So upgrading to Win7 would mean $100-$150 for the OS, plus $500 for new hardware
Your prices seem a bit inflated to me. According to dell's website you can get an optiplex 390 with a pentium G* CPU, 2GB of ram, a 250GB hard drive and windows 7 pro for $414.
but yeah unless there are obvious productivity, reliability or security issues (and by obvious I mean either something is noticablly slow or breakdowns/breakins happen noticablly often.) or power is VERY expensive in your locality it's going to be difficult to justify upgrading especially in times when cash is tight .
*Don't let the pentium name put you off, the pentium G is a dual core sandy bridge processor with more than enough performance for most tasks.
Do you really think Chileans are going to pay the 400-600 USD an unlocked/unsubsidized phone costs?
Not all phones are that expensive you know and besides customers pay for their phones one way or another, it's just a case of whether they do it explicitly or whether it is hidden in the cost of a cellphone contract.
I think there are two distinct but intertwined issues here
1: allowing carriers to offer subsidised phones in exchange for signing up to a 1-3 year contract. 2: allowing those same carriers to lock the subsidised phone so that even after the contract expires you are still locked in unless you get a new phone
When theese two factors are both present customers are basically forced into paying for a new phone every 1-3 years whether they actually want one or not since moving carrier would mean getting a new phone and your existing carrier has little motivation to lower prices for a customer who can't move without signing up to another phone contract with hidden phone purchase plan. This is hugely wasteful as huge number of mobile phones are made that people wouldn't buy if they had to pay for them directly.
If the government allows 1 but not 2 and assuming the phone networks are technically compatible* people can still get a "subsidised" phone but when their contract expires they can take that phone to any provider. This in turn gives the providers and incentive to offer and compete on cheaper "sim only" phone deals for the newly freed up customers. Phones will get used until people actually want/need to buy a replacement rather than being replaced on an arbitary schedule set by the carriers.
* the US has the additional problem that it's mobile phone networks are a mess of two competing sets of standards (GSM/UMTS verses IS-95/CDMA2000) so unlike most other places even if artificial barriers to taking your phone with you were removed your options for moving would still be limited.
Given the choice of those two chips I would go for the x3 (assuming there are no plans to overclock). Three cores at 3.3GHz are going to beat two cores of the same architecture at 2.2GHz in pretty much any workload. Add that to the fact that afaict the athlon in question is a newer and better architecture than the phenom in question and I really see no reason to go for the x4. Generally you should only start looking to more cores after you hit the point of diminishing returns in speed of individual cores.
Though i'd suggest taking a look at intel's offerings too. The i3-2100 seems to be similar in price and afaict it's better than either of those chips (with hyperthreading and better architecture more than making up for the lower core count and slightly lower clockspeed). This will depend a bit on what pricing is like where you live though.
That sounds like a good reason to be opposed to fossil fuels in general as a power source. As I understand it though coal is worse than oil or gas in this regard.
It would add other confusion of it's own though. For example the confusion of having one workday spread across two calender days in some regions and the confusion when you arrive somewhere of "what time is daytime here".
Funny, I haven't heard of that many deaths from solar power.........
That is because the deaths solar power causes are pretty mundane. Stuff like falling off roofs and similar.
It's like car crashes vs train and plane crashes. Car crashes happen all the time but they are too mundane to be interesting so they aren't usualy mentioned beyond the local news and sometimes not even there. Train and plane crashes kill far fewer people but when one does happen it's big news.
But, when does demand spike ?
When the temperature outside gets furthest from the temperature people want inside. In some places that is hot days, in other places it is cold days.
It really depends on your definition of better. In particular which parameter are you trying to minimise. Lets assume that a designer has a maximum acceptable resistance from their design calculations. Three things they might might be trying to minimise, cross sectional area, mass or cost.
Copper resistivity is 16.78 Ohm-meter, Aluminum is more than 28.
Your figures are nonsensically (10^9 times larger than the figures on wikipedia) but aluminium does have a higher resistivity than copper. However it also has a lower density and a lower cost per tonne.
Note: in the calculations below R is resistance, p is resistivity, A is cross sectional area, L is length, V is volume, d is density amd M is mass.
