Resolution played a huge part in performance considerations. But there was also the problem with card support.
Remember UniVBE? You needed that to get high res display, but it was very, VERY slow. I managed to get 1024x768 with 256 colors on my 386, and about all it was good for was viewing photos because they took about a second to draw. I managed to get Win 3.1 run at that resolution and could watch the windows slowly appear on the screen.
One of the first games I remember playing in high res graphics was a chess game precisely because lack of performance wasn't such a big deal for that kind of thing.
Exactly what is that cost? Is Mother Nature going to send me a bill at the end of the month?
There is a bill, but it won't come for a while. When it does though, it'll be with a lot of interest.
For instance, if enough ice melts, we could end up with a New Orleans on many coastal cities. Significant climate change will result in consequences for agriculture.
Like somebody else said on an earlier article, I can piss in the well and the water will remain drinkable. If everybody does it though, it won't remain like that forever. And everybody's contribution to that will be tiny so nobody will feel like it's their fault.
Not only that, but is the government going to use those tax revenues to somehow pay that cost so that there's no net impact of me polluting?
Well, the tax makes it harder for you to pollute as much, because it costs more. That's a start.
Though I agree that the money gained should be used for something like research of more environmentally friendly cars.
You care that he can earn money for something 10 years old? WTF?
He does because most people do work, and get paid for it once. They don't get to sit on their butt and receive royalties for something they did 10 years ago. And I think that's a good thing, because it motivates people to do other things.
What if people build things that take 10 years to come out ahead?
Then they get screwed. Sucks for them.
What do you care that it's 10 years?
There has to be a reasonable limit, and 10 years is rather too long for computer related things. I'm sure there have been enough new developments in compression that his book misses a significant amount of relevant research that's been done since then. Think of what you were doing with your computer 10 years ago.
But here's another thing. Your argument seems to be that if things may take 10 years to develop, we should allow copyright to be long enough for them to be able to benefit. But why 10 years? For instance, from wikipedia:
The history of the steam engine stretches back as far as the first century AD; the first recorded rudimentary steam engine being the aeolipile described by Hero of Alexandria.[3] In the following centuries, the few engines known about were essentially experimental devices used by inventors to demonstrate the properties of steam, such as the rudimentary steam turbine device described by Taqi al-Din[4] in 1551 and Giovanni Branca[5] in 1629.
The first practical steam-powered 'engine' was a water pump, developed in 1698 by Thomas Savery. It proved only to have a limited lift height and was prone to boiler explosions, but it still received some use for mines and pumping stations.
Do you believe we should find Hero of Alexandria's ancestors and pay them money for the steam engine? If 1600 years is too long, then why 10 years isn't?
See, thing is most people aren't going to go out of their way to find a way to pay for your book.
If you were my favourite author, maybe I'd make a special effort to reward you for your work. But I have no clue who you are, so when given a choice between "PDF" and "out of print", I will go with the PDF obviously. Then if I really, REALLY like the PDF maybe I will try to figure out where to buy a copy. But if your website still says "out of stock" by then I'll probably just give up on it, unless it's something wonderful.
You have to understand that most people find it very unusual to have to figure out how to give you money. If you want money, you're expected to make it giving you some very easy, by for instance, actually having a book to sell. If your volume is so low that you can't get it printed, at least put up a paypal donation link.
Actually, VACUUM has been automatic for a few years now. The dedicated daemon has gone away too and been merged into the server process.
And sometimes it may make sense to manually scheduling VACUUM. For instance if you never delete from a table there's no need to bother trying to vacuum it. If you have a very high load, it may make sense to defer it until a low load time. It also makes sense to manually run it after making big changes to the database (like restoring a backup for instance).
But in current releases it works fine on its own and there's no need to mess with it unless you have a special need for it.
Surely the whole point about a list of infinite size is that, by definition, it contains every possible permutation, and therefore nothing can be excluded.
No, you can have a list of all the prime numbers, which is infinite, but doesn't have any numbers divisible by 2.
You can also have an infinite list of every number except 3.
Lithium isn't the only thing that batteries can be made from.
Electric cars need electricity, which can be stored in many forms. If there's not enough lithium perhaps we'll use NiMH batteries, or flywheels, or ultracapacitors, or superconductors, or...
And unlike with oil, lithium running out shouldn't be a huge problem. Existing car batteries won't stop working. All that will be needed is to figure out a new system for new cars, and a compatible way to replace worn out batteries.
