No, it lasted and dominated because IBM
happened to choose it to be the cpu for their
PC.
Initially, yes. However, it lasted this long
because Intel worked very hard to keep it
alive. If the x86 trailed, for example, the
PowerPC-based Macintosh by 50% in performance,
many things may be very different.
Had that not happened, x86 would be at best a footnote, along with the 65XX, Z80, etc.
The 6502 and Z-80 are not "footnotes". They
deserve prominent spots in CPU history marking
the beginning of personal computing and
affordable gaming consoles. When the
x86's time finally comes, it will also be a
major milestone marking the maturing of
personal computing.
It's lasted, and dominated, because in many ways it's a good design.
Depends on what you mean by good. If you mean
the Darwinian sense, then yes, it's phenomenally
successful.
However, you write like a person who has never
had to work under the 8086 real mode in assembly
language. Here are a few things wrong with it
(the whole family, over the years):
Too few registers
Registers have special purposes, and are
not generic enough
Many instructions are very rarely used
Did not have a supervisor mode (pre 386)
or MMU support
Unbelievably lame 16-bit segmentation
Overcomplicated memory protection (few
if any OSes take advantage of segmentation)
These are design failings that are not "in the
eye of the beholder". Intel overcame the first
two by going to a hidden RISCy core with many
more registers, the third by implementing
many rarely used instructions in microcode,
the next two by essentially discarding the
8086 and 80286 architectures in going to the
80386. Intel deserves a lot of credit, but
they had to work very hard to overcome these
problems.
Comparing it to the 68000 is left as an
exercise for the reader.
I'd love to find an offbeat processor like this on a board which still accepted standard PCI cards, or at least a few USB peripherals.
Here. They're a little pricey, but that's what you pay for getting
out of the mainstream. They can run BSD or Linux, and a development kit (including a really good IDE) is available for free. If you can't afford a new one, there's a vibrant used market as well.
I'm not joking. A Mac is the best way to experiment outside the x86, partly because it can still double as a real personal computer and run popular apps.
the institution should _not_ have any more say than any ISP over how student use what is in effect their _own home connection_.
In fact, an ISP can tell you that you're
paying for (and entitled to) X GB per month.
Many of them use an approximation of "unlimited"
access, but I've known people warned for
"abusing" supposedly "unlimited" dial-up.
Now, the question is whether Cornell is in
effect changing the terms of service illegally?
They have no choice of ISP's.
That's no different than where I live. With
no access to DSL (and apartments typically
don't like satellite dishes), a cable monopoly
is my only reasonable choice of broadband
provider. Why is that special?
as long as my only concern is fiscal, and my lawyers are smart, there's no reason to care about trust.
So you would hire a 14 year old high school
drop-out orphan to design a web site for you
Fortune 10 company, as long as there's a proper
contract? I expect that he will be one of the
lowest bidders.
it's not economic to sell something as cheap
as a single match, or even ten matches.
My point was that a music vendor would rather
sell you a $15 CD, and make a $10 profit (for
example), rather than try to sell you 15
different songs at $1, even if each song
still earns the same profit margin of 66 cents.
Oversimplified, they make $10 by putting 3
good songs in an album right now. Selling
individually, those same songs get them less
than $2, so they must make more good stuff to
get the same revenue.
So, no, I'm not certain they're happy about
this.
there are at least a few songs in a typical album that I will enjoy that I didn't hear before buying it.
Excellent point. Cheap downloadable singles
will also place pressure on the producers to
pack the album with better stuff.
And those countries are stupid. Why the hell do you ban cloning?
I wasn't commenting on the merits of that decision,
just showing that humans are not bound by some
endless need to reproduce. In the case of cloning, many made an ethical decision to not pursue one form of reproduction.
the commpanies make just as much money so they're happy
Gosh, no. If you're allowed to purchase low-value goods piecemeal, you might buy it from someone else next time. That's why you have to buy 500 matches at a time, even if you just need one.