Let us consider a cable 1 kilometer long and with a resistance of 1 ohm (not an unreasonable design requirement IMO). First let us consider what it's cross sectional area would need to be. R=p(l/A) A=p(l/R) For copper A=16.78*10^-9*(10^3/1)=16.78*10^-6 square meters (16.78 square millimeters) For aluminium A=28.2*10^-9*(10^3/1)=28.2*10^-6 square meters (28.2 square millimeters)
So the copper cable has a smaller CSA. Now lets figure out it's volume V=Al For copper V=16.78*10^-6*10^3=16.78*10^-3 cubic meters For aluminimum V=28.2*10^-6*10^3=28.2*10^-3 cubic meters Now for it's mass M=Vd For copper M=16.78*10^-3*8.94*10^3=150 kilograms For aluminium M=28.2*10^-3*2.70*10^3=76.1 kilograms
So the alumimum cable is just over half the weight of the copper cable.
Add to that the price per ton for copper being over three times higher than the price per ton of aluminium and the aluminium cable is likely to be far cheaper (plastics cost and armouring costs will be a little higher for the copper cable but afaict this is not significant overall).
So the copper cable is smaller but the aluminium is half the weight and far cheaper. Which is "better" depends on your design priorities.
Aluminium is not used much in building wiring despite it's lower cost and lower weight. Afaict the main reason for this is that it needs special termination practices to avoid terminations failing (and often getting dangerously hot in the process). This isn't too much of a problem in an aircraft or a distribution network where you can enforce proper termination practices but it's a big problem in building wiring which is rarely under tight control.
There is no such thing as a "perfect quality" cable.
Yes some applications do need cables with extremely high specifications and/or extremely long lengths and those cables do cost serious money. but you wouldn't be buying such cables at worst buy or indeed from any supplier that doesn't provide proper datasheets with proper specficiations (not buzzword bingo).
Assuming a random statistical distribution of failure
The problem is while some failure will be random others are likely to be caused by things (maybe the driver putting on or losing weight or feeling sore and sitting differenetly because of it) that will cause a retry to fail too.
If the US governement wanted to they could put a lot of pain on bitcoin. For example they could go after the financial transactions with offshore bitcoin exchanges (just as they now go after financial transactions offshore online poker sites). They could announce that anyone caught running an unlicensed bitcoin exchange was open to prosecution and that authorised exchanges were only to accept coins whose history could be traced back to either an authorised miner or to before the introduction of the rules. They could further introduce laws stating that authorised miners and exchanges were required to keep transaction records including bitcoin addresses, that using or operating a "mixing service" was a felony and so on..
Personally I think the reason they haven't done any of this yet is that bitcoin is too small for them to care. Afaict there are currently just under 8 million bitcoins in existence (and some unknown proportion of those bitcoins have been rendered irretrievable as the associated keys are lost) at $30 per bitcoin that gives a theoretical "market cap" of arround $240 million. The actual amount of money involved is probablly a hell of a lot smaller than the aforementioned theoretical market cap.
There is nothing in C90 that forbids 64-bit integers.
It doesn't forbid them but it doesn't standardise them either. Whether they are provided or not and the mechanism used to provide them is up to the individual implementation and as such any code that relies on them becomes implementation dependent.
What is unique and very interesting about the Raz is HDMI output
The beagle/panda series has a HDMI connector (though the signal on it is apparently only DVI, not sure if the same is true of the pi).
What is really unique about the Pi is the price. Afaict the sheevaplug is $99+shipping and the begleboard is arround $140 (the beagleboard is sold through distributors)
As a beagleboard owner afaict the biggest issue is storage, SD cards SUCK at handling the random access workloads that come from tasks like updating packages.
umm, I wouldn't call them "more eager", debian have supported arm since before ubuntu existed and are currently in the process of bringing up a new hardfloat arm port to complement their existing softfloat port.
To me the key features of a "full blown linux box" are
1: enough storage to install a regular linux distro 2: a MMU so you can run proper linux kernel rather than uclinux/uclibc 3: enough ram to reasonablly run the aforementioned regular linux distro.
You have clearly never been to a city with well designed public transport. London's underground will get you very close to your destination and it will get you there quickly...............Manchester is currently rolling out more trams and has train stops in most of the places the trams don't go.
I live in the manchester conurbation and have been to the london conurbation before.
Don't read too much into the metrolink, it saves some end effects for people travelling into the city from it's four lines but mostly it was built by converting existing passenger railway lines (resulting in long shutdowns for users of those lines during the conversion) not by establishing new routes.