Actually watch your bandwidth usage. You'll be surprised how easy it is to use 1GB.
If you visit image heavy websites (dark roasted blend for instance), youtube, play games, etc, it's very easy to use up 500MB in one day without really trying.
Then add completely normal downloads to that. 3 computers downloading the same 200MB of updates. A few game demos amounting to a few GB. I've even heard of a keyboard driver with a 100MB.
Since I booted this morning, my laptop and desktop together downloaded about 50MB, while doing nothing special: some web browsing, some slashdot reading, catching up with a couple web comics, and checking the mail. Extrapolating that usage, that would amount to 74GB over a year.
No, he spoke about transistors. Transistors is what a CPU is made of.
You can think of it as a huge rail network. The system relies on the timing to be right. If the rails break down, or a train slows down too much, you can end up with a huge crash.
Now what he also said is that failures aren't necessarily huge and dramatic. You may have an absolutely vital part break, then the computer one day stops working. But there exist subtle failure modes. For instance bad capacitors can cause random crashes with no easily determinable origin. Bad RAM may cause all sorts of issues, like none at all (if you're very lucky), a bad pixel on the screen, or disk corruption.
I've had hardware die in different ways, and it usually doesn't suddenly die one day. I've had a laptop screen fail over a period of weeks, getting worse and worse over time. Hard disks may keep working with some bad sectors. Components with overheating or soldering problems often display intermittent problems before really dying. Worn out fans keep on spinning for some time before failing. LCD backlights and OLEDs gradually decrease in brightness, and so on.
How long it takes for something to break in this manner depends on the component, usage and environment. But no hardware lasts forever, and doesn't necessarily just die one day, having worked perfectly the day before.
Individual components can degrade in terms of becoming slower to change state, less able to hold a charge, increasing in resistance, decreasing in voltage tolerance, and so on. But those are individual component characteristics.
The things made from those, like CPUs and motherboards work on the assumption of that its component parts have characteristics within some tolerance value. When some component goes out of the acceptable range, all sorts of things can happen. Critical components failing can result in the CPU getting fried, getting constant crashes or the computer not turning on. You might also get lucky and had some unused part fail to no ill effect.
So no, you won't find evidence that the computer will become 25% slower. What possibly gets slower is something deep inside the CPU. Eventually it doesn't change state fast enough, and the CPU starts working incorrectly.
Humans have bad speed, but *very* good endurance. A determined human in good shape can simply jog at a comfortable pace after many animals until they're so exhausted they can't make a step. Not to mention that a human can carry high calorie foods and water to gain even more of an advantage.
If you look around, Linux managed to get into plenty places already.
For instance, large chains of supermarkets run with Linux on the points of sale, the server that runs in the shop, and the central servers.
If you want to work with Linux, the best bet is to look for a place that runs lots of servers. It's not doing that great in desktop usage yet, but it's very commonly used on servers, embedded hardware and other applications where normal people don't even know what's running.
"Works" means it successfully performs the task assigned to it. In businesses there are many single purpose machines. If a print server prints, it works. If a file server serves files, it works. It may have a too old version of DirectX, but in that environment, nobody cares about that.
Furthermore, in those environments, that it works is a requirement. If it didn't work they wouldn't be waiting for MS to release a new version, they'd be using something else instead.
The public's lack of interest in NASA has nothing to do with pop culture, it's that NASA isn't doing anything interesting anymore.
In the days of the space race, it was very exciting. The "us vs them" competition with the USSR, people doing things never done before, going to places nobody ever went before. And all that done by launching people with experimental hardware done on monumental scales.
Where is all that now? NASA keeps shuttling people between ground and space station, where they do amazing experiments, such as how spiders react to a lack of gravity. The hardware is still big and dangerous, but now if something goes wrong it's not because it's new unproven tech, but because the management at NASA is stupid. The Mars robots are about the most interesting thing being done these days, and even that isn't all that exiciting.
What's needed is not nonsense like Britney getting involved in the space program, but a space program that does something the public would be interested in. A colony on the moon, and astronauts on Mars would be a start. Sure it's dangerous and expensive, but it's also very much exciting, and the danger didn't stop anybody in the race of getting to the moon.
If what needs to be done is currently being successfully done, then an upgrade isn't needed, as simple as that.
