of which are internally connected to a single USB-hub on the PCB.
"has a good Privacy Policy, and doesn't have a history of breaches or allowing snooping." -- Runbox fits all of those. Norwegians have very strong laws regarding privacy, which should please you, and the company doesn't do any advertising or crawling through your emails for tracking or anything like that. It's not a free service nor is it the cheapest one available, but I've been their customer for several years and I would at least recommend one to take a look at their offerings.
I did mention they are quite expensive, but they do have prices listed in the shop quite clearly. Maybe you missed the two links I gave above, but there's their online-shop behind the first link with a selection of the various Up^2 - boards at http://up-shop.org/4-up-boards . They even sell a UPS-like expansion board for it on the shop, too, so one doesn't have to hack one together themselves necessarily -- though, at $89 it may not be the world's best value. I suppose it depends on one's capability of cobbling one together themselves.
Well, perhaps I should've included some link, so here are two: http://www.up-board.org/upsqua... https://www.kickstarter.com/pr... But yes, they have a stupid name for the boards. Luckily, the name doesn't make the hardware any worse.
If one was going the DIY-way, I'd rather recommend Up^2. It's an actual x86-board, so it can run all the usual x86-stuff, there's a proper mPCI-E - slot for mSATA- and/or NVMe-drives or whatever mPCI-E card you may want to use, an M.2 2230 E-key for real, proper WiFi-cards, a SATA-connector, 3x USB3.0 (and a couple USB2.0-ports via a pin-header), a lot, lot more capable GPU than Raspberry Pi's one, built-in eMMC (the top-end model has a 128GB one) and so on.
The thing is, an RPi makes for a really crappy desktop-experience. The Up^2 is significantly more expensive, but it's also significantly more capable and much better suited for desktop-use.
I actually really liked the first Resident Evil - movie, the second was ok, but the rest of them I just yawned and rolled my eyes through and now afterwards I just can't recall those movies at all -- it's like I had never even seen them in the first place, which I find quite funny. Now, I have not played the games, I do not have any connection to the lore or anything, so I only watched those movies as exactly that: movies about some mutating viral zombie-thingamabob. Seeing as how exceedingly quickly the movies hit the rock bottom quality-wise I have a hard time imagining any reboots can get any worse; either they'll stay as bad, or get better.
What's stopping you from using QoS and other kinds of filtering-techniques on OpenWRT/LEDE? I use QoS on my router running LEDE without an issue.
A standard router is better optimized H/W-wise for these tasks, just get one that is supported by OpenWRT and/or LEDE. OpenWRT/LEDE gives you SSH and everything else just as well, if you want that, plus it's actually all optimized for router-use.
I take it this stupid article refers to NAT routers, and not cable modems at all.
These devices are actually both routers and cable-modems.
I have access to a Puma6-based device and sure, the dual-core Atom is fast enough to do a lot of stuff, but the single ARM-core is excruciatingly slow. And guess what? All the cable-management stuff is relegated to the ARM-core, the web-UI runs on the ARM-core, nearly everything runs on it and the x86-cores, in the meantime, just sit idle -- they are only used for NAS-functionality, streaming DVB-C content and Google Music. It's ridiculous how stupid the whole thing is. The box is also ridiculously easy to cause to crash, it's really easy to break into, even without physical access to the device in the first place and so on.
Indeed, this isn't anything new and all sorts of conductive inks have been around for a good while. You can even modify an inkjet-printer to print with conductive ink, or you can use conductive filament on a 3D-printer. The problem has always been the high resistance and I do not see anything in the article indicating that they'd have solved this.
Staring at the score a single reviewer gives is, IMHO, stupid, but taking a look at what thousands of people think gives a much better image of what it's actually like. On Rotten Tomatoes, IMDB and the likes I always look at the user-score, not the critic-score, and pretty often the user-score more-or-less matches with my opinion.
No. HTTPS protects the traffic between the server and your machine, but it doesn't hide the URL you've requested, as you can only establish an HTTPS-connection in the first place after you've contacted the URL.
They meant "mouthing the password," it's just poorly worded. There's e.g. the excerpt of "Third, lip passwords don't rely on speech recognition, meaning they can be used in noisy environments." in the article, which obviously wouldn't work if you had to actually say the password out loud -- the background noise would just drown you out. The system just relies on lip shape and mouth movement, not actually hearing anything.
The official Orange Pi - images are total fucking crapshoot, being so bad they make even your mum's cameltoe look appealing in comparison! It's not the availability of apps that is the problem, it's the support for the boards and all of their features, including Mali-drivers, or the closed, undocumented WiFi-chips, and so on that is the problem! Xunlong ain't doing shit to help get drivers mainlined in the kernel, they just produce a shitty image that barely boots and then hope the community will do all the hard work for them.
So, what's the point with these "snaps?" How do they make the situation any better? Oh, they don't? Weeellll...
Um, the schematics for the Orange Pi - boards have been available on their website from the beginning. I would know, considering I've fucking downloaded and read through them!