Anyway if either where you are coming from or where you are going to is close enough to the city center then the combination of city center traffic congestion and the fact that all railways lead to the city centre mean trains etc* can be a pretty good option.
However lets try some places further out in those conurbations and on different radials. Hell lets give public transport the advantage by planning between stations and by ignoring the time you spend waiting for your trips initial departure..
Try say watford juntion to hertford north** in the london conurbation. Transport-direct tells me 37 minuites by car 1 hour 21 mins at best (and usually more) by public transport Try davenport to ashton-under-lyne in the manchester conurbation. Transport-direct tells me 18 minuites by car, an hour at best (and usually more) by public transport
* trains, mostly seggregates trams, underground etc ** hertford east is even worse.
Well I just looked at the one round here and saw the following
1: it's city center only, I wouldn't want to travel into the city center by public transport just to drive back out again 2: they insist that cars are returned to the same place they are picked up from. That basically means you have to pay for the car for the whole duration of your trip and you can't really use the car to come home if you miss the last bus/train unless you want to pay for a 24 hour rental and return it the following day. 3: it's pretty pricey, at £4.16 per hour or £33.60 for a day (and that's if you buy the discount card) if you are doing one day out each weekend plus an hour or two during the week thats about £2K per year on rentals, there is a mileage charge on top of that. I guess if you can't stand driving an older car or if you find it difficult to get insurance that may actually be appealing but for someone who has been driving for a while and owns their car outright it's probablly more expensive than continuing to own a car.
Are car clubs in other places better or is this fairly typical?
As I understand it the problem with "voting" is that it requires the three systems to produce identical (or very close to identical) results under non-fault conditions. That means you have to write an extremely detailed specification and that in turn means you can end up with all the teams implementing the same bug either because the bug was in the specification or because the specification while strictly correct lead all the teams into making the same mistake.
Trains are great when you have lots of people going to/from the same place. The trouble is in a large conurbation while there are a lot of people going to/from the city center there are also many people who would like to travel between two points further out in the conurbation that are fairly close together but on different radials. Doing this by public transport typically either means catching a slow bus (slow because to get enough passengers to make it viable it has to stop frequently and drive on the slow roads through places rather than the fast roads round places) or taking a very roundabout train route. If you enjoy exploring the countryside it gets even worse with many places effectively cut off from you completely.
It's possible to live without a car but it means planning your life arround public transport (including choosing where you live to have a fast public transport link to where you work) and putting up with the fact that any journeys other than your regular commute (which you chose your place of living based on) are going to be very slow. Especially in the evenings and on sundays when there are less busses and trains.
IMO the only way car ownership and use will significantly reduce is if using a car simply becomes unaffordable for the vast majority of people.
Despite the larger screen it seems that an 11.6 inch macbook air is about the same size and lighter in weight than an EEEPC 1000 and it has a screen with 768 pixels of vertical resoloution a dual core 1.6 GHz (1.8 if you choose the right BTO option) sandy bridge processor (still fairly weak compared to regular laptop processor but far more powefull than an atom).
I see a few advantages of sealed unit watercoolers compared to their heatpipe based competition
1:they tend to be smaller and lighter 2: there is some flexibility in radiator placement (good for those building small form factor systems) 3: they exhaust the CPU heat straight out of the case from the rad rather than relying on general case airflow to take it out 4: a large portion of their weight is mounted on the case rather than the motherboard. That means less risk of damage when moving the machine.
Even if you're not overclocking, water cooling is good
The mid to high end cooling market gave up on conventional heatsinks ages ago. Nowadays they use either heatpipes or watercooling (the so-called "high end air coolers" are heatpipe based). Conventional heatsinks aren't very good at getting lots of heat out of small spaces.
Both methods can take heat quietly from from a small and/or difficult to cool space (PC expansion cards suck from an air cooling perspective because they the motherboard and the connector plate on adjacent sides) and take it to a large radiator that can be more easilly cooled.
The advantage of heatpipes is you don't need a pump.
The advantage of watercooling is flexibility. Heatpipes tend to be (I don't know if they technically have to be) sealed copper pipes, slight bending is possible but there is little flexibility in the system and so the radiator has to be attatched to the pickup. This isn't a problem for a laptop where custom mounts can be designed in but in a conventional desktop it means you end up with the whole cooling assembly bolted to the heatsink mounts on the motherboard. This is bad for robustness and means you can't easilly route the heated air directly out of the case.
