People who want a bunch of computing resources, can't/don't want to operate it remotely and don't work in a fixed location.
It's a lot more conviniant to grab a big laptop from the back of your car and set it up on a desk at a temporary work location then to do the same with a desktop and it's associated collection of perhiperals and cables.
They do limit simulationus streams, but that only partially solves the problem.
Firstly a limit of one simultaneous stream still allows sharing if people live on different schedules. Secondly they know that families won't pay four times as much as single men. So following what the cable/sattelite providers do they allow an account to be shared within a household with a relatively small up charge for allowing multiple streams at once.
The problem for these streaming services is defining "same household", for the cable providers it was easy, they only run a cable to one house. For the sattelite providers they can insist all your decoder boxes are connected to the same phone line, but one of these streaming services big selling points is you can use your existing equpiment and take them with you on the go.
So the question becomes how do you detect people who are sharing accounts in violation of the terms and conditions while minimizing the impact on those who are operating within the rules
Well there are some pretty distinct generations, which each had different driving goals and had incompatible radio layers.
1G: analog mobile phones. 2G: digital mobile phones, primerally voice but with some data services. 3G: packet data as a design driver but circuit-switched services still a major part of the offerting. 4G: everything is packet data, voice and text are services that run on top of (high-priority) data service. 5G: move to higher frequency shorter distance radio to massively increase data capacity.
In between the major generations there were incremental improvements GSM got packet data services bolted on (intially the only data service was circuit-switched and very slow) and then they got the data rates of that service cranked up. 3G got HSPA and HSPA+ which cranked up the data rates a bit. In the USA it seems these often got marketed as being the generation after the one they really belonged to.
The total revenue of all ISPs in the world is apparently about 600 billion dollars per year. A falcon 9 launch costs about 60 million dollars. So total worldwide ISP revenue could pay for about ten thousand launches per year. Lets ignore the cost of the sattelites themselves and the availability of radio channels to communicate with them. Lets say that goes on for 10 years making a hundred thousand launches. Lets say each one launches 25 small sattelites that weigh a ton each and are about the size of a car with an area of about 5 square meters.
So in our somewhat ludicrous scenario we have a total of 1.25*10^7 square meters of satellites. The surface area of the earth is about 5*10^14 square meters.
Satellites won't be blocking out the sun any time soon.
As someone who lives in the UK afaict UK prime is better than us prime. AIUI US prime only gives free 2-day delivery while UK prime gives free next day or if you prefer nominated day on most Amazon-stocked items (some items seem to have more restrictive options, probablly as a result of being in different warehouses). Also unlike most next-day and nominated day services Amazon offers weekend deliveries at no extra cost.
Prime is great when you are doing something like preparing for an event when you want to order a bunch of stuff over a short period (but not all at once because you inevitablly find you need more stuff as the preperations proceed). I haven't yet found motivation to stay subscribed all the time though, I just buy a month when I want it.
I know amazon UK has a video service, but I dunno if it's better or worse than the US one, I haven't tried it personally.
Fortunately 99% of the infrastructure is already there
I'm not at all convinced of that. From some quick searching is seems a house in the UK averages about 500 watts (it peaks much higher than that of course). The average driver drives about 8000 miles per year which works out to and a Tesla model 3 supposedly takes 26KWH per 100 miles which works out to about 2000 KWH per year or about 230 watts on average and there are about 1.3 cars per household so that is an extra 300 watts per household. That does not seem like an insignificant increase to me.
If charging is done the "dumb" way I would expect the peaks from EV charging to line up with other peaks in residential demand, People are likely to plug in their EV at about the same time they start cooking their dinner. I think if EV charging is going to work on a large scale without requiring major grid reinforcements then EV chargers will need to be smarter and tied into time of day based electricity pricing.
It is relatively easy to program robots to move standardised containers around a well-ordered warehouse. It is much harder to deal with the massive variation of individual items. So when an item needs to be moved between containers (including but not limited to ingress and egress) that is done by humans.
Companies are going to want to decide if Computer Calls count as legitimate calls or not.
And ultimately that is going to come down to relative volumes. If most of the calls coming through this new google service are legitimate then companies will just tolerate the bullshit ones just like they tolerate bullshit from humans. If most of the calls coming through are time/money wasting bullshit then they will probablly start hanging up on them as soon as they hear the calls are from google.
