However, most of the world can buy a used desktop for around $50 if they want - sure its not the same as a new upmarket model, but I would prefer a P4 based desktop with a screen and keyboard (not running Windows, obviously) over a $50 tablet, new or used, if the object was to create stuff.
Hell is something close to debugging PHP with embedded SQL on a low resolution tablet.
And yes, I have been to Africa recently, and yes I could get a P4 with a CRT and PS2 keyboard and mouse
for under $50 (it did look past its prime though). I could also buy a Nigerian Guinness for about $0.30 and a nutritiously sound meal for about $1.50. An experienced local would obviously pay less than me for the meal or the computer, unless he had drunk too much Nigerian Guinness.
A lot of people there already had $50 tablets two years ago. Some even had PCs with Linux.
The problem in Africa is not access to hardware, it is, to some degree, understanding the benefits of the hardware (particularly as compared to the merits of dressing up and partying). However, you could access mainframes in 1963 here in the UK. How many people had a use for a mainframe in 1963? Hell, how many people would have known what one did, even if they were in the computer room? (it was enough to make Ross Perot filthy rich). However, the clothes and parties here in the UK in 1963 were pretty crap unless you were a cabinet minister (see Profumo).
The solution to this problem is time not hardware.
Yes, but if you don't have one at home, then the implication is you have one outside, and quite possibly there is someone you would prefer does not know - some black hats like this kind of opportunity. Some data stores are not particularly secure.
OTOH, you might be doing a favour for a neighbour, and get wrongly accused by the blackmailer - this could lead to a major opportunity for a crime drama - anyone have the number for Bellisarios?
Nearly every modern serious data storage (even some high-range SD flash cards: see Transcend) uses some form of error correction.
You say it like its a good thing!
Error correction works fine for one, or possibly a small number of errors, such as you might get in DRAM, but if you get a lot of errors like on a bad disk or tape, it is capable of munging the data and declaring it fixed. And there is no way to know how many errors you have got. If you have errors, you get another tape out of the cupboard. (You do have more than one backup, don't you?) SD cards? how would you know what's going on inside?
You do not get an algorithm to "fix" the data if its your life on the line (or your $$$). Of course, if its only prÃn, then maybe the fixing might even improve it.
I can still recall the horrors of 200bpi 7-track 1/2" tape NRZI with the "choose your own parity" feature.
The big advantage of 200bpi was that you could sprinkle iron filings on the tape and read the bit patterns for disaster recovery. (Not that I would want to read more than a couple of 80 column card images that way).
Here's a secret about tape; you drop it, it's as fragile as a HDD.
NO, its not. If you drop it, the puck (the bit that the loading mechanism uses to pick the end of the tape) may fall out of its retaining slots. It can be put back in place if you are moderately careful. If the plastic case is not broken, the tape is probably readable.
I have dropped a fair number of tapes from desk height over the years (have been using them since the 1970's and designed both hardware and software for tape drives). None has failed as a result. I have also dropped a few H/Ds as well - some were damaged by falls of a few inches (they are actually more robust if operating). I have restored many tapes after 30 years. You will have a hard time finding an ST506 interface that connects to a modern computer.
I also seem to have significant problems with bit-rot on both Windows and Linux. This is noticeable as jpgs which have problems after sitting idle on the disk for a year or two, and occasionally docs and odts that won't read. Less of a problem with SCSI disks AFAICT, so I suspect hardware, but I did use DOS 4.0, so, I am not sure its not software.
I have definitely had brand new server grade HDs fail to start after 3 years on the shelf. I doubt used ones are more reliable.
At the moment, maintaining a robot costs as much as employing a human.
You think a robot looks like R2D2? No. Most look like one thread on you Core2duo.
Actually, many tasks that previously required people now don't even require hardware, and if they do, its more likely to be self-service supermarket checkout hardware or ATM than something from a Transformers movie.
A self-driving tractor does not have a robo-farmer riding on it (Personally, I cant wait to see a Playboy cover with a Sexbot driving a tractor, but then, perhaps I am kinky)
This is the same, or very similar, to an Intel bug described about a month ago:
The issue in both cases is either:
a) The device can be set to operate under conditions that are known to cause it to be unreliable (be out of spec)
b) The device fails to operate reliably when operated within spec
If it is (a) then perhaps the manufacturer should test devices more thoroughly - and then blow fuses to limit operation within spec. If it is (b) the manufacturer should test the devices more thoroughly.
