All that statisitically significant means is "if the null hypothosis* is true the chances of results as bad or worse (for the null hypothosis) as those we obtained are below a given value".
It does not tell you what the correct alternate hypothosis is nor does it tell you with certainty that the null hypothosis is wrong.
And in particular finding something isn't statistically significant doesn't nessacerally mean the null hypothosis is right.
* The null hypothosis basically says "the results are random"
My understanding was that youtube had reached some backroom deals with music companies where the companies agree not to attempt to take down the music videos in exchange for a slice of ad-revenue and/or pointing users at places to legally buy the music in quetion but because it was only an agreement not to take enforcement action not an actual license agreement the artists themselves weren't seeing any of the money.
In terms of total hours it doesn't really seem much quicker than the more expensive "next day" services. It's just rather than "order in the early evening receive in the morning" it's "order in the early morning receive in the evening". I'm sure there are occasional situations where that is handy but it's still considerablly slower than driving to a nearby retailer. Heck it's considerablly slower than catching the bus to a nearby retailer.
When the shit hits the fan the difference between having something in an hour or so and having it in 10 hours or so is likely to be pretty significant. When the shit hasn't hit the fan a couple of days is usually fine.
The good thing is that this is the way it would work if you sold a used CD -- nothing stops you from ripping it before selling it, right? So there's nothing inherently different about the process.
One difference is with a CD you can pretty easilly tell the original from a copy (assuming here copies made with typical home equipment). Yes there is nothing stopping someone reselling the original and keeping a copy but they can't sell their copy twice without one of the buyers getting a dodgy copy.
Sites like redigi may do some checks to try and stop double resale but as the clones pop up that will get harder and harder to ensure.
One option would be to require all sites to register sale and resale of digital downloads in a central database with the person the database entry points at being considered the legitimate owner but that would raid concerns of it's own.
They don't want the best effort internet traffic to swamp the premium voice traffic. So if they want to go full VOIP they need to build QOS into every aspect of the network. Further they have to work out how you will handle phones switching mid call from 4G voip to 2G/3G circuit switched voice when they go out of 4G coverage. Finally all the carriers and phone vendors need to agree on this so it can be incorporated into mass market phones.
Plus even if they do get voice over 4G sorted they will still need the 2G/3G functionality so the phone works in areas without 4G network coverage.
It'll work just don't expect it to perform very well since everything will be forced over a single USB2 link (and remember USB2 is half duplex).
If you want a low power linux filesever you will probablly be better off looking for a hackable NAS device since the SOCs used in those will have native sata and native ethernet (unlike the one in the Pi which has neither).
Afaict the arm core in the pi is only armv6 and this serverely limits your choice of linux distros. For example the new hardfloat port of debian is armv7 or higher only.
It's not really intended to be a STB (it can be used as one but that's not it's primary purpose)
In time there will be an official case because one is needed for school deployments to placate H&S and reduce the risk of damage but it's not a priority right now.
There are a number of third party case designs but I doubt any of the third party case designers is prepared to put up the upfront money needed to get low per-unit costs.
Afaict they chose broadcom because at least one of the project founders had a job there and was able to use that job to get a sweetheart deal on SOC pricing (large volume pricing even for relatively small volume orders). That was what made a proper linux computer at such a low pricepoint possible without risking an unrealistic ammount of capital (noone really knows to what extent this will succeed yet).
This datasheet is nice to have but it's not vital. Exact details on what the IO lines can and cannot do is pretty tangetical to the PIs main purpose. Details of the video core are still secret but afaict that is the case with all the "mobile phone SOCs"
The makers of "mobile phone SOCs" are pretty secretive about their GPUs. I think there are two reasons for this, firstly to keep information out of competitors hands. Secondly to prevent the "mobile phone SOCs" from competiting with more expensive "ARM+DSP SOCs" intended for general processing applicaitons.
When running a company that makes such chips you essentially have two choices
1: don't release anything to those who haven't signed an NDA 2: go through your documentation and carefully separate the bits that you can tell the public about and the bits you are keeping
The question is whether the mindshare that can be bought though releasing a redacted datasheet is worth the cost and risk (that they left in some information they were trying to keep secret) of doing so. Afaict broadcom think that in the general case the answer is no but in the specific case of the Pi they have been persuaded to make an exception.
From what I can gather broadcom don't like dealing with small customers so if you are planning to grab a design chunk for your own design and you aren't shipping a huge number of units you might want to look elsewhere.
Plus for smaller runs it may well be cheaper to just buy the pis (and sit them on a support board made on a cheap 2 layer process that adds any extra hardware you need) than to design and have built a board of that caliber.