The question will be can google open this service up so most normal people can use it while at the same time excluding troublemakers from using it.
The USA and EU (at least the parts of it i'm familiar with) set up their phone systems differently.
In the USA mobile phones get ordinary geopgrahic phone numbers and the amount charged by the receiving telco to the originating telco is the same for landlines and mobiles. The recipiant pays (either explicitly or as part of the cost of their plan) for the call to be delievered from the terrestrial phone network to their mobile.
In most if not all of the EU mobile phones get phone numbers from a special range. The recipiant doesn't pay anything for incoming calls (unless they are roaming outside Europe or are diverting a landline number to a mobile or some other unusal corner case). Instead the originating telco pays more for a call to a mobile than for a call to a landline.
This has two effects, firstly there is the direct cost impact on the telemarketers. Secondly it means they can't claim (truthfully or otherwise) ignorance about the fact that the number they are calling is a mobile.
The USB controller on "A series" models is only used for the single USB port. On "B series" models it is used for a USB hub with ethernet chip that drives the USB and Ethernet ports.
The SD card, HDMI, DSI and CSI interfaces and the interfaces available on the GPIO header go directly to the SoC. The wifi is connected to the SoC by SDIO. The bluetooth is connected to the SoC via serial.
The original 256M Pi1 A+ was replaced by the 512M Pi1 A+ which is still available. Farnell. CPC, RS, Allied and newark are all showing it as in-stock.
As for power draw according to https://medium.com/@ghalfacree... the idle power draw is significantly higher than an A+ but lower than any B-series pi. The full-load power draw is lower than a 3B+ but higher than everything else.
Yeah, I was looking at the price for a complete system.
Looking at the bits seperately, the CPU prices do indeed seem reasonable, the mainboard prices on the other hand make small core count systems prohibitively expensive.
Basically it went pretty much nowhere until after the major RIRs ran out of IPv4 addresses for regular allocations (they held back a small pool for special purposes), then it picked up substantially.
postgresql seems to be the go-to FOSS relational database for those who care about data integrity, it's been around for decades and there seem to be plenty of paid support options if you need them. I am told it doesn't scale as well as Oracle or DB2 though.
The best glide speed (if it's making an emergency landing on a road then it probablly has an engine failure) of a small light aircraft is in the same ballpark as highway traffic speeds, so the relative speeds are likely to be managable as long as the road isn't too busy.
Aside from the fact that noone reads the manual that paragraph is largely useless because
1. it doesn't define "high concentrations" or "near". 2. noone bothers to monitor for helium because it is regarded as benign. If you were in a place where there could be particularlly big releases you might monitor for oxygen displacement, but from some googling it seems that oxygen depletion sensors warn at 19% and alarm at 17%, normal oxygen level is about 21%, so you are talking about 9.5% of the air having to be displaced before an oxygen displacement meter warns and about 18% of the air having to be displaced before it alarms.
shit we've perfected time telling technology since like 100 years in the paleolithic
bullshit.
like cave dudes had a triangle that axed the sun what time it was
Which only works during the daytime and requires a scale whose location relative to compass directions is surveyed.
how do you fuck up a watch
An apple watch isn't just a watch, just like a smartphone isn't just a phone. It's a complex multifunction cellular-connected device, that means much more opertunity for bugs and a desire for the manufactuerer to add features over time. That means firmware updates.
Sometimes firmware updates go wrong, vendors can reduce the probability and impact with internal testing and with staged rollouts but on a sufficiently large and varied userbase the occasional fuckup due to a combination of factors not covered in testing is pretty much inevitable.
Using BGP is the normal way routes are exchanged between carriers on the Internet. It is absoloutely normal for carriers in different countries to have BGP sessions with each other.
The problem is a combination of laziness and resource limitations mean that carriers and other networks end up trusting each other. Sure filters can be put in place in theory but on a link where thousands of prefixes are normally exchanged maintaining those filters is both a a PITA and a resource drain on the routers.
Adding to that many networks are cheapskates. Rather than take the shortest path to a destination they will take the cheapest. i.e. they will prefer sending the traffic to a peer or downstream over sending it to an upstream.
The result of this is it's easy for traffic to get diverted, either accidentally or maliciously, and as long as the traffic reaches it's destination without undue delays it is very likely that no one will notice.