There are plenty businesses out there which run servers on kernel 2.4, and desktops on Windows 2000, with frontend apps written in VB6.
The server keeps on chugging, the desktop keeps working by virtue of simple reimaging when it breaks, and the apps continue doing whatever simple task they were made for 10 years back.
When businesses start evaluating upgrading isn't when a new version of something comes out, but when they run into difficulties. Like server hardware that works with kernel 2.4 is hard to find, support for Win2K is inexistent, frontend app written in VB6 breaks on Vista for some obscure reason.
At once company I worked at they had their product made to be installed on Red Hat 9, because that's what the client used, and certainly none of them hurried in the slightest with even evaluating any sort of upgrades. Why do it, when things keep working? It seems it was only last year they started thinking that maybe supporting something that runs a newer kernel would be a good thing. Even that was a "should do some day" sort of thing, without being an official project or having a deadline.
Nonsense. If that was true, then why don't we have a state-sanctioned monopoly on all foodstuffs so we don't run the risk of 'unreliable' supply? I mean, food is so crucial.
Because it's not needed? There's no problem with 50 producers competing for who can deliver the cheapest rice, because there's no problem with all of them making their products available for sale, and it still must pass government quality standards.
That doesn't work with things like water though. Would you want to have 10 sets of water pipes, with all the street digging that implies, and 10x more frequent pipe breakage? The space available for piping is very limited as well.
In this situation the way to go is not having 10 sets of pipes, but have one, highly regulated delivery network (water, power, fiber), and competition in the supply of that network (powerstations, water filtering plants, ISPs).
Done correctly, the delivery network lacks any reason to prefer or favor one provider over another, and the providers lack the ability to deny access to each other, since they don't own the delivery network. The consumer can then freely choice which they want, and the entry barrier for a new provider is low because it doesn't require digging up streets.
If that still doesn't work, you can try upgrading from the commandline, with:
$ sudo do-release-upgrade -d
Since this is on the commandline, you can press Ctrl+Z at any time to temporarily interrupt the upgrade process. Then use the "fg" command to resume it again.
Still, overheating isn't something that should normally happen. Maybe you have a clogged or defective fan?
ondemand actually happens to be the best governor.
In theory, "powersave", by keeping the CPU frequency at a minimum would save some power in comparison. In practice, it doesn't. This is because doing anything at all prevents the CPU from entering the lowest power using modes (which go beyond simply dropping in frequency).
So it's more efficient to make the CPU run at full blast, do whatever needs to be done, then go to sleep (C3, not suspend to RAM), than to do the same work at a lower clock speed, keeping the CPU active 3 or 4 times longer. By C2 the clock isn't active anymore, which is a huge gain on anything the "powersave" governor can provide.
There's nothing wrong with fishing for more funding.
The important thing isn't the efficiency, but the price/performance ratio.
1% efficient cells that are dirt cheap still aren't worth installing on your roof. 95% efficient cells at $50K per square meter are only of interest for satellite applications.
But, a 30% efficient cell that's reasonably cheap is a whole lot more interesting than a 40% one that costs 5 times as much. Taking a cheap 10% efficient tech and making it 3 times better without making it 3 times more expensive is a very useful thing.
There was an article some years ago about monkeys (not sure if chimpanzees or not) being trained to use money.
Researchers taught them that discs of metal could be exchanged for food and such things. They got all sorts of interesting behaviors out of it, including the monkeys attempting to fake the money.
One uncomfortable discovery was discovering that some of them were actually using that money to pay for sex.
This seems even better than this one. Food for sex is a straightforward exchange. Tokens that can be used to obtain food for sex is more complicated, and shows a deeper understanding.
That's a fine beginning, and I don't think there are many people who disagree with it.
The problem is when you take it too far. Then you get much stranger laws, such as "protecting the institution of marriage", making it illegal to have homosexual sex, prohibition, criminalization of victimless crimes, and calls for anybody acting "suspiciously" to be reported.
Those things don't really favor society. It doesn't benefit society as a whole when laws tell you to look at an amateur chemist as a new unabomber in the making. It does however benefit some paranoid politician who will get credit for passing laws that are completely counterproductive to society.
And that's the real problem. Too far down that road it's not longer really for society's benefit. It's all full things done for the carreer of politicians under the guise of benefit for the society. And unfortunately many of the society's members can't look far ahead enough to see what they're getting into.