Sealed unit watercoolers give a bit more flexibility, enough to bolt the radiator to the outside of the case (which is obviously superior to just having it sit above the motherboard.
Full custom water loops are even more flexible but are more expensive and more hassle.
especially for GPU heavy machines.
Agreed, PC expansion slots were simply never designed for good cooling (if they were they would have the backplane opposite the connector plate rather than adjacent to it). So the only way GPU vendors can make a workable integrated cooling system for high power cards is to use a leafblower like setup.
Unfortunately they don't make third party cooling easy either, pretty much every GPU model and sometimes even different cards with the same GPU needs a different waterblock so you are unlikely to see sealed unit watercooling for GPUs.
They are also very quiet considering the gigantic heat output.
Probablly about a kilowatt or so at most. That is like a fan heater on it's low setting, it's not that gigantic really.
Warhammer models (bought online in the US in the US) seem to range from about $4 per model (sold in packs of 10) for basic troops to about $70 a model for big vehicles. You can buy "starter kits" for about $100. Moulding your warhammer army out of fields metal would probablly be 10 times more expensive than just buying the models.
A toy moulding set that only had enough material for a couple of soilders and cost over $150 or so just wouldn't be attractive because you can get much better presents in that price range.
It's just too damn expensive for toys due to the indium content.
Retail price seems to be over $600 per kilo. Bulk metal prices will be cheaper of course but you are still talking hundreds of dollars per kilo. Looks like a metal toy soilder is about 100g so you are talking about $60 per toy soilder moulded.
In electronics a large part of the cost is upfront so niche products are FAR more expensive than common products of similar complexity. This applies to some extent to assemblies (compare the cost of a PXIe controller module to the cost of a laptop for example) and applies to an even greater extent to semiconductor products.
Powerful FPGAs are a niche product and hence expensive. GPUs are a mass market product and hence cheap. So if a GPU can do your calculations reasonablly efficiently it's probablly the cheaper option.
Slashdot Mirror
So upgrading to Win7 would mean $100-$150 for the OS, plus $500 for new hardware
Your prices seem a bit inflated to me. According to dell's website you can get an optiplex 390 with a pentium G* CPU, 2GB of ram, a 250GB hard drive and windows 7 pro for $414.
but yeah unless there are obvious productivity, reliability or security issues (and by obvious I mean either something is noticablly slow or breakdowns/breakins happen noticablly often.) or power is VERY expensive in your locality it's going to be difficult to justify upgrading especially in times when cash is tight .
*Don't let the pentium name put you off, the pentium G is a dual core sandy bridge processor with more than enough performance for most tasks.
Do you really think Chileans are going to pay the 400-600 USD an unlocked/unsubsidized phone costs?
Not all phones are that expensive you know and besides customers pay for their phones one way or another, it's just a case of whether they do it explicitly or whether it is hidden in the cost of a cellphone contract.
I think there are two distinct but intertwined issues here
1: allowing carriers to offer subsidised phones in exchange for signing up to a 1-3 year contract.
2: allowing those same carriers to lock the subsidised phone so that even after the contract expires you are still locked in unless you get a new phone
When theese two factors are both present customers are basically forced into paying for a new phone every 1-3 years whether they actually want one or not since moving carrier would mean getting a new phone and your existing carrier has little motivation to lower prices for a customer who can't move without signing up to another phone contract with hidden phone purchase plan. This is hugely wasteful as huge number of mobile phones are made that people wouldn't buy if they had to pay for them directly.
If the government allows 1 but not 2 and assuming the phone networks are technically compatible* people can still get a "subsidised" phone but when their contract expires they can take that phone to any provider. This in turn gives the providers and incentive to offer and compete on cheaper "sim only" phone deals for the newly freed up customers. Phones will get used until people actually want/need to buy a replacement rather than being replaced on an arbitary schedule set by the carriers.
* the US has the additional problem that it's mobile phone networks are a mess of two competing sets of standards (GSM/UMTS verses IS-95/CDMA2000) so unlike most other places even if artificial barriers to taking your phone with you were removed your options for moving would still be limited.
