Note that this doesn't mean custom compiling the kernel is a bad idea not because of what you can strip out but because compiling stuff into the kernel saves the need for an initrd.
Actually it does with a 7200RPM drive rotational latency alone is going to eat over 4ms on average (sometimes you will be lucky and it will be near zero, sometimes you will be unlucky and it will be double that) for each access and actually reading a file is likely to take multiple accesses. Still reading in a single file (even a fairly large one) from a modern drive takes a pretty small ammount of time if you only do one.
The thing is most disk based operating systems i've seen don't work like that, they are made up of a large number of files which are loaded one after another. Worse those files often represent programs/services/etc which have some other delays (waiting for hardware, waiting for interruptions from the user etc) in them. Such a design is much slower to boot than a single system image, especially if it wasn't explicitly designed for speed.
Linux itself loads extremely quickly (measured from making selection in the bootloader to first running userland code) because it's a single image being pulled in from disk in a most likely sequential manner. It's all the stuff built on top of it that takes ages.
It takes a lot of alcohol to poision and afaict it breaks down organically into harmless stuff so releases aren't a concern.
Afaict liquid mercury isn't hugely dangerous simply because the body won't absorb much but vapours of mercury are worse and organic compounds of mercury are even worse. This gives a good reason for controlling it's use.
but has fallen into disuse Do you have any sources for that claim or is it just a guess based on your personal experiance (my personal experiance is the opposite)?
Unfortunately the regime being on the internet and the general public being on the internet are very different things. My understanding is in north korea the regime are but the general public aren't (even those of the general public who can afford computers etc).
If they are using windows they have hidden it very well. Could well be a winCE base or so though (afaict wince can be made to look pretty much however the device vendor likes).
Yeah, I have a FSH8 (which is closely related to the FSH4 but goes to 8GHz) in my office and it's a really nicely designed peice of kit. It's light, the controls are responsive and well designed, the dyamic range is good (at least compared to the anritsu I don't have anything else to compare it too).
Much nicer to use than the anritsu MS2036A we have.
Yeah they aren't cheap, the low end of the prices listed on that google search seem to tally with what the base price would have been (I had to work the education discount backwards to figure out the base price) for ordering direct from R&S (we bought a FSH8 recently at uni).
No idea if the police would have got a discount , what the custom setup actually consisted of (it may well have just been setting up the right frequency ranges) or whether they bought any extra options (extra options can seriously add to the cost but i'd think the base model hardware would be sufficiant for this job).
An interesting side story here is that apparently adobe will be supporting VP8 in flash. Whether they will actually get arround to it (afaict they haven't done so yet) and whether they will support all of webm remains unclear though.
Might depend on your browser. The PNG guys decided to make PNG static image format only and refused to endorse APNG prefering to push the incompatible and overcomplicated MNG format instead. Afaict no browser supports MNG and only firefox supports APNG (even iceweasel doesn't support APNG because they use the system libpng rather than mozilla's patched version).
I wonder what the situation is with VP8 on flash, adobe have said they will support it but I haven't found any evidence they actually have followed through and if so whether they support the VP8+vorbis+matroska combination google is pushing.
Support for webm in flash would give it VERY good overall coverage and afaict would leave the idevices as the only sigificant reason to continue offering h.264.
Too bad CDs aren't as durable as LPs and cassettes. Hmm, my experiance has been that pressed CDs are pretty durable. You can damage them with really bad scratches but the same applies to LPs to a greated extent and cassetted have their own durability problems (They can get chewed up badly if there are problems in a player, this was especially an issue with automotive front loading decks afaict).
It's the recordable CDs that have a habbit of dying from issues with the dye:(.
While true in a sense afaict in nearly all cases the CD is the "most original" copy available and the various download formats are lossy encodings of what is on the CD (sometimes you will see a lossless copy of what was on the CD and very rarely you will see a download release that is better than the CD). Therefore when you want a source to encode at your preffered format and bitrate the CD is the best source available.
We need everyone to agree on one codec Unfortunately that is simply not likely to happen:(
Mozilla can't implement h.264 without both paying quite a lot of money and more importantly violating the core principles of being an opensource project and they refuse to use OS level implementations because they believe in consistency across platforms.
Apple and nokia refused to implment theora claiming submarine patent risks. Whether that was the real reason or whether the real reason was a vested interest in pushing h.264 remains unclear.
IIRC MS has been sitting on the h.264 side as well though i'm not sure of their reasons.
Google has brought VP8 along as a third option but so far they don't seem to be doing a very good job of bringing the likes of apple arround to it.