Linux BIOS isn't for people who just think their box boots too slowly.
It's for people with large clusters, or important servers, where having an important server spend several minutes in the BIOS is a considerable annoyance. It also adds management facilities not present on many motherboards.
Picture having to go around with a keyboard and monitor, server by server, to change a BIOS setting on hundreds of servers.
Slashdot Mirror
Resolution played a huge part in performance considerations. But there was also the problem with card support.
Remember UniVBE? You needed that to get high res display, but it was very, VERY slow. I managed to get 1024x768 with 256 colors on my 386, and about all it was good for was viewing photos because they took about a second to draw. I managed to get Win 3.1 run at that resolution and could watch the windows slowly appear on the screen.
One of the first games I remember playing in high res graphics was a chess game precisely because lack of performance wasn't such a big deal for that kind of thing.
There is a bill, but it won't come for a while. When it does though, it'll be with a lot of interest.
For instance, if enough ice melts, we could end up with a New Orleans on many coastal cities. Significant climate change will result in consequences for agriculture.
Like somebody else said on an earlier article, I can piss in the well and the water will remain drinkable. If everybody does it though, it won't remain like that forever. And everybody's contribution to that will be tiny so nobody will feel like it's their fault.
Well, the tax makes it harder for you to pollute as much, because it costs more. That's a start.
Though I agree that the money gained should be used for something like research of more environmentally friendly cars.
He does because most people do work, and get paid for it once. They don't get to sit on their butt and receive royalties for something they did 10 years ago. And I think that's a good thing, because it motivates people to do other things.
Then they get screwed. Sucks for them.
There has to be a reasonable limit, and 10 years is rather too long for computer related things. I'm sure there have been enough new developments in compression that his book misses a significant amount of relevant research that's been done since then. Think of what you were doing with your computer 10 years ago.
But here's another thing. Your argument seems to be that if things may take 10 years to develop, we should allow copyright to be long enough for them to be able to benefit. But why 10 years? For instance, from wikipedia:
Do you believe we should find Hero of Alexandria's ancestors and pay them money for the steam engine? If 1600 years is too long, then why 10 years isn't?
Oops. You're right there.
See, thing is most people aren't going to go out of their way to find a way to pay for your book.
If you were my favourite author, maybe I'd make a special effort to reward you for your work. But I have no clue who you are, so when given a choice between "PDF" and "out of print", I will go with the PDF obviously. Then if I really, REALLY like the PDF maybe I will try to figure out where to buy a copy. But if your website still says "out of stock" by then I'll probably just give up on it, unless it's something wonderful.
You have to understand that most people find it very unusual to have to figure out how to give you money. If you want money, you're expected to make it giving you some very easy, by for instance, actually having a book to sell. If your volume is so low that you can't get it printed, at least put up a paypal donation link.
Actually, VACUUM has been automatic for a few years now. The dedicated daemon has gone away too and been merged into the server process.
And sometimes it may make sense to manually scheduling VACUUM. For instance if you never delete from a table there's no need to bother trying to vacuum it. If you have a very high load, it may make sense to defer it until a low load time. It also makes sense to manually run it after making big changes to the database (like restoring a backup for instance).
But in current releases it works fine on its own and there's no need to mess with it unless you have a special need for it.
No, you can have a list of all the prime numbers, which is infinite, but doesn't have any numbers divisible by 2.
You can also have an infinite list of every number except 3.
Lithium isn't the only thing that batteries can be made from.
Electric cars need electricity, which can be stored in many forms. If there's not enough lithium perhaps we'll use NiMH batteries, or flywheels, or ultracapacitors, or superconductors, or...
And unlike with oil, lithium running out shouldn't be a huge problem. Existing car batteries won't stop working. All that will be needed is to figure out a new system for new cars, and a compatible way to replace worn out batteries.
Actually watch your bandwidth usage. You'll be surprised how easy it is to use 1GB.
If you visit image heavy websites (dark roasted blend for instance), youtube, play games, etc, it's very easy to use up 500MB in one day without really trying.
Then add completely normal downloads to that. 3 computers downloading the same 200MB of updates. A few game demos amounting to a few GB. I've even heard of a keyboard driver with a 100MB.
Since I booted this morning, my laptop and desktop together downloaded about 50MB, while doing nothing special: some web browsing, some slashdot reading, catching up with a couple web comics, and checking the mail. Extrapolating that usage, that would amount to 74GB over a year.