What you can't tell on an encrypted communications channel (without finding a way to bypass the encyrption) is what is being sent. In particular whether the information being sent is the minimum required to complete the action at hand or whether additional data is being exfiltrated.
One way this can happen is through repeated failure.
You try to transfer something, the transfer nearly completes but then fails for some reason. When you re-try it it fails again for the same reason. If you don't have a mechanism to limit the number of retries or implement exponential backoff then you can cause data traffic many times the size of the data you are trying and failing to transfer.
I had this happen with email once, the sender was using nullmailer and trying to send mails that were over the recieving server's size limit. The receiving server (postfix IIRC) would reject the mail, but only after it had been sent over the network. nullmailer at the time had no concept of a permanent failure or exponential back-off and just kept trying to send the mail once per minuite.
Though the 3 man crew of apollo 1 died in a space rocket on the launch pad.
And the 3 man crew of apollo 13 only survived because of on the fly improvisation to allow the command module's CO2 scrubber cartridges to be used in the lunar module.
until the fucking crapfest the shuttles were.
According to Wikipedia there were 33 manned American space flights in the pre-shuttle era and during that era one crew died in an American spacecraft (though admittedly not during a space flight).
Compare to the 135 flights with two loss of crew incidents of the shuttle and the 139 manned flights with two loss of crew incidents of the soyuz.
Challenger died due to ignoring 1 and 2.
Agreed
Columbia due to a lack of a Plan B, let alone Plan C
I am more inclined to blame the side-mount stack that left the heat shield in a vulnerable location.
Heat shield failure is something that is very difficult to plan for. If you find the problem after you enter the atmosphere it's too late to do anything about it. Even if you find the problem before re-entering you are still screwed unless you have a space-station handy or a rescue vessel on stand-by.
Slashdot Mirror
People who want a bunch of computing resources, can't/don't want to operate it remotely and don't work in a fixed location.
It's a lot more conviniant to grab a big laptop from the back of your car and set it up on a desk at a temporary work location then to do the same with a desktop and it's associated collection of perhiperals and cables.
They do limit simulationus streams, but that only partially solves the problem.
Firstly a limit of one simultaneous stream still allows sharing if people live on different schedules. Secondly they know that families won't pay four times as much as single men. So following what the cable/sattelite providers do they allow an account to be shared within a household with a relatively small up charge for allowing multiple streams at once.
The problem for these streaming services is defining "same household", for the cable providers it was easy, they only run a cable to one house. For the sattelite providers they can insist all your decoder boxes are connected to the same phone line, but one of these streaming services big selling points is you can use your existing equpiment and take them with you on the go.
So the question becomes how do you detect people who are sharing accounts in violation of the terms and conditions while minimizing the impact on those who are operating within the rules
Well there are some pretty distinct generations, which each had different driving goals and had incompatible radio layers.
1G: analog mobile phones.
2G: digital mobile phones, primerally voice but with some data services.
3G: packet data as a design driver but circuit-switched services still a major part of the offerting.
4G: everything is packet data, voice and text are services that run on top of (high-priority) data service.
5G: move to higher frequency shorter distance radio to massively increase data capacity.
In between the major generations there were incremental improvements GSM got packet data services bolted on (intially the only data service was circuit-switched and very slow) and then they got the data rates of that service cranked up. 3G got HSPA and HSPA+ which cranked up the data rates a bit. In the USA it seems these often got marketed as being the generation after the one they really belonged to.
Lets do some back of the envelope calculations.
The total revenue of all ISPs in the world is apparently about 600 billion dollars per year. A falcon 9 launch costs about 60 million dollars. So total worldwide ISP revenue could pay for about ten thousand launches per year. Lets ignore the cost of the sattelites themselves and the availability of radio channels to communicate with them. Lets say that goes on for 10 years making a hundred thousand launches. Lets say each one launches 25 small sattelites that weigh a ton each and are about the size of a car with an area of about 5 square meters.
So in our somewhat ludicrous scenario we have a total of 1.25*10^7 square meters of satellites. The surface area of the earth is about 5*10^14 square meters.
Satellites won't be blocking out the sun any time soon.