You may know that (eg) Intel sell processors "locked" to prevent over clocking. This prevents (a). It obviously fails to prevent (b) either the manufacturer chose not to lock the device as (or the buyer chose not to buy locked ones) and the suer was "free" to use the devices out of spec, or the article describes devices where the tests were inadequate.
In reality, device performance is not consistent within a batch, and devices are sorted for performance - hence processors with different speed and power options. This has been true since the beginning of TTL.
As devices have higher part count (see Moore's law) they have a higher probability of failure - since there are more failure modes, there is a much higher time-to-test. Time to test maps directly to device cost. Because time-to-test adds to cost, semiconductor devices are not tested 100%*: some parameters are, and others are only sampled to ensure that the tests are identifying the duds.
The problem here is that the parameters tested by sampling may not be as reliably characterised as they are believed to be. If you assume that (for example) all static ram cells in the chip have essentially the same logic levels and speeds within a certain margin, and that margin has a wider spread between devices under circumstances that have not been identified, then testing some sample registers won't tell you that others are not reliable on chips with this unknown and unidentified problem.
Complexity does not scale linearly with transistor count - it is partly that, but it also scales with number of modules, module complexity, and number of interfaces between modules (hardware equivalent of API's not API instances). A more complex CPU has more of all three of these factors.
Any way you look at it, a more complex chip will be more likely to fail in modes that are hard to identify.
About 15 years ago, I was part of a team that identified a problem in a CPU of fairly low complexity caused by data leakage between pipeline stages in a processor used in safety critical applications (AFAIK, no one actually died as a result of these failures). These failure modes are very hard to find. This one took about a man-year of very expensive engineers using very expensive equipment.
I predict that Moore's law will eventually be hit the Thwacks Barrier: Processor complexity will reach the stage where a processor cannot be adequately tested within a timescale that makes it worth producing.
I therefore hereby, formally pronounce that testability will be the barrier that ends Moore's law.
*Some/. users who are old enough to afford lawns may recall the national Semiconductors Mil-spec scandal: devices were sold as 10% tested when in fact they were only sampled, because the failure rate was "very low". No Aircraft carriers or space rockets were actually lost, but crimes were found to be committed anyway.
As someone who was 4,000 miles away at the time, I could smell the stink from here.
The IBM team wanted CPM, but Garry Kidall was flying in his private plane when the IBM people called at his office. He simply did not believe his luck and assumed it was a hoax. The IBM guys went/phoned home "no deal here" Mrs Gates must have been present when the call came in and said "My son has an OS you could use" - not knowing the difference between a Basic interpreter and an OS. why else would the IBM guys have gone and asked a school kid if they could buy his OS? when he did not even have one?
I worked for a company that had a perfectly good OS at the time. Could we sell it? Hell no. Intel had an OS - would they sell it? hell no! Were there other alternatives known to work? probably (see DECUS).
Without any research or due diligence at all, IBM, known for their lawyers, signed a contract that shafted IBM as much as it shafted the rest of the industry.
That is what sinks.
(And we all believe the executives sold their shares "by accident" after the data leaked, and before the news was published? What part of gullible to you not understand?)
100 million Volkswagen diesel shit cars sending mass pollution into the sky and killing people. The US caught them doctoring the test results!
You, sir, are the problem here.
The cars "passed the test". How much of your science syllabus do you remember after the final exam? The politicians only asked them to pass the exam, not to meet the requirements on the road. being good Germans, they "Obeyed the law".
In reality, the problem is not 1.6 litre Volkswagens its the laws. If you want cars not to pollute on the road, pass a law that says "don't pollute on the road". Since trucks have 16 litre engines (not 1.6 litre), they produce a lot more NOx. Also, they run for about 6 hours a day, in place of about 2 (figures made upon the spot).
The real problem is Urea injection (AdBlu). This was introduced to neutralise the NOx emissions in the tailpipe of Diesels. However, due to low power to weight ratio, when you drive a truck in traffic, your foot is typically "on the floor" or "off the pedal". This means the exhaust temperature goes between 900 deg C and 0 deg C and back in seconds. The amount of N2 in the exhaust depends on temperature in the combustion chamber, not the amount of Diesel burned - the N2 comes from the air, not the Diesel. How is the microprocessor going to know how much Urea to inject when it cannot know how much N2 has been dissociated? It can't.
However, the AdBlu process has been favoured by politicians over other solutions to meet the very tight controls on CO2 (There are other solutions, but they do not involve selling large quantities of pig's piss to trucks).
In short, The vehicle manufacturers did as they were told, and the result was predictable. No one wants to admit it for political and financial reasons, and Volkswagen is not American.