They should talk to [...] VMware for the emulation for x86.
The technique used by VMWare is known as
virtualization, which still relies on having
an x86 chip to actually process instructions.
For VMWare to support the PowerPC, they have
to essentially create a new product. MOL
and Plex86 also use the same technique.
Of course, with 1.3 GHz Durons costing just
$30 in retail, I wonder why nobody has come
up with a x86-on-a-PCI with (a lot less)
associated software to solve this problem.
China is persuing this project for political reasons too. It is a Jim Collins style Big Hairy Audacious Goal, the whole point is that it is hard. But China wants to do it to prove it is a major power and that the US and others should not underestimate them.
China is also at risk of political meltdown.
Rapid introduction of capitalism over the past
two decades dramatically increased the gap
between the rich and the poor. The violation
of long espoused communist ideals is pretty
clear, and not easy to overlook if you're the
ones left behind. Serious tension is
inevitable over the coming years.
BHAGs unite the country and give people a
sense of purpose for their sacrifices, at
least temporarily.
what rises to the level that I as a consumer should pay attention[?] Certainly development process does not. Corrupt labor practices, unchecked by law, might.
I don't know if I made clear enough that
different people will feel different levels
of responsibility towards companies they pay,
and I'm not judging anybody.
In my ideal world, shoddy development practices
like 100 hours of work a week will show up in
the final product, and consumers will recognize
and reject it. All without really paying
attention to how things are done, because the
end result speaks for itself.
In the less than ideal real world, if you don't
want to contribute to a cycle of something you
dislike, then you need to take stand with your
dollars. It will mean that you get new games
fewer and farther in between, and they might
even be more expensive, but maybe the developers
will manage to stay in the business longer, and
maybe make a better game.
That's what stupid life does. Every
other species on this earth must rely on
nature to throttle back their overpopulation,
but humans now have the knowledge to control
our own population. Several countries already
have a negative population growth rate,
despite prosperity.
So, yes, we can choose not to mine the moon
for whatever reason. After all, many countries
are choosing not to clone humans.
I'm not going to check on [refusal to pay overtime] before I buy a game though.
An emerging line of thinking is that consumers
can and should gain visibility into how things
are made, and actively choose the companies that
use processes they like. This is the line of
thinking that pushes Nike to end child labor
practices, tuna fishermen to use dolphin-friendly
nets, etc.
In this case, individuals have very little
recourse against an abusive company. First of
all, the job market is really bad. Secondly,
they are probably well paid, relative to the
rest of the society, and are probably "exempt"
employees. Finally, court cases can take
many years and much expense, and is rightfully
only a matter of the last resort.
I'm not trying to be sanctimonious. I'm just
pointing out that you have in your hand the
power to affect the way these companies do
business, and you're choosing to ignore it.
That's not wrong, but I find it unfortunate.
It should work for every PhD candidate doing work in Oxford on computational number theory, like he is.
Not if you're named John Smith. Google has its
limits.
Besides, as cool as that sounds (yes, I have a
graduate degree from an internationally famous
university, so I'm not just jealous), it's
no guarantee that people will even acknowledge
your resume these days. I really wonder when
this person last looked for a job the hard way.
I draw the line at my name. If that, and Google, isn't enough for a potential employer to know if they are interested enough to interview me, I'm not interested in working for them.
How many jobs have you actually found, initially
supplying only your name?
You think maybe this tactic, even if it does
work for you in this terrible job market, will
work for somebody with a rather common name?
that's just another disincentive for me to go out & buy a Mac.
There will always be a faster or otherwise
better computer coming soon.
Most people buy a computer when the need comes
up, such as when the old one is broken. If
it's simply too slow, then they can generally
wait a month or two if they think that something
much better is coming up really soon.
If you can wait up to 18 months, then at any
point in time you should wait the 18 months.
Your money will go that much further, thanks
to Moore's Law. You're simply not the target market.