I think the only way this will truely be solved is when the patents on h.264 expire:(
The reason that good codecs stick around is that there's a lot of hardware that will play/display them. A lot of people still have DVD-players so MPEG-2 still gets used because that's what the player expects, even though MPEG-2 isn't all that good compared to h.264. MP3 is still around because there are still tons of MP3 players and almost any device that can output audio continues to include MP3 support because it's cheap to do so. Exactly and assuming bandwidth costs continue to trend downwards then at some point the benefit of wider compatibility comes to outweigh the cost of larger file-sizes. This has already happened for audio and still images and at some point in the future I would expect it to happen for video too. Age also brings the benefit of a higher likelihood of being free of patent issues (unfortunately I believe in the US there is still the possibility of a submarine patent that was filed years ago finally being granted).
Suppose the game used a 10Hz tick (it's probablly faster) and at each tick there were 4 possible combinations inputs (there are probablly more). In 3 hours that's 4^36000 = 2^720 > 10^216 (2^10 > 10^3) possible combinations of inputs. opps sorry those are the figures for one hour. still makes the point though.
Seems like we have an AC who doesn't understand exponentials.
Suppose the game used a 10Hz tick (it's probablly faster) and at each tick there were 4 possible combinations inputs (there are probablly more). In 3 hours that's 4^36000 = 2^720 > 10^216 (2^10 > 10^3) possible combinations of inputs.
Of course most of those will be stupid but there is still an unimaginablly large number of sane possibilities.
It fills-up and then you're done Not really true with cellular. You can increase bandwidth per unit of area by packing the basestations more densely. Obviously this doesn't help if you have a huge number of people in exactly the same place but that is rarely the case.
Software implementation of new instruction sets was always trailing years. Trailing years isn't enough though. Patents last 20 years or so.
And is more that good enough to drive Windows XP I'm not so sure about that. Afaict the only 233MHz pentium had MMX as did all K6 chips. I dunno if there are any patent issues with MMX but if there are it could be a problem.
which is still current. Currently supported but afaict no longer availiable to OEMs and by 2013 when the pentium becomes 20 years old it will be very near the end of support.
Windows 7 doesn't explicitly specify families but I would be surprised if it ran on anything less than a pentium pro equivilent.
plus there is the side fact that the PRNG is NOT a true random number generator. So it is quite possible that getting perfect results from it every time is impossible.
The only metric worth anything is performance per dollar. BULLSHIT
From a customers point of view what matters depends on whether the workload can/will be split across multiple nodes or not. If it can't/won't the question is performance at a given price point and whether the extra performance from moving to a higher price point is worth the extra money. If it can and will then what matters is performance per dollar of the ENTIRE SYSTEM (potentially including things like hosting, power, software and even admin costs).
Consistency is also valuable to larger customers. Windows doesn't get on particually well with being imaged between systems of different CPU vendors (see for example the issue with XP where an image transfered from an intel system to an AMD one would go into a reboot loop after installing a service pack). Spending slightly more on each box may well be worth it to keep your low and high end boxes on the same platform (afaict this is why LGA775 still rules the corporate desktop market, it's not that good value in any particular segment but it can cover most of the range of requirements/pricepoints reasonablly well).
And of course there is the issue of performance at what. That chart shows quad cores as being the best value, of course that is true if and only if your workload can actually use all those cores.
But this isn't about the customers point of view, it's about the suppliers point of view.
From a suppliers point of view what matters is profit. You make the most profit by having high margins and high volume. You make least profit by having low margins and low volume. If your profits on production aren't high enough to cover your R&D costs you lose money overall.
Afaict Intel has both high volume and high margins while AMD has tight margins (they have to sell their top end product at the price intel sells it's midrange for and I doubt there is much difference in manufacture cost between intels top end and AMDs top end) and lower volume.
Intel charges more because it has brand recognition which was propped up with the illegal activities of this convicted monopolist. That is part of it but I think it's far from the whole story. AMD may be struggling less if Intel had played fair but I still think they would be struggling.
IANAL, someone enlighten me on when X86 patent runs out (am I even in the right ballpark, legally) ? I mean, X86 has been around.. forever. It depends what you mean by x86. x86 is being constantly extended and those extensions are almost certainly being patented.
Any patents on 8086 itself will almost certainly have long expired. Hell even the 80486 is probablly pretty much clear by now (though I can't say for sure due to craziness of patent law in some countries including the US) and IIRC there are a few companies out there making 486 clones for the embedded market (though I can't seem to find any details right now).
BUT IIRC Intel and AMD both hold a lot of patents on both modern extensions to x86 and on fast ways of implementing x86 and are probablly unwilling to license them. This means the only option for a new entrant is to go ahead anyway and hope they have enough retaliatory patents to get a settlement that lets them continue.
Slashdot Mirror
Note that this doesn't mean custom compiling the kernel is a bad idea not because of what you can strip out but because compiling stuff into the kernel saves the need for an initrd.