Given the choice of those two chips I would go for the x3 (assuming there are no plans to overclock). Three cores at 3.3GHz are going to beat two cores of the same architecture at 2.2GHz in pretty much any workload. Add that to the fact that afaict the athlon in question is a newer and better architecture than the phenom in question and I really see no reason to go for the x4. Generally you should only start looking to more cores after you hit the point of diminishing returns in speed of individual cores.
Though i'd suggest taking a look at intel's offerings too. The i3-2100 seems to be similar in price and afaict it's better than either of those chips (with hyperthreading and better architecture more than making up for the lower core count and slightly lower clockspeed). This will depend a bit on what pricing is like where you live though.
That sounds like a good reason to be opposed to fossil fuels in general as a power source. As I understand it though coal is worse than oil or gas in this regard.
It would add other confusion of it's own though. For example the confusion of having one workday spread across two calender days in some regions and the confusion when you arrive somewhere of "what time is daytime here".
Funny, I haven't heard of that many deaths from solar power .........
That is because the deaths solar power causes are pretty mundane. Stuff like falling off roofs and similar.
It's like car crashes vs train and plane crashes. Car crashes happen all the time but they are too mundane to be interesting so they aren't usualy mentioned beyond the local news and sometimes not even there. Train and plane crashes kill far fewer people but when one does happen it's big news.
But, when does demand spike ?
When the temperature outside gets furthest from the temperature people want inside. In some places that is hot days, in other places it is cold days.
Aluminum is NOT a better conductor.
It really depends on your definition of better. In particular which parameter are you trying to minimise. Lets assume that a designer has a maximum acceptable resistance from their design calculations. Three things they might might be trying to minimise, cross sectional area, mass or cost.
Copper resistivity is 16.78 Ohm-meter, Aluminum is more than 28.
Your figures are nonsensically (10^9 times larger than the figures on wikipedia) but aluminium does have a higher resistivity than copper. However it also has a lower density and a lower cost per tonne.
Note: in the calculations below R is resistance, p is resistivity, A is cross sectional area, L is length, V is volume, d is density amd M is mass.
Let us consider a cable 1 kilometer long and with a resistance of 1 ohm (not an unreasonable design requirement IMO). First let us consider what it's cross sectional area would need to be.
R=p(l/A)
A=p(l/R)
For copper
A=16.78*10^-9*(10^3/1)=16.78*10^-6 square meters (16.78 square millimeters)
For aluminium
A=28.2*10^-9*(10^3/1)=28.2*10^-6 square meters (28.2 square millimeters)
So the copper cable has a smaller CSA. Now lets figure out it's volume
V=Al
For copper
V=16.78*10^-6*10^3=16.78*10^-3 cubic meters
For aluminimum
V=28.2*10^-6*10^3=28.2*10^-3 cubic meters
Now for it's mass
M=Vd
For copper
M=16.78*10^-3*8.94*10^3=150 kilograms
For aluminium
M=28.2*10^-3*2.70*10^3=76.1 kilograms
So the alumimum cable is just over half the weight of the copper cable.
Add to that the price per ton for copper being over three times higher than the price per ton of aluminium and the aluminium cable is likely to be far cheaper (plastics cost and armouring costs will be a little higher for the copper cable but afaict this is not significant overall).
So the copper cable is smaller but the aluminium is half the weight and far cheaper. Which is "better" depends on your design priorities.
Aluminium is not used much in building wiring despite it's lower cost and lower weight. Afaict the main reason for this is that it needs special termination practices to avoid terminations failing (and often getting dangerously hot in the process). This isn't too much of a problem in an aircraft or a distribution network where you can enforce proper termination practices but it's a big problem in building wiring which is rarely under tight control.
There is no such thing as a "perfect quality" cable.
Yes some applications do need cables with extremely high specifications and/or extremely long lengths and those cables do cost serious money. but you wouldn't be buying such cables at worst buy or indeed from any supplier that doesn't provide proper datasheets with proper specficiations (not buzzword bingo).
Assuming a random statistical distribution of failure
The problem is while some failure will be random others are likely to be caused by things (maybe the driver putting on or losing weight or feeling sore and sitting differenetly because of it) that will cause a retry to fail too.
If the US governement wanted to they could put a lot of pain on bitcoin. For example they could go after the financial transactions with offshore bitcoin exchanges (just as they now go after financial transactions offshore online poker sites). They could announce that anyone caught running an unlicensed bitcoin exchange was open to prosecution and that authorised exchanges were only to accept coins whose history could be traced back to either an authorised miner or to before the introduction of the rules. They could further introduce laws stating that authorised miners and exchanges were required to keep transaction records including bitcoin addresses, that using or operating a "mixing service" was a felony and so on..