The pic32 while a high end microcontroller is still a microcontroller. The normal way of running it is without an OS with the whole application and all libraries built into one big binary. While someone did port an ancient version of BSD to it there is no way you would get a regular modern OS on the thing.
People ran (and still run judging by recent traffic in #debian-arm) debian on the NSLU2 and that only had 32 megabytes of memory at least until you started soldering.
but it only has one USB port, add a USB hub and a USB to ethernet adaptor (or a single device that combines them) and you are right back up arround the price of the model B with a less conviniant form factor.
The A will have it's uses but as an initial board (not a deployment board) for the tinkerers/hobbyists the B is a better fit. AIUI the point of this initial batch is to get the board into the hands of such people so any issues can be found and so a software ecosystem can form.
"Charzinski's 2010 paper shows the beneficial effects of caching on performance. Table 1 shows that the average top 500 home page goes from 507K and 64.7 requests upon initial cache-cleared load to 98.5K and 16.1 requests."
As I suspected it seems like the "nearly a megabyte" headline figure you quoted is for loading a page with nothing in cache and the average pageload is likely to be much lower.
Re:The ocean frontier - not
on
Remembering Sealab
·
· Score: 4, Informative
Going through the list there Jules Undersea Lodge: a converted two bedroom research facility. Located in a lake 21 feet down. Dive entry. Utter Inn: a box just below the surface with surface entry not much different from the lower sections of a boat except there is water between the upper and lower sections. Only one room. Hydropolis: looks like it was intended to be a proper hotel though only barely underwater and surface entry but the article you linked claims it as "under construction" but wikipedia links to another article that claims it is "nothing more than a pile of blueprints". Looks like it got nixed in the wake of the credit crunch. Poseidon Undersea Resorts: this does actually sound like an undersea hotel but from their website it is not at all clear whether it was ever finished or not. Trying to get a "booking request form" out of their website gives the message "Thank you for your interest in Poseidon Resorts. We welcome you to contact us after September 15, 2009.". This suggests the website hasn't been updated in years. Istanbul: I can't find any evidence of this underwater hotel actually exiting either.
The same could be said for space, there are a load of satellites and a single manned space station for research but there are no space cities or space hotels, there have only been 6 manned landings on the moon and afaict no human has left LEO since.
The fact is both space and deep ocean are hostile environments. Space has the additional issue of being extremely expensive to get to and from.
Do you have a source for that claim? (i'm not nessacerally saying it's wrong but the handful of pages I took a quick look at all seemed to be much smaller and I'd like to know how the number was arrived at and what assumptions were made)
The physical circuits are long gone in most places but on conventional digital phone networks keeping a call open keeps a reserved slice of bandwidth (a "virtual circuit") for the length of the call. Afaict on the landline side while telcos are starting to roll out VOIP there is still a lot of conventional virtual circuit based network around too. On the mobile side afaict virtually all phones in use today are still using circuit switched voice protocols (LTE is supposed to change this). IIRC circuit switched calls* also have priority over packet switched data traffic at least in GSM.
*You can have a circuit switched data call with GSM but almost noone does it anymore.
Do you really think anyone is going to wonder how many minutes they get out of their data plan when loading a typical web page is equivalent to 5 minutes of talk time?
4 megabytes or so seems insanely high for one web page to me.
Afaict the tag means that they have inspected the plane and are convinced it's safe enough to let you fly it but not convinced it's safe enough to let you run commercial operations with it.
Of course actually being safe and convincing the authorities something is safe are a very different matter.
AIUI ARM do HDL design of processor cores, then they pass that HDL on to other companies who make complete chip designs based on it. Those companies in turn pass the designs onto fabs (which may be in-house or external) for manufacture. IIRC some vendors also do their own HDL work and only license the basic architectural design from ARM.
You didn't make it clear whether you were replying to the first part of my post and being sarcastic or replying to the second part and agreeing with it. I will assume the latter.
The likes of rabbits and cane toads are precisely the kind of species you DO have to worry about. Cane toads produce tiny eggs which lay dormant in the water for some time before hatching. Rabbits hide underground. Both are also small and fast breeding. The result is it's nearly impossible to wipe them out.
Apple aren't really relavent here, they were just using apple's brand to try and make themselves look more legit. I'm sure a legitimate service could find someone to license a brand off if they didn't feel like building their own.
However as you allude to an equally good legitimate service cannot and will not happen (at least not without government intervention). The fundamental problem is that since licenses do cost money and some content owners are likely to either refuse outright or charge unreasonable rates a legitimate service is doomed to end up with both higher prices and less content than a pirate service.
Slashdot Mirror
or just to plain old random chance.
All that statisitically significant means is "if the null hypothosis* is true the chances of results as bad or worse (for the null hypothosis) as those we obtained are below a given value".