Disclaimers: Yes I am an engineer, was a truck driver, live in Europe, am not German, and don't currently own a car of any type.
What exactly would happen?
Most likely, the leak would be small, the air would get in slowly, and the loss of speed would just be annoying.
However, there are other possibilities, critically dependent on the size and number of leaks, the speed of the vehicle at the time, and how close the leaks are to the vehicle.
If the leak happens on one side of the vehicle, while it is passing - eg because the vibration caused by the vehicle passing is "the last straw that breaks the camel's back" then the vehicle will likely be deflected and, several hundred yards down the line, scrape against the side, causing it to be shredded into tiny pieces.
Or, the leak may be some distance ahead of the vehicle, causing it to brake, probably not fiercely, and come to a halt slowly.
or, the leak could be behind the vehicle, propelling it forward to speeds beyond the maximum safe speed, causing the vehicle to come off the track at the firs slight bend, and shred, much like previous scenario.
However, it is far more likely that the real life result will be, much like all previous "atmospheric" railways, and there will be so many small leaks, and two new ones for every one they fix (like Microsoft bugs) that the bloody thing will never actually work at all. I admit that "rats eating the leather valves" may not be a major problem, but "they don't make things like they used to" probably will. I suspect "buying cheap bits from Ali Express" won't help a lot either.
It is my sworn statement that the above is entirely and completely in line with the statements made under oath to the relevant Parliamentary Enquiries and Royal Commissions by Dr Silas P Lardner.
Depends on what you mean by 100% uptime. I have achieved 100% client availability. Some machines go down, but that does not need to take down customer availability - if you can get Carp to work, and use reliable kit.
I have had Sun kit run 3 years non-stop. Generally, this would not happen because it does not need to - you build a new machine and migrate tasks before upgrading the old one. If you have a "whole data centre" to play with, its not difficult. Even if your "whole data centre" is a 21U rack, if you have the right kit, its not hard.
"How did you hear about us?" is one of the most reliable, and direct sources of information about how someone found out about your product
A lot of web sites ask this, generally with a drop-down. I frequently find that either I don't know, or I heard about them in multiple ways, mostly, neither answer is available, so I select the most irrelevant.
Seriously, guys, If you are trying to collect this information, do it properly on a small percentage of transactions. If I get asked 3 times cos I buy from you three times, then I will probably go elsewhere for the fourth. I ran a polling business 25 years ago, and we knew this then. The only two reasons for polling 100% of customers are (a) stupidity, and (b) evil intent.
Slashdot Mirror
However, when doing CAD, I definitely want more pixels, please.
They are so idiotic, only left wing nut jobs can believe them.
Oh, wait ...
I am quietly confident you do not work in the demolition industry.
Hell is something close to debugging PHP with embedded SQL on a low resolution tablet.
And yes, I have been to Africa recently, and yes I could get a P4 with a CRT and PS2 keyboard and mouse for under $50 (it did look past its prime though). I could also buy a Nigerian Guinness for about $0.30 and a nutritiously sound meal for about $1.50. An experienced local would obviously pay less than me for the meal or the computer, unless he had drunk too much Nigerian Guinness.
A lot of people there already had $50 tablets two years ago. Some even had PCs with Linux.
The problem in Africa is not access to hardware, it is, to some degree, understanding the benefits of the hardware (particularly as compared to the merits of dressing up and partying). However, you could access mainframes in 1963 here in the UK. How many people had a use for a mainframe in 1963? Hell, how many people would have known what one did, even if they were in the computer room? (it was enough to make Ross Perot filthy rich). However, the clothes and parties here in the UK in 1963 were pretty crap unless you were a cabinet minister (see Profumo).
The solution to this problem is time not hardware.
Buy Sparc
Are you implying it was an "inhuman problem"?
Are they eating Lithium-Ion batteries, or lead-acid? or do they, in reality, have a mixed diet?
Enquiring minds need to know!
OTOH, you might be doing a favour for a neighbour, and get wrongly accused by the blackmailer - this could lead to a major opportunity for a crime drama - anyone have the number for Bellisarios?
The workers were not slaves. There are payroll records to prove it.
You say it like its a good thing!
Error correction works fine for one, or possibly a small number of errors, such as you might get in DRAM, but if you get a lot of errors like on a bad disk or tape, it is capable of munging the data and declaring it fixed. And there is no way to know how many errors you have got. If you have errors, you get another tape out of the cupboard. (You do have more than one backup, don't you?) SD cards? how would you know what's going on inside?