Personally I think the reason they haven't done any of this yet is that bitcoin is too small for them to care. Afaict there are currently just under 8 million bitcoins in existence (and some unknown proportion of those bitcoins have been rendered irretrievable as the associated keys are lost) at $30 per bitcoin that gives a theoretical "market cap" of arround $240 million. The actual amount of money involved is probablly a hell of a lot smaller than the aforementioned theoretical market cap.
There is nothing in C90 that forbids 64-bit integers.
It doesn't forbid them but it doesn't standardise them either. Whether they are provided or not and the mechanism used to provide them is up to the individual implementation and as such any code that relies on them becomes implementation dependent.
What is unique and very interesting about the Raz is HDMI output
The beagle/panda series has a HDMI connector (though the signal on it is apparently only DVI, not sure if the same is true of the pi).
What is really unique about the Pi is the price. Afaict the sheevaplug is $99+shipping and the begleboard is arround $140 (the beagleboard is sold through distributors)
As a beagleboard owner afaict the biggest issue is storage, SD cards SUCK at handling the random access workloads that come from tasks like updating packages.
umm, I wouldn't call them "more eager", debian have supported arm since before ubuntu existed and are currently in the process of bringing up a new hardfloat arm port to complement their existing softfloat port.
To me the key features of a "full blown linux box" are
1: enough storage to install a regular linux distro
2: a MMU so you can run proper linux kernel rather than uclinux/uclibc
3: enough ram to reasonablly run the aforementioned regular linux distro.
The pi has all of those
You have clearly never been to a city with well designed public transport. London's underground will get you very close to your destination and it will get you there quickly...............Manchester is currently rolling out more trams and has train stops in most of the places the trams don't go.
I live in the manchester conurbation and have been to the london conurbation before.
Don't read too much into the metrolink, it saves some end effects for people travelling into the city from it's four lines but mostly it was built by converting existing passenger railway lines (resulting in long shutdowns for users of those lines during the conversion) not by establishing new routes.
Anyway if either where you are coming from or where you are going to is close enough to the city center then the combination of city center traffic congestion and the fact that all railways lead to the city centre mean trains etc* can be a pretty good option.
However lets try some places further out in those conurbations and on different radials. Hell lets give public transport the advantage by planning between stations and by ignoring the time you spend waiting for your trips initial departure..
Try say watford juntion to hertford north** in the london conurbation. Transport-direct tells me 37 minuites by car 1 hour 21 mins at best (and usually more) by public transport
Try davenport to ashton-under-lyne in the manchester conurbation. Transport-direct tells me 18 minuites by car, an hour at best (and usually more) by public transport
* trains, mostly seggregates trams, underground etc
** hertford east is even worse.
Well I just looked at the one round here and saw the following
1: it's city center only, I wouldn't want to travel into the city center by public transport just to drive back out again
2: they insist that cars are returned to the same place they are picked up from. That basically means you have to pay for the car for the whole duration of your trip and you can't really use the car to come home if you miss the last bus/train unless you want to pay for a 24 hour rental and return it the following day.
3: it's pretty pricey, at £4.16 per hour or £33.60 for a day (and that's if you buy the discount card) if you are doing one day out each weekend plus an hour or two during the week thats about £2K per year on rentals, there is a mileage charge on top of that. I guess if you can't stand driving an older car or if you find it difficult to get insurance that may actually be appealing but for someone who has been driving for a while and owns their car outright it's probablly more expensive than continuing to own a car.
Are car clubs in other places better or is this fairly typical?
As I understand it the problem with "voting" is that it requires the three systems to produce identical (or very close to identical) results under non-fault conditions. That means you have to write an extremely detailed specification and that in turn means you can end up with all the teams implementing the same bug either because the bug was in the specification or because the specification while strictly correct lead all the teams into making the same mistake.
Trains are great when you have lots of people going to/from the same place. The trouble is in a large conurbation while there are a lot of people going to/from the city center there are also many people who would like to travel between two points further out in the conurbation that are fairly close together but on different radials. Doing this by public transport typically either means catching a slow bus (slow because to get enough passengers to make it viable it has to stop frequently and drive on the slow roads through places rather than the fast roads round places) or taking a very roundabout train route. If you enjoy exploring the countryside it gets even worse with many places effectively cut off from you completely.