I can't understand why so many insist on comparing Intel's 32 bit solutions to this chip which is a 64 bit solution.
Because at the end of the day, it's how fast a
task completes that we're concerned about. If
Intel's 32-bit chips get it done faster, then
that's what many people will use. (Other people
consider more than raw speed, of course.)
I think the days of selling computers based on Mhz just drew to a close.
You might be right, but I'm afraid that it'll
merely be replaced with another simple but
meaningless indicator of system speed. The
main problem that people don't want to
educate themselves about these things.
I would argue that operating systems target a class of architecture -- not just the CPU, but the underlying system as well.
Yes. This trait is referred to as the scalability of the OS. A more scalable design is usually a good thing.
some chips are faster at certain OS-related things [...], but that seems more a nod to usability than a nod to any specific design of
OS.
The 80386 is a really good example of what I'm
talking about. Following closely after the
80286, it introduced a new, incompatible
protected mode, as well as paging (paged
segmentation, to be precise). These are non-trivial features, and cater specifically
to the post MS-DOS market, enabling the rise
of workstation class OSes like Windows NT
and Linux. Obviously, the more OS designs you
can readily support the better.
The virtual 8086 mode, on the other hand,
seems to suggest a vision of a multi-tasking
OS running multiple sessions of MS-DOS. It
certainly doesn't point in the Unix direction.
hardware MMU/paging isn't necesarily required for multitasking, it just makes it far more efficient, and it provides you a level of protection
I was being loose on the word "required",
referring mostly to marketing and customer
requirements, rather than strict technical
ones. Sorry for the confusion.
That's mostly right, but not entirely. CPUs
do target specific classes of operating systems.
Earlier CPUs like the 8086 don't support memory
protection, or paging. These are required for
a multitasking, multi-user OS, and were added
when CPUs began to target this sort of OS.
Linux, for example, basically won't work without
a MMU.
Also, the target OS (or class of OS) also affects optimization choices. For example, a microkernel OS is likely to have many more user-kernel
mode transitions than a monolithic OS, so if
you were targeting microkernels you'd want to
optimize that somehow.
You can call it quid pro quo if you want, but the only real solution to copyright infringement is exactly more reasonable prices and stricter enforcement.
Even for just one component, you're far better off buying the bundle here in the US.
That's part of the point. If Microsoft "unbundles" Word from Office and sells it for just a small fraction less, then it's really not giving any meaningful customer choice.
Word should, by all rights, cost at most half
of Office. I would not be at all surprised if
that still gave Microsoft the same profit as
the entire Office. The bundling is more likely
intended at leveraging one monopoly to create
another.
the real safeguard comes when every single individual in the nation is armed, and willing to die. It's less likely a solider is going to be so eager to destroy the rights of his own people if he knows that means killing his own friends, neighbors, and family in the process.
The Tienanmen Square Massacre is an instructive example. The troops actually around Beijing
were well informed of the people's demands,
and sympathized with the people. In fact,
they refused orders to attack. Beijing had
to move uninformed troops from elsewhere to
carry out the attack, telling them that the
people in the Square were enemies.
The People Power Revolution happened because
the troops were informed, and did not fire
on the civilians.
The key is whether the troops know what is
going on. Soldiers are trained specifically
to attack people who are well-armed and
willing to die, and will probably follow
orders unless they know what they're doing
is wrong.
Bears don't commit armed robbery. [...]
The Second Amendment was put in place by our founding fathers [...]
You managed to completely miss the point.
Once you invoke the Second Amendment as a
moral justification to participate in the
design of a "controversial" device, you've
made an ethical decision on the technology.
You've decided that ethical considerations
in favor of the technology outweigh the
potential abuses.
Therefore proving my point that technology
is not independent of ethics. Engineers
shouldn't go to work completely oblivious
of the uses of the technology they develop.