You do not get an algorithm to "fix" the data if its your life on the line (or your $$$). Of course, if its only prÃn, then maybe the fixing might even improve it.
The big advantage of 200bpi was that you could sprinkle iron filings on the tape and read the bit patterns for disaster recovery. (Not that I would want to read more than a couple of 80 column card images that way).
NO, its not. If you drop it, the puck (the bit that the loading mechanism uses to pick the end of the tape) may fall out of its retaining slots. It can be put back in place if you are moderately careful. If the plastic case is not broken, the tape is probably readable.
I have dropped a fair number of tapes from desk height over the years (have been using them since the 1970's and designed both hardware and software for tape drives). None has failed as a result. I have also dropped a few H/Ds as well - some were damaged by falls of a few inches (they are actually more robust if operating). I have restored many tapes after 30 years. You will have a hard time finding an ST506 interface that connects to a modern computer.
I also seem to have significant problems with bit-rot on both Windows and Linux. This is noticeable as jpgs which have problems after sitting idle on the disk for a year or two, and occasionally docs and odts that won't read. Less of a problem with SCSI disks AFAICT, so I suspect hardware, but I did use DOS 4.0, so, I am not sure its not software.
I have definitely had brand new server grade HDs fail to start after 3 years on the shelf. I doubt used ones are more reliable.
You think a robot looks like R2D2? No. Most look like one thread on you Core2duo.
Actually, many tasks that previously required people now don't even require hardware, and if they do, its more likely to be self-service supermarket checkout hardware or ATM than something from a Transformers movie.
A self-driving tractor does not have a robo-farmer riding on it (Personally, I cant wait to see a Playboy cover with a Sexbot driving a tractor, but then, perhaps I am kinky)
The issue in both cases is either:
a) The device can be set to operate under conditions that are known to cause it to be unreliable (be out of spec)
b) The device fails to operate reliably when operated within spec
If it is (a) then perhaps the manufacturer should test devices more thoroughly - and then blow fuses to limit operation within spec. If it is (b) the manufacturer should test the devices more thoroughly.
You may know that (eg) Intel sell processors "locked" to prevent over clocking. This prevents (a). It obviously fails to prevent (b) either the manufacturer chose not to lock the device as (or the buyer chose not to buy locked ones) and the suer was "free" to use the devices out of spec, or the article describes devices where the tests were inadequate.
In reality, device performance is not consistent within a batch, and devices are sorted for performance - hence processors with different speed and power options. This has been true since the beginning of TTL. As devices have higher part count (see Moore's law) they have a higher probability of failure - since there are more failure modes, there is a much higher time-to-test. Time to test maps directly to device cost. Because time-to-test adds to cost, semiconductor devices are not tested 100%*: some parameters are, and others are only sampled to ensure that the tests are identifying the duds. The problem here is that the parameters tested by sampling may not be as reliably characterised as they are believed to be. If you assume that (for example) all static ram cells in the chip have essentially the same logic levels and speeds within a certain margin, and that margin has a wider spread between devices under circumstances that have not been identified, then testing some sample registers won't tell you that others are not reliable on chips with this unknown and unidentified problem.
Complexity does not scale linearly with transistor count - it is partly that, but it also scales with number of modules, module complexity, and number of interfaces between modules (hardware equivalent of API's not API instances). A more complex CPU has more of all three of these factors. Any way you look at it, a more complex chip will be more likely to fail in modes that are hard to identify.
About 15 years ago, I was part of a team that identified a problem in a CPU of fairly low complexity caused by data leakage between pipeline stages in a processor used in safety critical applications (AFAIK, no one actually died as a result of these failures). These failure modes are very hard to find. This one took about a man-year of very expensive engineers using very expensive equipment.
I predict that Moore's law will eventually be hit the Thwacks Barrier: Processor complexity will reach the stage where a processor cannot be adequately tested within a timescale that makes it worth producing.
I therefore hereby, formally pronounce that testability will be the barrier that ends Moore's law.
*Some /. users who are old enough to afford lawns may recall the national Semiconductors Mil-spec scandal: devices were sold as 10% tested when in fact they were only sampled, because the failure rate was "very low". No Aircraft carriers or space rockets were actually lost, but crimes were found to be committed anyway.
The IBM team wanted CPM, but Garry Kidall was flying in his private plane when the IBM people called at his office. He simply did not believe his luck and assumed it was a hoax. The IBM guys went/phoned home "no deal here" Mrs Gates must have been present when the call came in and said "My son has an OS you could use" - not knowing the difference between a Basic interpreter and an OS. why else would the IBM guys have gone and asked a school kid if they could buy his OS? when he did not even have one?