It does not tell you what the correct alternate hypothosis is nor does it tell you with certainty that the null hypothosis is wrong.
And in particular finding something isn't statistically significant doesn't nessacerally mean the null hypothosis is right.
* The null hypothosis basically says "the results are random"
My understanding was that youtube had reached some backroom deals with music companies where the companies agree not to attempt to take down the music videos in exchange for a slice of ad-revenue and/or pointing users at places to legally buy the music in quetion but because it was only an agreement not to take enforcement action not an actual license agreement the artists themselves weren't seeing any of the money.
That undestanding may be out of date though.
In terms of total hours it doesn't really seem much quicker than the more expensive "next day" services. It's just rather than "order in the early evening receive in the morning" it's "order in the early morning receive in the evening". I'm sure there are occasional situations where that is handy but it's still considerablly slower than driving to a nearby retailer. Heck it's considerablly slower than catching the bus to a nearby retailer.
When the shit hits the fan the difference between having something in an hour or so and having it in 10 hours or so is likely to be pretty significant. When the shit hasn't hit the fan a couple of days is usually fine.
The good thing is that this is the way it would work if you sold a used CD -- nothing stops you from ripping it before selling it, right? So there's nothing inherently different about the process.
One difference is with a CD you can pretty easilly tell the original from a copy (assuming here copies made with typical home equipment). Yes there is nothing stopping someone reselling the original and keeping a copy but they can't sell their copy twice without one of the buyers getting a dodgy copy.
Sites like redigi may do some checks to try and stop double resale but as the clones pop up that will get harder and harder to ensure.
One option would be to require all sites to register sale and resale of digital downloads in a central database with the person the database entry points at being considered the legitimate owner but that would raid concerns of it's own.
They don't want the best effort internet traffic to swamp the premium voice traffic. So if they want to go full VOIP they need to build QOS into every aspect of the network. Further they have to work out how you will handle phones switching mid call from 4G voip to 2G/3G circuit switched voice when they go out of 4G coverage. Finally all the carriers and phone vendors need to agree on this so it can be incorporated into mass market phones.
Plus even if they do get voice over 4G sorted they will still need the 2G/3G functionality so the phone works in areas without 4G network coverage.
It'll work just don't expect it to perform very well since everything will be forced over a single USB2 link (and remember USB2 is half duplex).
If you want a low power linux filesever you will probablly be better off looking for a hackable NAS device since the SOCs used in those will have native sata and native ethernet (unlike the one in the Pi which has neither).
I'd add another, armv7 instruction set support.
Afaict the arm core in the pi is only armv6 and this serverely limits your choice of linux distros. For example the new hardfloat port of debian is armv7 or higher only.
It's not really intended to be a STB (it can be used as one but that's not it's primary purpose)
In time there will be an official case because one is needed for school deployments to placate H&S and reduce the risk of damage but it's not a priority right now.
There are a number of third party case designs but I doubt any of the third party case designers is prepared to put up the upfront money needed to get low per-unit costs.
Afaict they chose broadcom because at least one of the project founders had a job there and was able to use that job to get a sweetheart deal on SOC pricing (large volume pricing even for relatively small volume orders). That was what made a proper linux computer at such a low pricepoint possible without risking an unrealistic ammount of capital (noone really knows to what extent this will succeed yet).
This datasheet is nice to have but it's not vital. Exact details on what the IO lines can and cannot do is pretty tangetical to the PIs main purpose. Details of the video core are still secret but afaict that is the case with all the "mobile phone SOCs"
The makers of "mobile phone SOCs" are pretty secretive about their GPUs. I think there are two reasons for this, firstly to keep information out of competitors hands. Secondly to prevent the "mobile phone SOCs" from competiting with more expensive "ARM+DSP SOCs" intended for general processing applicaitons.
When running a company that makes such chips you essentially have two choices
1: don't release anything to those who haven't signed an NDA
2: go through your documentation and carefully separate the bits that you can tell the public about and the bits you are keeping
The question is whether the mindshare that can be bought though releasing a redacted datasheet is worth the cost and risk (that they left in some information they were trying to keep secret) of doing so. Afaict broadcom think that in the general case the answer is no but in the specific case of the Pi they have been persuaded to make an exception.
From what I can gather broadcom don't like dealing with small customers so if you are planning to grab a design chunk for your own design and you aren't shipping a huge number of units you might want to look elsewhere.
Plus for smaller runs it may well be cheaper to just buy the pis (and sit them on a support board made on a cheap 2 layer process that adds any extra hardware you need) than to design and have built a board of that caliber.