As for the Second Amendment, most people
don't think Iraq's armed forces stands a
chance against the US. Do you think your
"well-regulated militia" really stands a
chance if the US Armed Forces can be turned
on its citizens? (IOW, the real safeguard
of your liberties comes from the Armed
Forces siding with the people in such an
event, not with an independent militia.)
what if the body armor is wore by gun touting armed robbers
All I meant by this example is that some
technologies have only one purpose. In this
case, to kill a human, because only humans
wear body armor. If you have a problem with
killing humans of any sort, you should stay
away from a job developing armor piercing
bullets. I said the technology is not ethics
neutral, I didn't say it was unethical.
The engineer must come to an independent
conclusion on whether he/she can live with
all its effects.
life is not black and white as some portray
Exactly. To see technology as completely ethics neutral is to ignore the gray areas.
all the more reason to invade them rather than North Korea - Iraq won't launch nukes it doesn't have. Attacking North Korea in no way assures they won't get a nuke off
Let's see if I got this right.
Invade Iraq because it wants nukes. In fact,
better invade them now before they surely
have nukes.
Don't invade North Korea, because they have
nukes.
Sounds like the logical thing for any country
to do is to hide their nuclear program as
well as possible until you're ready to
detonate one. Is that the world order
America really wants to see?
I submit that if America - as the sole
superpower - respects and treats each country
without concern for nuclear capability, then
people wouldn't be working so hard to get
some.
Slashdot Mirror
Initially, yes. However, it lasted this long because Intel worked very hard to keep it alive. If the x86 trailed, for example, the PowerPC-based Macintosh by 50% in performance, many things may be very different.
Had that not happened, x86 would be at best a footnote, along with the 65XX, Z80, etc.
The 6502 and Z-80 are not "footnotes". They deserve prominent spots in CPU history marking the beginning of personal computing and affordable gaming consoles. When the x86's time finally comes, it will also be a major milestone marking the maturing of personal computing.
Depends on what you mean by good. If you mean the Darwinian sense, then yes, it's phenomenally successful.
However, you write like a person who has never had to work under the 8086 real mode in assembly language. Here are a few things wrong with it (the whole family, over the years):
- Too few registers
- Registers have special purposes, and are
not generic enough
- Many instructions are very rarely used
- Did not have a supervisor mode (pre 386)
or MMU support
- Unbelievably lame 16-bit segmentation
- Overcomplicated memory protection (few
if any OSes take advantage of segmentation)
These are design failings that are not "in the eye of the beholder". Intel overcame the first two by going to a hidden RISCy core with many more registers, the third by implementing many rarely used instructions in microcode, the next two by essentially discarding the 8086 and 80286 architectures in going to the 80386. Intel deserves a lot of credit, but they had to work very hard to overcome these problems.Comparing it to the 68000 is left as an exercise for the reader.
Here. They're a little pricey, but that's what you pay for getting out of the mainstream. They can run BSD or Linux, and a development kit (including a really good IDE) is available for free. If you can't afford a new one, there's a vibrant used market as well.
I'm not joking. A Mac is the best way to experiment outside the x86, partly because it can still double as a real personal computer and run popular apps.
In fact, an ISP can tell you that you're paying for (and entitled to) X GB per month. Many of them use an approximation of "unlimited" access, but I've known people warned for "abusing" supposedly "unlimited" dial-up.
Now, the question is whether Cornell is in effect changing the terms of service illegally?
They have no choice of ISP's.
That's no different than where I live. With no access to DSL (and apartments typically don't like satellite dishes), a cable monopoly is my only reasonable choice of broadband provider. Why is that special?
So you would hire a 14 year old high school drop-out orphan to design a web site for you Fortune 10 company, as long as there's a proper contract? I expect that he will be one of the lowest bidders.
My point was that a music vendor would rather sell you a $15 CD, and make a $10 profit (for example), rather than try to sell you 15 different songs at $1, even if each song still earns the same profit margin of 66 cents. Oversimplified, they make $10 by putting 3 good songs in an album right now. Selling individually, those same songs get them less than $2, so they must make more good stuff to get the same revenue.