I worked for a company that had a perfectly good OS at the time. Could we sell it? Hell no. Intel had an OS - would they sell it? hell no! Were there other alternatives known to work? probably (see DECUS).
Without any research or due diligence at all, IBM, known for their lawyers, signed a contract that shafted IBM as much as it shafted the rest of the industry.
That is what sinks.
(And we all believe the executives sold their shares "by accident" after the data leaked, and before the news was published? What part of gullible to you not understand?)
You, sir, are the problem here.
The cars "passed the test". How much of your science syllabus do you remember after the final exam? The politicians only asked them to pass the exam, not to meet the requirements on the road. being good Germans, they "Obeyed the law".
In reality, the problem is not 1.6 litre Volkswagens its the laws. If you want cars not to pollute on the road, pass a law that says "don't pollute on the road". Since trucks have 16 litre engines (not 1.6 litre), they produce a lot more NOx. Also, they run for about 6 hours a day, in place of about 2 (figures made upon the spot).
The real problem is Urea injection (AdBlu). This was introduced to neutralise the NOx emissions in the tailpipe of Diesels. However, due to low power to weight ratio, when you drive a truck in traffic, your foot is typically "on the floor" or "off the pedal". This means the exhaust temperature goes between 900 deg C and 0 deg C and back in seconds. The amount of N2 in the exhaust depends on temperature in the combustion chamber, not the amount of Diesel burned - the N2 comes from the air, not the Diesel. How is the microprocessor going to know how much Urea to inject when it cannot know how much N2 has been dissociated? It can't.
However, the AdBlu process has been favoured by politicians over other solutions to meet the very tight controls on CO2 (There are other solutions, but they do not involve selling large quantities of pig's piss to trucks).
In short, The vehicle manufacturers did as they were told, and the result was predictable. No one wants to admit it for political and financial reasons, and Volkswagen is not American.
Disclaimers: Yes I am an engineer, was a truck driver, live in Europe, am not German, and don't currently own a car of any type.
FTFY
You have no idea how kinky some people are these days!
I, for one, am bitterly disappointed by this bad news.
Are you an Internet marketer, by any chance?
There is a ton of data on that available from Crossrail in London. (The money itself is generally measured in metric megatons).
What if several small children are decapitated by the paper cut?
(At least as likely as most of the other scenarios put forward here).
However, there are other possibilities, critically dependent on the size and number of leaks, the speed of the vehicle at the time, and how close the leaks are to the vehicle.
If the leak happens on one side of the vehicle, while it is passing - eg because the vibration caused by the vehicle passing is "the last straw that breaks the camel's back" then the vehicle will likely be deflected and, several hundred yards down the line, scrape against the side, causing it to be shredded into tiny pieces.
Or, the leak may be some distance ahead of the vehicle, causing it to brake, probably not fiercely, and come to a halt slowly.
or, the leak could be behind the vehicle, propelling it forward to speeds beyond the maximum safe speed, causing the vehicle to come off the track at the firs slight bend, and shred, much like previous scenario.
However, it is far more likely that the real life result will be, much like all previous "atmospheric" railways, and there will be so many small leaks, and two new ones for every one they fix (like Microsoft bugs) that the bloody thing will never actually work at all. I admit that "rats eating the leather valves" may not be a major problem, but "they don't make things like they used to" probably will. I suspect "buying cheap bits from Ali Express" won't help a lot either.
It is my sworn statement that the above is entirely and completely in line with the statements made under oath to the relevant Parliamentary Enquiries and Royal Commissions by Dr Silas P Lardner.
I have had Sun kit run 3 years non-stop. Generally, this would not happen because it does not need to - you build a new machine and migrate tasks before upgrading the old one. If you have a "whole data centre" to play with, its not difficult. Even if your "whole data centre" is a 21U rack, if you have the right kit, its not hard.
The essential thing is to avoid MS.
A lot of web sites ask this, generally with a drop-down. I frequently find that either I don't know, or I heard about them in multiple ways, mostly, neither answer is available, so I select the most irrelevant.
Seriously, guys, If you are trying to collect this information, do it properly on a small percentage of transactions. If I get asked 3 times cos I buy from you three times, then I will probably go elsewhere for the fourth. I ran a polling business 25 years ago, and we knew this then. The only two reasons for polling 100% of customers are (a) stupidity, and (b) evil intent.