As someone who lives in the UK afaict UK prime is better than us prime. AIUI US prime only gives free 2-day delivery while UK prime gives free next day or if you prefer nominated day on most Amazon-stocked items (some items seem to have more restrictive options, probablly as a result of being in different warehouses). Also unlike most next-day and nominated day services Amazon offers weekend deliveries at no extra cost.
Prime is great when you are doing something like preparing for an event when you want to order a bunch of stuff over a short period (but not all at once because you inevitablly find you need more stuff as the preperations proceed). I haven't yet found motivation to stay subscribed all the time though, I just buy a month when I want it.
I know amazon UK has a video service, but I dunno if it's better or worse than the US one, I haven't tried it personally.
Debian has an unofficial x32 port. I don't know how usable it is though.
Fortunately 99% of the infrastructure is already there
I'm not at all convinced of that. From some quick searching is seems a house in the UK averages about 500 watts (it peaks much higher than that of course). The average driver drives about 8000 miles per year which works out to and a Tesla model 3 supposedly takes 26KWH per 100 miles which works out to about 2000 KWH per year or about 230 watts on average and there are about 1.3 cars per household so that is an extra 300 watts per household. That does not seem like an insignificant increase to me.
If charging is done the "dumb" way I would expect the peaks from EV charging to line up with other peaks in residential demand, People are likely to plug in their EV at about the same time they start cooking their dinner. I think if EV charging is going to work on a large scale without requiring major grid reinforcements then EV chargers will need to be smarter and tied into time of day based electricity pricing.
why there are any humans in warehouses.
Simple, because humans are more flexible.
It is relatively easy to program robots to move standardised containers around a well-ordered warehouse. It is much harder to deal with the massive variation of individual items. So when an item needs to be moved between containers (including but not limited to ingress and egress) that is done by humans.
Intel is out of the CPU biz now in all sectors
Lol, intel and AMD are the only CPU vendors with any significant presence in the mid range.
Companies are going to want to decide if Computer Calls count as legitimate calls or not.
And ultimately that is going to come down to relative volumes. If most of the calls coming through this new google service are legitimate then companies will just tolerate the bullshit ones just like they tolerate bullshit from humans. If most of the calls coming through are time/money wasting bullshit then they will probablly start hanging up on them as soon as they hear the calls are from google.
The question will be can google open this service up so most normal people can use it while at the same time excluding troublemakers from using it.
The USA and EU (at least the parts of it i'm familiar with) set up their phone systems differently.
In the USA mobile phones get ordinary geopgrahic phone numbers and the amount charged by the receiving telco to the originating telco is the same for landlines and mobiles. The recipiant pays (either explicitly or as part of the cost of their plan) for the call to be delievered from the terrestrial phone network to their mobile.
In most if not all of the EU mobile phones get phone numbers from a special range. The recipiant doesn't pay anything for incoming calls (unless they are roaming outside Europe or are diverting a landline number to a mobile or some other unusal corner case). Instead the originating telco pays more for a call to a mobile than for a call to a landline.
This has two effects, firstly there is the direct cost impact on the telemarketers. Secondly it means they can't claim (truthfully or otherwise) ignorance about the fact that the number they are calling is a mobile.
The USB controller on "A series" models is only used for the single USB port. On "B series" models it is used for a USB hub with ethernet chip that drives the USB and Ethernet ports.
The SD card, HDMI, DSI and CSI interfaces and the interfaces available on the GPIO header go directly to the SoC. The wifi is connected to the SoC by SDIO. The bluetooth is connected to the SoC via serial.
The original 256M Pi1 A+ was replaced by the 512M Pi1 A+ which is still available. Farnell. CPC, RS, Allied and newark are all showing it as in-stock.
As for power draw according to https://medium.com/@ghalfacree... the idle power draw is significantly higher than an A+ but lower than any B-series pi. The full-load power draw is lower than a 3B+ but higher than everything else.
Yeah, I was looking at the price for a complete system.
Looking at the bits seperately, the CPU prices do indeed seem reasonable, the mainboard prices on the other hand make small core count systems prohibitively expensive.
How's that been been going?
Basically it went pretty much nowhere until after the major RIRs ran out of IPv4 addresses for regular allocations (they held back a small pool for special purposes), then it picked up substantially.
