What chip, though? That's the problem. IBM has only been marketing relatively low-end cores (PPC 460) that are in between ARM-class and x64-class. That probably wouldn't be enough.
TFA suggests another tri-core 970 derivative, probably a process-shrunk version of the one in the 360, but I'm skeptical.
I'm curious to see what kind of hardware this thing has - there's not really a strong candidate for a processor right now. IBM's recent work on embedded PPC cores has been more along the lines of multicore low-clocked low-power designs (the PPC A2 and 4xx), which would be much faster than the 750 core used in the GameCube and the Wii but would still be unlikely to match the single-thread performance of the 970's in the Xbox 360 and the PS3. ARM isn't close to fast enough right now to deliver high-end performance and to emulate the Wii. Maybe Nintendo's building an x86 system, like the original Xbox.
Sure they were. The processors, despite initial hype, were nothing special (two cycles to complete an integer instruction? Really?). They were coupled to small amounts of relatively slow RAM. The GPU's were midrange to high-end for the day, but are barely comparable to integrated graphics now.
I didn't say that you couldn't buy them bare, just that it was fairly unusual. I doubt it makes up the difference between Windows and Linux ($6.3bn vs $2.5bn, 1.5mn units vs 450k units, according to IDC in Q4 2010.)
Very few. Most servers from major OEM's (HP, IBM, Fujitsu) which make up a large majority of server sales will include an OS, whether its Windows, Linux, or a proprietary system.
Windows ships more servers by both revenue and volume than Linux. Mobile devices, if you mean consumer electronics, are largely Linux, but embedded systems (especially lower-end ones) are not.
Nobody is voting to ban coal power production any time in the next thirty years, due to the annoying fact that it would result in a total collapse of the United States economy.
I have to wonder how much power that would use if it ran on mainframes or large UNIX servers rather than unreliable and relatively slow clusters of small machines. It's strange that none of the "new generation" websites are choosing to go with bigger systems, despite the fact that they tend to perform better on both performance and power/performance.
Even more bizarrely, the trunk directory now contains builds of Firefox 6. Two years ago, these "major releases" would have been point releases at best.
Nowhere does it say the Tandem/Compaq/HP NonStop ran Windows. It didn't. They just licensed some middleware and OS components to Microsoft, which is nothing special. They did the same thing with SCO around the same time.
A modern x64 core (Nehalem, Sandy Bridge, arguably K10) is not only faster than ARM, but so spectacularly faster that there's really no comparison. Atom - you know, the slow-ass netbook processor - is much faster than the Cortex A9 at the same clock speed, running integer workloads. That's pretty much the most favorable comparison you can get for ARM, and it still loses out badly.
ARM isn't standardized like x86 is, so probably not... at least not easily. IBM PC clones use a fairly standard set of firmware and peripherals, whereas ARM-based machines tend to be largely custom, just with a degree of binary compatibility between them. Getting Windows running on an iDevice would take serious work.
The NonStop didn't, and doesn't, run Windows. It runs a custom operating system called NSK, which is somewhat unique (each core runs a copy of the OS).
But how many of those background tasks actually require meaningful CPU time, rather than just running a few hundred thousand instructions and blocking on I/O?
County DSP-like coprocessors is letting the marketdroids win. Tegra 2 is a dual-core processor, and a somewhat crippled one - OMAP4 outperforms it in just about every application.
At least based on SPECint benchmarks, ARM is substantially slower than Atom at the same clock speed, and C2D is years beyond Atom. A quad Snapdragon might be only 20-30% slower for properly threaded integer code, but it would be hugely slower on floating point and on the most important factor: memory and I/O bandwidth.
Maintaining a project the size of a complete, commercial-grade, operating system takes incredible resources. Saying "let the open-source community handle it" without commercial backers isn't viable.
First glance at benchmarks indicates that this is still a bit slower than Power7, and has a similar or more expensive price. Considering we're a few months away from the Power7+ kicker, I don't expect this to have much adoption outside Windows Server users.
Slashdot Mirror
The 243MHz "Starlet" is a small component of the processor, basically a service chip. The main processor is a 729MHz PowerPC G3 derivative.
What chip, though? That's the problem. IBM has only been marketing relatively low-end cores (PPC 460) that are in between ARM-class and x64-class. That probably wouldn't be enough.