Actually it does with a 7200RPM drive rotational latency alone is going to eat over 4ms on average (sometimes you will be lucky and it will be near zero, sometimes you will be unlucky and it will be double that) for each access and actually reading a file is likely to take multiple accesses. Still reading in a single file (even a fairly large one) from a modern drive takes a pretty small ammount of time if you only do one.
The thing is most disk based operating systems i've seen don't work like that, they are made up of a large number of files which are loaded one after another. Worse those files often represent programs/services/etc which have some other delays (waiting for hardware, waiting for interruptions from the user etc) in them. Such a design is much slower to boot than a single system image, especially if it wasn't explicitly designed for speed.
Linux itself loads extremely quickly (measured from making selection in the bootloader to first running userland code) because it's a single image being pulled in from disk in a most likely sequential manner. It's all the stuff built on top of it that takes ages.
It takes a lot of alcohol to poision and afaict it breaks down organically into harmless stuff so releases aren't a concern.
Afaict liquid mercury isn't hugely dangerous simply because the body won't absorb much but vapours of mercury are worse and organic compounds of mercury are even worse. This gives a good reason for controlling it's use.
but has fallen into disuse
Do you have any sources for that claim or is it just a guess based on your personal experiance (my personal experiance is the opposite)?
Unfortunately the regime being on the internet and the general public being on the internet are very different things. My understanding is in north korea the regime are but the general public aren't (even those of the general public who can afford computers etc).
If they are using windows they have hidden it very well. Could well be a winCE base or so though (afaict wince can be made to look pretty much however the device vendor likes).
Yeah, I have a FSH8 (which is closely related to the FSH4 but goes to 8GHz) in my office and it's a really nicely designed peice of kit. It's light, the controls are responsive and well designed, the dyamic range is good (at least compared to the anritsu I don't have anything else to compare it too).
Much nicer to use than the anritsu MS2036A we have.
Yeah they aren't cheap, the low end of the prices listed on that google search seem to tally with what the base price would have been (I had to work the education discount backwards to figure out the base price) for ordering direct from R&S (we bought a FSH8 recently at uni).
No idea if the police would have got a discount , what the custom setup actually consisted of (it may well have just been setting up the right frequency ranges) or whether they bought any extra options (extra options can seriously add to the cost but i'd think the base model hardware would be sufficiant for this job).
An interesting side story here is that apparently adobe will be supporting VP8 in flash. Whether they will actually get arround to it (afaict they haven't done so yet) and whether they will support all of webm remains unclear though.
Might depend on your browser.
The PNG guys decided to make PNG static image format only and refused to endorse APNG prefering to push the incompatible and overcomplicated MNG format instead. Afaict no browser supports MNG and only firefox supports APNG (even iceweasel doesn't support APNG because they use the system libpng rather than mozilla's patched version).
20 years from when? How much life is left in these patents?
ARS claims "roughly 2025".
I wonder what the situation is with VP8 on flash, adobe have said they will support it but I haven't found any evidence they actually have followed through and if so whether they support the VP8+vorbis+matroska combination google is pushing.
Support for webm in flash would give it VERY good overall coverage and afaict would leave the idevices as the only sigificant reason to continue offering h.264.
Too bad CDs aren't as durable as LPs and cassettes.
Hmm, my experiance has been that pressed CDs are pretty durable. You can damage them with really bad scratches but the same applies to LPs to a greated extent and cassetted have their own durability problems (They can get chewed up badly if there are problems in a player, this was especially an issue with automotive front loading decks afaict).
It's the recordable CDs that have a habbit of dying from issues with the dye :(.
While true in a sense afaict in nearly all cases the CD is the "most original" copy available and the various download formats are lossy encodings of what is on the CD (sometimes you will see a lossless copy of what was on the CD and very rarely you will see a download release that is better than the CD). Therefore when you want a source to encode at your preffered format and bitrate the CD is the best source available.
We need everyone to agree on one codec :(
Unfortunately that is simply not likely to happen
Mozilla can't implement h.264 without both paying quite a lot of money and more importantly violating the core principles of being an opensource project and they refuse to use OS level implementations because they believe in consistency across platforms.
Apple and nokia refused to implment theora claiming submarine patent risks. Whether that was the real reason or whether the real reason was a vested interest in pushing h.264 remains unclear.
IIRC MS has been sitting on the h.264 side as well though i'm not sure of their reasons.
Google has brought VP8 along as a third option but so far they don't seem to be doing a very good job of bringing the likes of apple arround to it.
I think the only way this will truely be solved is when the patents on h.264 expire :(
The reason that good codecs stick around is that there's a lot of hardware that will play/display them. A lot of people still have DVD-players so MPEG-2 still gets used because that's what the player expects, even though MPEG-2 isn't all that good compared to h.264. MP3 is still around because there are still tons of MP3 players and almost any device that can output audio continues to include MP3 support because it's cheap to do so.