It's possible to live without a car but it means planning your life arround public transport (including choosing where you live to have a fast public transport link to where you work) and putting up with the fact that any journeys other than your regular commute (which you chose your place of living based on) are going to be very slow. Especially in the evenings and on sundays when there are less busses and trains.
IMO the only way car ownership and use will significantly reduce is if using a car simply becomes unaffordable for the vast majority of people.
Despite the larger screen it seems that an 11.6 inch macbook air is about the same size and lighter in weight than an EEEPC 1000 and it has a screen with 768 pixels of vertical resoloution a dual core 1.6 GHz (1.8 if you choose the right BTO option) sandy bridge processor (still fairly weak compared to regular laptop processor but far more powefull than an atom).
PCMCIA was originally designed as a memory card form factor. It was later thickened up for use as an expansion card form factor.
Also I think being thicker would have doomed it sooner as laptops got thinner.
I see a few advantages of sealed unit watercoolers compared to their heatpipe based competition
1:they tend to be smaller and lighter
2: there is some flexibility in radiator placement (good for those building small form factor systems)
3: they exhaust the CPU heat straight out of the case from the rad rather than relying on general case airflow to take it out
4: a large portion of their weight is mounted on the case rather than the motherboard. That means less risk of damage when moving the machine.
Even if you're not overclocking, water cooling is good
The mid to high end cooling market gave up on conventional heatsinks ages ago. Nowadays they use either heatpipes or watercooling (the so-called "high end air coolers" are heatpipe based). Conventional heatsinks aren't very good at getting lots of heat out of small spaces.
Both methods can take heat quietly from from a small and/or difficult to cool space (PC expansion cards suck from an air cooling perspective because they the motherboard and the connector plate on adjacent sides) and take it to a large radiator that can be more easilly cooled.
The advantage of heatpipes is you don't need a pump.
The advantage of watercooling is flexibility. Heatpipes tend to be (I don't know if they technically have to be) sealed copper pipes, slight bending is possible but there is little flexibility in the system and so the radiator has to be attatched to the pickup. This isn't a problem for a laptop where custom mounts can be designed in but in a conventional desktop it means you end up with the whole cooling assembly bolted to the heatsink mounts on the motherboard. This is bad for robustness and means you can't easilly route the heated air directly out of the case.
Sealed unit watercoolers give a bit more flexibility, enough to bolt the radiator to the outside of the case (which is obviously superior to just having it sit above the motherboard.
Full custom water loops are even more flexible but are more expensive and more hassle.
especially for GPU heavy machines.
Agreed, PC expansion slots were simply never designed for good cooling (if they were they would have the backplane opposite the connector plate rather than adjacent to it). So the only way GPU vendors can make a workable integrated cooling system for high power cards is to use a leafblower like setup.
Unfortunately they don't make third party cooling easy either, pretty much every GPU model and sometimes even different cards with the same GPU needs a different waterblock so you are unlikely to see sealed unit watercooling for GPUs.
They are also very quiet considering the gigantic heat output.
Probablly about a kilowatt or so at most. That is like a fan heater on it's low setting, it's not that gigantic really.
Warhammer models (bought online in the US in the US) seem to range from about $4 per model (sold in packs of 10) for basic troops to about $70 a model for big vehicles. You can buy "starter kits" for about $100. Moulding your warhammer army out of fields metal would probablly be 10 times more expensive than just buying the models.
A toy moulding set that only had enough material for a couple of soilders and cost over $150 or so just wouldn't be attractive because you can get much better presents in that price range.
It's just too damn expensive for toys due to the indium content.
Retail price seems to be over $600 per kilo. Bulk metal prices will be cheaper of course but you are still talking hundreds of dollars per kilo. Looks like a metal toy soilder is about 100g so you are talking about $60 per toy soilder moulded.
In electronics a large part of the cost is upfront so niche products are FAR more expensive than common products of similar complexity. This applies to some extent to assemblies (compare the cost of a PXIe controller module to the cost of a laptop for example) and applies to an even greater extent to semiconductor products.
Powerful FPGAs are a niche product and hence expensive. GPUs are a mass market product and hence cheap. So if a GPU can do your calculations reasonablly efficiently it's probablly the cheaper option.