Nice try, but programmers still count the number of things the same way as everybody else.
If "contients" is a collection, then continents.Count will be 7, while continents.Min is 0, and continents.Max is 6.
No, he spoke about transistors. Transistors is what a CPU is made of.
You can think of it as a huge rail network. The system relies on the timing to be right. If the rails break down, or a train slows down too much, you can end up with a huge crash.
Now what he also said is that failures aren't necessarily huge and dramatic. You may have an absolutely vital part break, then the computer one day stops working. But there exist subtle failure modes. For instance bad capacitors can cause random crashes with no easily determinable origin. Bad RAM may cause all sorts of issues, like none at all (if you're very lucky), a bad pixel on the screen, or disk corruption.
I've had hardware die in different ways, and it usually doesn't suddenly die one day. I've had a laptop screen fail over a period of weeks, getting worse and worse over time. Hard disks may keep working with some bad sectors. Components with overheating or soldering problems often display intermittent problems before really dying. Worn out fans keep on spinning for some time before failing. LCD backlights and OLEDs gradually decrease in brightness, and so on.
How long it takes for something to break in this manner depends on the component, usage and environment. But no hardware lasts forever, and doesn't necessarily just die one day, having worked perfectly the day before.
The chip won't slow down, it'll simply fail.
Individual components can degrade in terms of becoming slower to change state, less able to hold a charge, increasing in resistance, decreasing in voltage tolerance, and so on. But those are individual component characteristics.
The things made from those, like CPUs and motherboards work on the assumption of that its component parts have characteristics within some tolerance value. When some component goes out of the acceptable range, all sorts of things can happen. Critical components failing can result in the CPU getting fried, getting constant crashes or the computer not turning on. You might also get lucky and had some unused part fail to no ill effect.
So no, you won't find evidence that the computer will become 25% slower. What possibly gets slower is something deep inside the CPU. Eventually it doesn't change state fast enough, and the CPU starts working incorrectly.
Actually it's perfectly workable.
Humans have bad speed, but *very* good endurance. A determined human in good shape can simply jog at a comfortable pace after many animals until they're so exhausted they can't make a step. Not to mention that a human can carry high calorie foods and water to gain even more of an advantage.
If you look around, Linux managed to get into plenty places already.
For instance, large chains of supermarkets run with Linux on the points of sale, the server that runs in the shop, and the central servers.
If you want to work with Linux, the best bet is to look for a place that runs lots of servers. It's not doing that great in desktop usage yet, but it's very commonly used on servers, embedded hardware and other applications where normal people don't even know what's running.
"Works" means it successfully performs the task assigned to it. In businesses there are many single purpose machines. If a print server prints, it works. If a file server serves files, it works. It may have a too old version of DirectX, but in that environment, nobody cares about that.
Furthermore, in those environments, that it works is a requirement. If it didn't work they wouldn't be waiting for MS to release a new version, they'd be using something else instead.
I said it before, and I'll say it again:
The public's lack of interest in NASA has nothing to do with pop culture, it's that NASA isn't doing anything interesting anymore.
In the days of the space race, it was very exciting. The "us vs them" competition with the USSR, people doing things never done before, going to places nobody ever went before. And all that done by launching people with experimental hardware done on monumental scales.
Where is all that now? NASA keeps shuttling people between ground and space station, where they do amazing experiments, such as how spiders react to a lack of gravity. The hardware is still big and dangerous, but now if something goes wrong it's not because it's new unproven tech, but because the management at NASA is stupid. The Mars robots are about the most interesting thing being done these days, and even that isn't all that exiciting.
What's needed is not nonsense like Britney getting involved in the space program, but a space program that does something the public would be interested in. A colony on the moon, and astronauts on Mars would be a start. Sure it's dangerous and expensive, but it's also very much exciting, and the danger didn't stop anybody in the race of getting to the moon.
If what needs to be done is currently being successfully done, then an upgrade isn't needed, as simple as that.
There are plenty businesses out there which run servers on kernel 2.4, and desktops on Windows 2000, with frontend apps written in VB6.
The server keeps on chugging, the desktop keeps working by virtue of simple reimaging when it breaks, and the apps continue doing whatever simple task they were made for 10 years back.