The pic32 while a high end microcontroller is still a microcontroller. The normal way of running it is without an OS with the whole application and all libraries built into one big binary. While someone did port an ancient version of BSD to it there is no way you would get a regular modern OS on the thing.
Also it's MIPS not arm but meh on that.
People ran (and still run judging by recent traffic in #debian-arm) debian on the NSLU2 and that only had 32 megabytes of memory at least until you started soldering.
but it only has one USB port, add a USB hub and a USB to ethernet adaptor (or a single device that combines them) and you are right back up arround the price of the model B with a less conviniant form factor.
The A will have it's uses but as an initial board (not a deployment board) for the tinkerers/hobbyists the B is a better fit. AIUI the point of this initial batch is to get the board into the hands of such people so any issues can be found and so a software ecosystem can form.
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=461309 shows why xmms was kicked out of debian. Presumablly ubuntu just followed them. Not sure about the others.
http://www.websiteoptimization.com/speed/tweak/average-web-page/ [websiteoptimization.com]
"Charzinski's 2010 paper shows the beneficial effects of caching on performance. Table 1 shows that the average top 500 home page goes from 507K and 64.7 requests upon initial cache-cleared load to 98.5K and 16.1 requests."
As I suspected it seems like the "nearly a megabyte" headline figure you quoted is for loading a page with nothing in cache and the average pageload is likely to be much lower.
Going through the list there
Jules Undersea Lodge: a converted two bedroom research facility. Located in a lake 21 feet down. Dive entry.
Utter Inn: a box just below the surface with surface entry not much different from the lower sections of a boat except there is water between the upper and lower sections. Only one room.
Hydropolis: looks like it was intended to be a proper hotel though only barely underwater and surface entry but the article you linked claims it as "under construction" but wikipedia links to another article that claims it is "nothing more than a pile of blueprints". Looks like it got nixed in the wake of the credit crunch.
Poseidon Undersea Resorts: this does actually sound like an undersea hotel but from their website it is not at all clear whether it was ever finished or not. Trying to get a "booking request form" out of their website gives the message "Thank you for your interest in Poseidon Resorts. We welcome you to contact us after September 15, 2009.". This suggests the website hasn't been updated in years.
Istanbul: I can't find any evidence of this underwater hotel actually exiting either.
The same could be said for space, there are a load of satellites and a single manned space station for research but there are no space cities or space hotels, there have only been 6 manned landings on the moon and afaict no human has left LEO since.
The fact is both space and deep ocean are hostile environments. Space has the additional issue of being extremely expensive to get to and from.
Do you have a source for that claim? (i'm not nessacerally saying it's wrong but the handful of pages I took a quick look at all seemed to be much smaller and I'd like to know how the number was arrived at and what assumptions were made)
The physical circuits are long gone in most places but on conventional digital phone networks keeping a call open keeps a reserved slice of bandwidth (a "virtual circuit") for the length of the call. Afaict on the landline side while telcos are starting to roll out VOIP there is still a lot of conventional virtual circuit based network around too. On the mobile side afaict virtually all phones in use today are still using circuit switched voice protocols (LTE is supposed to change this). IIRC circuit switched calls* also have priority over packet switched data traffic at least in GSM.
*You can have a circuit switched data call with GSM but almost noone does it anymore.
Do you really think anyone is going to wonder how many minutes they get out of their data plan when loading a typical web page is equivalent to 5 minutes of talk time?
4 megabytes or so seems insanely high for one web page to me.
The tag does not reflect on it's safety.
Afaict the tag means that they have inspected the plane and are convinced it's safe enough to let you fly it but not convinced it's safe enough to let you run commercial operations with it.
Of course actually being safe and convincing the authorities something is safe are a very different matter.
AIUI ARM do HDL design of processor cores, then they pass that HDL on to other companies who make complete chip designs based on it. Those companies in turn pass the designs onto fabs (which may be in-house or external) for manufacture. IIRC some vendors also do their own HDL work and only license the basic architectural design from ARM.
You didn't make it clear whether you were replying to the first part of my post and being sarcastic or replying to the second part and agreeing with it. I will assume the latter.
The likes of rabbits and cane toads are precisely the kind of species you DO have to worry about. Cane toads produce tiny eggs which lay dormant in the water for some time before hatching. Rabbits hide underground. Both are also small and fast breeding. The result is it's nearly impossible to wipe them out.
Apple aren't really relavent here, they were just using apple's brand to try and make themselves look more legit. I'm sure a legitimate service could find someone to license a brand off if they didn't feel like building their own.
However as you allude to an equally good legitimate service cannot and will not happen (at least not without government intervention). The fundamental problem is that since licenses do cost money and some content owners are likely to either refuse outright or charge unreasonable rates a legitimate service is doomed to end up with both higher prices and less content than a pirate service.