Exactly and assuming bandwidth costs continue to trend downwards then at some point the benefit of wider compatibility comes to outweigh the cost of larger file-sizes. This has already happened for audio and still images and at some point in the future I would expect it to happen for video too. Age also brings the benefit of a higher likelihood of being free of patent issues (unfortunately I believe in the US there is still the possibility of a submarine patent that was filed years ago finally being granted).
I haven't but I'd expect even there an operator could densify their network if they had too. They just would have to start thinking 3D instead of 2D.
Suppose the game used a 10Hz tick (it's probablly faster) and at each tick there were 4 possible combinations inputs (there are probablly more). In 3 hours that's 4^36000 = 2^720 > 10^216 (2^10 > 10^3) possible combinations of inputs.
opps sorry those are the figures for one hour. still makes the point though.
Seems like we have an AC who doesn't understand exponentials.
Suppose the game used a 10Hz tick (it's probablly faster) and at each tick there were 4 possible combinations inputs (there are probablly more). In 3 hours that's 4^36000 = 2^720 > 10^216 (2^10 > 10^3) possible combinations of inputs.
Of course most of those will be stupid but there is still an unimaginablly large number of sane possibilities.
It fills-up and then you're done
Not really true with cellular. You can increase bandwidth per unit of area by packing the basestations more densely. Obviously this doesn't help if you have a huge number of people in exactly the same place but that is rarely the case.
Software implementation of new instruction sets was always trailing years.
Trailing years isn't enough though. Patents last 20 years or so.
And is more that good enough to drive Windows XP
I'm not so sure about that. Afaict the only 233MHz pentium had MMX as did all K6 chips. I dunno if there are any patent issues with MMX but if there are it could be a problem.
which is still current.
Currently supported but afaict no longer availiable to OEMs and by 2013 when the pentium becomes 20 years old it will be very near the end of support.
Windows 7 doesn't explicitly specify families but I would be surprised if it ran on anything less than a pentium pro equivilent.
plus there is the side fact that the PRNG is NOT a true random number generator. So it is quite possible that getting perfect results from it every time is impossible.
The only metric worth anything is performance per dollar.
BULLSHIT
From a customers point of view what matters depends on whether the workload can/will be split across multiple nodes or not. If it can't/won't the question is performance at a given price point and whether the extra performance from moving to a higher price point is worth the extra money. If it can and will then what matters is performance per dollar of the ENTIRE SYSTEM (potentially including things like hosting, power, software and even admin costs).
Consistency is also valuable to larger customers. Windows doesn't get on particually well with being imaged between systems of different CPU vendors (see for example the issue with XP where an image transfered from an intel system to an AMD one would go into a reboot loop after installing a service pack). Spending slightly more on each box may well be worth it to keep your low and high end boxes on the same platform (afaict this is why LGA775 still rules the corporate desktop market, it's not that good value in any particular segment but it can cover most of the range of requirements/pricepoints reasonablly well).
And of course there is the issue of performance at what. That chart shows quad cores as being the best value, of course that is true if and only if your workload can actually use all those cores.
But this isn't about the customers point of view, it's about the suppliers point of view.
From a suppliers point of view what matters is profit. You make the most profit by having high margins and high volume. You make least profit by having low margins and low volume. If your profits on production aren't high enough to cover your R&D costs you lose money overall.
Afaict Intel has both high volume and high margins while AMD has tight margins (they have to sell their top end product at the price intel sells it's midrange for and I doubt there is much difference in manufacture cost between intels top end and AMDs top end) and lower volume.
Intel charges more because it has brand recognition which was propped up with the illegal activities of this convicted monopolist.
That is part of it but I think it's far from the whole story. AMD may be struggling less if Intel had played fair but I still think they would be struggling.
Or perhaps it's a sign that bulldozer will suck and he wants to get out before the shit hits the fan.
IANAL, someone enlighten me on when X86 patent runs out (am I even in the right ballpark, legally) ? I mean, X86 has been around.. forever.
It depends what you mean by x86. x86 is being constantly extended and those extensions are almost certainly being patented.
Any patents on 8086 itself will almost certainly have long expired. Hell even the 80486 is probablly pretty much clear by now (though I can't say for sure due to craziness of patent law in some countries including the US) and IIRC there are a few companies out there making 486 clones for the embedded market (though I can't seem to find any details right now).
BUT IIRC Intel and AMD both hold a lot of patents on both modern extensions to x86 and on fast ways of implementing x86 and are probablly unwilling to license them. This means the only option for a new entrant is to go ahead anyway and hope they have enough retaliatory patents to get a settlement that lets them continue.