How big is big?
postgresql seems to be the go-to FOSS relational database for those who care about data integrity, it's been around for decades and there seem to be plenty of paid support options if you need them. I am told it doesn't scale as well as Oracle or DB2 though.
https://www.raptorcs.com/conte...
Expensive certainly but not totally impossible for a "mere mortal" to buy.
The best glide speed (if it's making an emergency landing on a road then it probablly has an engine failure) of a small light aircraft is in the same ballpark as highway traffic speeds, so the relative speeds are likely to be managable as long as the road isn't too busy.
Aside from the fact that noone reads the manual that paragraph is largely useless because
1. it doesn't define "high concentrations" or "near".
2. noone bothers to monitor for helium because it is regarded as benign. If you were in a place where there could be particularlly big releases you might monitor for oxygen displacement, but from some googling it seems that oxygen depletion sensors warn at 19% and alarm at 17%, normal oxygen level is about 21%, so you are talking about 9.5% of the air having to be displaced before an oxygen displacement meter warns and about 18% of the air having to be displaced before it alarms.
shit we've perfected time telling technology since like 100 years in the paleolithic
bullshit.
like cave dudes had a triangle that axed the sun what time it was
Which only works during the daytime and requires a scale whose location relative to compass directions is surveyed.
how do you fuck up a watch
An apple watch isn't just a watch, just like a smartphone isn't just a phone. It's a complex multifunction cellular-connected device, that means much more opertunity for bugs and a desire for the manufactuerer to add features over time. That means firmware updates.
Sometimes firmware updates go wrong, vendors can reduce the probability and impact with internal testing and with staged rollouts but on a sufficiently large and varied userbase the occasional fuckup due to a combination of factors not covered in testing is pretty much inevitable.
Using BGP is the normal way routes are exchanged between carriers on the Internet. It is absoloutely normal for carriers in different countries to have BGP sessions with each other.
The problem is a combination of laziness and resource limitations mean that carriers and other networks end up trusting each other. Sure filters can be put in place in theory but on a link where thousands of prefixes are normally exchanged maintaining those filters is both a a PITA and a resource drain on the routers.
Adding to that many networks are cheapskates. Rather than take the shortest path to a destination they will take the cheapest. i.e. they will prefer sending the traffic to a peer or downstream over sending it to an upstream.
The result of this is it's easy for traffic to get diverted, either accidentally or maliciously, and as long as the traffic reaches it's destination without undue delays it is very likely that no one will notice.
What you can't tell on an encrypted communications channel (without finding a way to bypass the encyrption) is what is being sent. In particular whether the information being sent is the minimum required to complete the action at hand or whether additional data is being exfiltrated.
One way this can happen is through repeated failure.
You try to transfer something, the transfer nearly completes but then fails for some reason. When you re-try it it fails again for the same reason. If you don't have a mechanism to limit the number of retries or implement exponential backoff then you can cause data traffic many times the size of the data you are trying and failing to transfer.
I had this happen with email once, the sender was using nullmailer and trying to send mails that were over the recieving server's size limit. The receiving server (postfix IIRC) would reject the mail, but only after it had been sent over the network. nullmailer at the time had no concept of a permanent failure or exponential back-off and just kept trying to send the mail once per minuite.
Unbalanced is putting it nicely, the spectrum of low pressure sodium lamps is practically monochromatic.
nobody died in a space flight in the US
Though the 3 man crew of apollo 1 died in a space rocket on the launch pad.
And the 3 man crew of apollo 13 only survived because of on the fly improvisation to allow the command module's CO2 scrubber cartridges to be used in the lunar module.
until the fucking crapfest the shuttles were.
According to Wikipedia there were 33 manned American space flights in the pre-shuttle era and during that era one crew died in an American spacecraft (though admittedly not during a space flight).
Compare to the 135 flights with two loss of crew incidents of the shuttle and the 139 manned flights with two loss of crew incidents of the soyuz.
Challenger died due to ignoring 1 and 2.
Agreed
Columbia due to a lack of a Plan B, let alone Plan C
I am more inclined to blame the side-mount stack that left the heat shield in a vulnerable location.
Heat shield failure is something that is very difficult to plan for. If you find the problem after you enter the atmosphere it's too late to do anything about it. Even if you find the problem before re-entering you are still screwed unless you have a space-station handy or a rescue vessel on stand-by.