So, no, I'm not certain they're happy about this.
there are at least a few songs in a typical album that I will enjoy that I didn't hear before buying it.
Excellent point. Cheap downloadable singles will also place pressure on the producers to pack the album with better stuff.
I wasn't commenting on the merits of that decision, just showing that humans are not bound by some endless need to reproduce. In the case of cloning, many made an ethical decision to not pursue one form of reproduction.
Gosh, no. If you're allowed to purchase low-value goods piecemeal, you might buy it from someone else next time. That's why you have to buy 500 matches at a time, even if you just need one.
The technique used by VMWare is known as virtualization, which still relies on having an x86 chip to actually process instructions. For VMWare to support the PowerPC, they have to essentially create a new product. MOL and Plex86 also use the same technique.
Of course, with 1.3 GHz Durons costing just $30 in retail, I wonder why nobody has come up with a x86-on-a-PCI with (a lot less) associated software to solve this problem.
China is also at risk of political meltdown. Rapid introduction of capitalism over the past two decades dramatically increased the gap between the rich and the poor. The violation of long espoused communist ideals is pretty clear, and not easy to overlook if you're the ones left behind. Serious tension is inevitable over the coming years.
BHAGs unite the country and give people a sense of purpose for their sacrifices, at least temporarily.
I don't know if I made clear enough that different people will feel different levels of responsibility towards companies they pay, and I'm not judging anybody.
In my ideal world, shoddy development practices like 100 hours of work a week will show up in the final product, and consumers will recognize and reject it. All without really paying attention to how things are done, because the end result speaks for itself.
In the less than ideal real world, if you don't want to contribute to a cycle of something you dislike, then you need to take stand with your dollars. It will mean that you get new games fewer and farther in between, and they might even be more expensive, but maybe the developers will manage to stay in the business longer, and maybe make a better game.
That's what stupid life does. Every other species on this earth must rely on nature to throttle back their overpopulation, but humans now have the knowledge to control our own population. Several countries already have a negative population growth rate, despite prosperity.
So, yes, we can choose not to mine the moon for whatever reason. After all, many countries are choosing not to clone humans.
An emerging line of thinking is that consumers can and should gain visibility into how things are made, and actively choose the companies that use processes they like. This is the line of thinking that pushes Nike to end child labor practices, tuna fishermen to use dolphin-friendly nets, etc.
In this case, individuals have very little recourse against an abusive company. First of all, the job market is really bad. Secondly, they are probably well paid, relative to the rest of the society, and are probably "exempt" employees. Finally, court cases can take many years and much expense, and is rightfully only a matter of the last resort.
I'm not trying to be sanctimonious. I'm just pointing out that you have in your hand the power to affect the way these companies do business, and you're choosing to ignore it. That's not wrong, but I find it unfortunate.
Not if you're named John Smith. Google has its limits.
Besides, as cool as that sounds (yes, I have a graduate degree from an internationally famous university, so I'm not just jealous), it's no guarantee that people will even acknowledge your resume these days. I really wonder when this person last looked for a job the hard way.
How many jobs have you actually found, initially supplying only your name?
You think maybe this tactic, even if it does work for you in this terrible job market, will work for somebody with a rather common name?
There will always be a faster or otherwise better computer coming soon.
Most people buy a computer when the need comes up, such as when the old one is broken. If it's simply too slow, then they can generally wait a month or two if they think that something much better is coming up really soon.
If you can wait up to 18 months, then at any point in time you should wait the 18 months. Your money will go that much further, thanks to Moore's Law. You're simply not the target market.
Because at the end of the day, it's how fast a task completes that we're concerned about. If Intel's 32-bit chips get it done faster, then that's what many people will use. (Other people consider more than raw speed, of course.)
I think the days of selling computers based on Mhz just drew to a close.
You might be right, but I'm afraid that it'll merely be replaced with another simple but meaningless indicator of system speed. The main problem that people don't want to educate themselves about these things.