TFA suggests another tri-core 970 derivative, probably a process-shrunk version of the one in the 360, but I'm skeptical.
I'm curious to see what kind of hardware this thing has - there's not really a strong candidate for a processor right now. IBM's recent work on embedded PPC cores has been more along the lines of multicore low-clocked low-power designs (the PPC A2 and 4xx), which would be much faster than the 750 core used in the GameCube and the Wii but would still be unlikely to match the single-thread performance of the 970's in the Xbox 360 and the PS3. ARM isn't close to fast enough right now to deliver high-end performance and to emulate the Wii. Maybe Nintendo's building an x86 system, like the original Xbox.
Sure they were. The processors, despite initial hype, were nothing special (two cycles to complete an integer instruction? Really?). They were coupled to small amounts of relatively slow RAM. The GPU's were midrange to high-end for the day, but are barely comparable to integrated graphics now.
I didn't say that you couldn't buy them bare, just that it was fairly unusual. I doubt it makes up the difference between Windows and Linux ($6.3bn vs $2.5bn, 1.5mn units vs 450k units, according to IDC in Q4 2010.)
Very few. Most servers from major OEM's (HP, IBM, Fujitsu) which make up a large majority of server sales will include an OS, whether its Windows, Linux, or a proprietary system.
Windows ships more servers by both revenue and volume than Linux. Mobile devices, if you mean consumer electronics, are largely Linux, but embedded systems (especially lower-end ones) are not.
Except that you have to deal with the garbage of running a cluster, having a higher cost of licensing, and a higher rate of failure.
Nobody is voting to ban coal power production any time in the next thirty years, due to the annoying fact that it would result in a total collapse of the United States economy.
I have to wonder how much power that would use if it ran on mainframes or large UNIX servers rather than unreliable and relatively slow clusters of small machines. It's strange that none of the "new generation" websites are choosing to go with bigger systems, despite the fact that they tend to perform better on both performance and power/performance.
Even more bizarrely, the trunk directory now contains builds of Firefox 6. Two years ago, these "major releases" would have been point releases at best.
Nowhere does it say the Tandem/Compaq/HP NonStop ran Windows. It didn't. They just licensed some middleware and OS components to Microsoft, which is nothing special. They did the same thing with SCO around the same time.
A modern x64 core (Nehalem, Sandy Bridge, arguably K10) is not only faster than ARM, but so spectacularly faster that there's really no comparison. Atom - you know, the slow-ass netbook processor - is much faster than the Cortex A9 at the same clock speed, running integer workloads. That's pretty much the most favorable comparison you can get for ARM, and it still loses out badly.
ARM isn't standardized like x86 is, so probably not... at least not easily. IBM PC clones use a fairly standard set of firmware and peripherals, whereas ARM-based machines tend to be largely custom, just with a degree of binary compatibility between them. Getting Windows running on an iDevice would take serious work.
The NonStop didn't, and doesn't, run Windows. It runs a custom operating system called NSK, which is somewhat unique (each core runs a copy of the OS).
WebKit was a derivative of KHTML, a GPL'd system.
How did you get to that conclusion? Does OS X have a tablet mode similar to Windows Phone 7's UI, or a ribbon interface?
What a strange troll...
But how many of those background tasks actually require meaningful CPU time, rather than just running a few hundred thousand instructions and blocking on I/O?
County DSP-like coprocessors is letting the marketdroids win. Tegra 2 is a dual-core processor, and a somewhat crippled one - OMAP4 outperforms it in just about every application.
At least based on SPECint benchmarks, ARM is substantially slower than Atom at the same clock speed, and C2D is years beyond Atom. A quad Snapdragon might be only 20-30% slower for properly threaded integer code, but it would be hugely slower on floating point and on the most important factor: memory and I/O bandwidth.
What massive-scale open-source project is commercial-grade without being largely commercial backed?
Maintaining a project the size of a complete, commercial-grade, operating system takes incredible resources. Saying "let the open-source community handle it" without commercial backers isn't viable.
Everything I've seen other than Java is, at best, half-baked on Android. Scala is the closest I've seen to full-featured.
The only "debatable" parts of Mono are the WinForms bits, which nobody really uses anyway. Keep on trollin' though.
First glance at benchmarks indicates that this is still a bit slower than Power7, and has a similar or more expensive price. Considering we're a few months away from the Power7+ kicker, I don't expect this to have much adoption outside Windows Server users.