When businesses start evaluating upgrading isn't when a new version of something comes out, but when they run into difficulties. Like server hardware that works with kernel 2.4 is hard to find, support for Win2K is inexistent, frontend app written in VB6 breaks on Vista for some obscure reason.
At once company I worked at they had their product made to be installed on Red Hat 9, because that's what the client used, and certainly none of them hurried in the slightest with even evaluating any sort of upgrades. Why do it, when things keep working? It seems it was only last year they started thinking that maybe supporting something that runs a newer kernel would be a good thing. Even that was a "should do some day" sort of thing, without being an official project or having a deadline.
Because it's not needed? There's no problem with 50 producers competing for who can deliver the cheapest rice, because there's no problem with all of them making their products available for sale, and it still must pass government quality standards.
That doesn't work with things like water though. Would you want to have 10 sets of water pipes, with all the street digging that implies, and 10x more frequent pipe breakage? The space available for piping is very limited as well.
In this situation the way to go is not having 10 sets of pipes, but have one, highly regulated delivery network (water, power, fiber), and competition in the supply of that network (powerstations, water filtering plants, ISPs).
Done correctly, the delivery network lacks any reason to prefer or favor one provider over another, and the providers lack the ability to deny access to each other, since they don't own the delivery network. The consumer can then freely choice which they want, and the entry barrier for a new provider is low because it doesn't require digging up streets.
Ubuntu packages are signed. The signature certifies that the package was mirrored as-is and not modified in any way.
Ok, I can suggest some workarounds then.
First, you can manually force the governor with this as root:
If that still doesn't work, you can try upgrading from the commandline, with:
Since this is on the commandline, you can press Ctrl+Z at any time to temporarily interrupt the upgrade process. Then use the "fg" command to resume it again.
Still, overheating isn't something that should normally happen. Maybe you have a clogged or defective fan?
ondemand actually happens to be the best governor.
In theory, "powersave", by keeping the CPU frequency at a minimum would save some power in comparison. In practice, it doesn't. This is because doing anything at all prevents the CPU from entering the lowest power using modes (which go beyond simply dropping in frequency).
So it's more efficient to make the CPU run at full blast, do whatever needs to be done, then go to sleep (C3, not suspend to RAM), than to do the same work at a lower clock speed, keeping the CPU active 3 or 4 times longer. By C2 the clock isn't active anymore, which is a huge gain on anything the "powersave" governor can provide.
There's nothing wrong with fishing for more funding.
The important thing isn't the efficiency, but the price/performance ratio.
1% efficient cells that are dirt cheap still aren't worth installing on your roof.
95% efficient cells at $50K per square meter are only of interest for satellite applications.
But, a 30% efficient cell that's reasonably cheap is a whole lot more interesting than a 40% one that costs 5 times as much. Taking a cheap 10% efficient tech and making it 3 times better without making it 3 times more expensive is a very useful thing.
There was an article some years ago about monkeys (not sure if chimpanzees or not) being trained to use money.
Researchers taught them that discs of metal could be exchanged for food and such things. They got all sorts of interesting behaviors out of it, including the monkeys attempting to fake the money.
One uncomfortable discovery was discovering that some of them were actually using that money to pay for sex.
This seems even better than this one. Food for sex is a straightforward exchange. Tokens that can be used to obtain food for sex is more complicated, and shows a deeper understanding.
That's a fine beginning, and I don't think there are many people who disagree with it.
The problem is when you take it too far. Then you get much stranger laws, such as "protecting the institution of marriage", making it illegal to have homosexual sex, prohibition, criminalization of victimless crimes, and calls for anybody acting "suspiciously" to be reported.
Those things don't really favor society. It doesn't benefit society as a whole when laws tell you to look at an amateur chemist as a new unabomber in the making. It does however benefit some paranoid politician who will get credit for passing laws that are completely counterproductive to society.
And that's the real problem. Too far down that road it's not longer really for society's benefit. It's all full things done for the carreer of politicians under the guise of benefit for the society. And unfortunately many of the society's members can't look far ahead enough to see what they're getting into.
Linux BIOS isn't for people who just think their box boots too slowly.
It's for people with large clusters, or important servers, where having an important server spend several minutes in the BIOS is a considerable annoyance. It also adds management facilities not present on many motherboards.
Picture having to go around with a keyboard and monitor, server by server, to change a BIOS setting on hundreds of servers.