Yes. This trait is referred to as the scalability of the OS. A more scalable design is usually a good thing.
some chips are faster at certain OS-related things [...], but that seems more a nod to usability than a nod to any specific design of OS.
The 80386 is a really good example of what I'm talking about. Following closely after the 80286, it introduced a new, incompatible protected mode, as well as paging (paged segmentation, to be precise). These are non-trivial features, and cater specifically to the post MS-DOS market, enabling the rise of workstation class OSes like Windows NT and Linux. Obviously, the more OS designs you can readily support the better.
The virtual 8086 mode, on the other hand, seems to suggest a vision of a multi-tasking OS running multiple sessions of MS-DOS. It certainly doesn't point in the Unix direction.
hardware MMU/paging isn't necesarily required for multitasking, it just makes it far more efficient, and it provides you a level of protection
I was being loose on the word "required", referring mostly to marketing and customer requirements, rather than strict technical ones. Sorry for the confusion.
That's mostly right, but not entirely. CPUs do target specific classes of operating systems.
Earlier CPUs like the 8086 don't support memory protection, or paging. These are required for a multitasking, multi-user OS, and were added when CPUs began to target this sort of OS. Linux, for example, basically won't work without a MMU.
Also, the target OS (or class of OS) also affects optimization choices. For example, a microkernel OS is likely to have many more user-kernel mode transitions than a monolithic OS, so if you were targeting microkernels you'd want to optimize that somehow.
You can call it quid pro quo if you want, but the only real solution to copyright infringement is exactly more reasonable prices and stricter enforcement.
That's part of the point. If Microsoft "unbundles" Word from Office and sells it for just a small fraction less, then it's really not giving any meaningful customer choice.
Word should, by all rights, cost at most half of Office. I would not be at all surprised if that still gave Microsoft the same profit as the entire Office. The bundling is more likely intended at leveraging one monopoly to create another.
The Tienanmen Square Massacre is an instructive example. The troops actually around Beijing were well informed of the people's demands, and sympathized with the people. In fact, they refused orders to attack. Beijing had to move uninformed troops from elsewhere to carry out the attack, telling them that the people in the Square were enemies.
The People Power Revolution happened because the troops were informed, and did not fire on the civilians.
The key is whether the troops know what is going on. Soldiers are trained specifically to attack people who are well-armed and willing to die, and will probably follow orders unless they know what they're doing is wrong.
You managed to completely miss the point. Once you invoke the Second Amendment as a moral justification to participate in the design of a "controversial" device, you've made an ethical decision on the technology. You've decided that ethical considerations in favor of the technology outweigh the potential abuses.
Therefore proving my point that technology is not independent of ethics. Engineers shouldn't go to work completely oblivious of the uses of the technology they develop.
As for the Second Amendment, most people don't think Iraq's armed forces stands a chance against the US. Do you think your "well-regulated militia" really stands a chance if the US Armed Forces can be turned on its citizens? (IOW, the real safeguard of your liberties comes from the Armed Forces siding with the people in such an event, not with an independent militia.)
All I meant by this example is that some technologies have only one purpose. In this case, to kill a human, because only humans wear body armor. If you have a problem with killing humans of any sort, you should stay away from a job developing armor piercing bullets. I said the technology is not ethics neutral, I didn't say it was unethical. The engineer must come to an independent conclusion on whether he/she can live with all its effects.
life is not black and white as some portray
Exactly. To see technology as completely ethics neutral is to ignore the gray areas.
Let's see if I got this right.
Invade Iraq because it wants nukes. In fact, better invade them now before they surely have nukes.
Don't invade North Korea, because they have nukes.
Sounds like the logical thing for any country to do is to hide their nuclear program as well as possible until you're ready to detonate one. Is that the world order America really wants to see?
I submit that if America - as the sole superpower - respects and treats each country without concern for nuclear capability, then people wouldn't be working so hard to get some.