Proposed System

On 9 Oct 2003 10:12:09 -0700, (MikeW) wrote:


And what's this obsession with undervolting. I can see not wanting to
overclock, but if you run the chips as designed, you can probably keep
them cool enough without too much noise, with intelligent case/cooling
system design.

Why not undervolt? So long as it's not such a low voltage to
intruduce instability there's nothing but benefit to it... due to the
way Intel tiers their CPUs in voltage groups, almost all of 'em but
the early releases at the highest speeds (per core revision) can run
undervolted, even overclocked up to a point.


Dave

On Thu, 09 Oct 2003 22:41:00 GMT
kony wrote:

On 9 Oct 2003 10:12:09 -0700, (MikeW) wrote:


And what's this obsession with undervolting. I can see not wanting
to overclock, but if you run the chips as designed, you can probably
keep them cool enough without too much noise, with intelligent
case/cooling system design.

Why not undervolt? So long as it's not such a low voltage to
intruduce instability there's nothing but benefit to it... due to the
way Intel tiers their CPUs in voltage groups, almost all of 'em but
the early releases at the highest speeds (per core revision) can run
undervolted, even overclocked up to a point.

It's like overclocking I guess--some people do it because they can.
With passive coolers available for every processor currently on the
market though there's no need to do it to achieve a quiet machine
though.


Dave


--
--
--John
Reply to jclarke at ae tee tee global dot net
(was jclarke at eye bee em dot net)

On Fri, 10 Oct 2003 00:45:11 +0000, "J.Clarke"
wrote:

On Thu, 09 Oct 2003 22:41:00 GMT
kony wrote:

On 9 Oct 2003 10:12:09 -0700, (MikeW) wrote:


And what's this obsession with undervolting. I can see not wanting
to overclock, but if you run the chips as designed, you can probably
keep them cool enough without too much noise, with intelligent
case/cooling system design.

Why not undervolt? So long as it's not such a low voltage to
intruduce instability there's nothing but benefit to it... due to the
way Intel tiers their CPUs in voltage groups, almost all of 'em but
the early releases at the highest speeds (per core revision) can run
undervolted, even overclocked up to a point.

It's like overclocking I guess--some people do it because they can.
With passive coolers available for every processor currently on the
market though there's no need to do it to achieve a quiet machine
though.


There aren't truely passive coolers available for AMD or Intel though,
they require a very dedicated fan, airflow, might as well be
considered active coolers with the fan simply moved or put to take for
multiple functions as with Dell ducted systems.

It really is only for simplicities' sake that most CPUs use the
excessive voltage that they do. It would be more confusing if Intel
released CPUs each using only as much voltage as needed, instead of
same voltage per family, up until more voltage is needed at a certain
speed-grade to keep initial yields up.

Undervoltage not only decreases heat, it lessens wear on the
motherboard, uses less energy overall, including a slight bit less to
re-cool the room the system is in during the warmer months. Given
Intel's path towards even hotter CPUs I expect we'll see more and more
people undervolting their CPUs.


Dave

On Fri, 10 Oct 2003 08:11:20 GMT
kony wrote:

On Fri, 10 Oct 2003 00:45:11 +0000, "J.Clarke"
wrote:

On Thu, 09 Oct 2003 22:41:00 GMT
kony wrote:

On 9 Oct 2003 10:12:09 -0700, (MikeW) wrote:


And what's this obsession with undervolting. I can see not
wanting to overclock, but if you run the chips as designed, you can
probably keep them cool enough without too much noise, with
intelligent case/cooling system design.

Why not undervolt? So long as it's not such a low voltage to
intruduce instability there's nothing but benefit to it... due to
the way Intel tiers their CPUs in voltage groups, almost all of 'em
but the early releases at the highest speeds (per core revision) can
run undervolted, even overclocked up to a point.

It's like overclocking I guess--some people do it because they can.
With passive coolers available for every processor currently on the
market though there's no need to do it to achieve a quiet machine
though.


There aren't truely passive coolers available for AMD or Intel though,
they require a very dedicated fan, airflow, might as well be
considered active coolers with the fan simply moved or put to take for
multiple functions as with Dell ducted systems.

Well, actually passive coolers have been constructed for AMD CPUs. And
since "everybody knows" that "Intel runs cooler" there should be no
problem doing the same for an Intel. Not off-the-shelf items but it has
been done.

Regardless, I've been running a Thunderbird 1400, which is one of the
hardest CPUs to keep cool, for several years now using the quietest fan
that Papst makes and it's been working fine. And there are bigger and
better heat sinks available now than there were then.

It really is only for simplicities' sake that most CPUs use the
excessive voltage that they do. It would be more confusing if Intel
released CPUs each using only as much voltage as needed, instead of
same voltage per family, up until more voltage is needed at a certain
speed-grade to keep initial yields up.

Uh-huh. Generally speaking I prefer stable systems, and to me that
means running everything within spec and not second guessing engineers
who have vastly more resources at their disposal than I do. But to each
his own. If you are certain that all CPUs run at "excessive voltage" I
would like to see your sources, and don't show me some overclocker site,
show me the calculations that demonstrate this and the test results that
confirm them.

Undervoltage not only decreases heat, it lessens wear on the
motherboard,

ROF,L. Any motherboard you buy today, unless something is defective,
will still be running fine long after it is so far obsolete that you
won't be able to give it away.

uses less energy overall, including a slight bit less to
re-cool the room the system is in during the warmer months.

How much less?

Given
Intel's path towards even hotter CPUs I expect we'll see more and more
people undervolting their CPUs.

Why? We don't see many AMD people undervolting, and they've got more
incentive than Intel people.

Running a CPU or any other component out of spec is something you get
away with, not correction of an error on the part of the designers.
Trying to sell it as anything else does nobody a service.




Dave






--
--
--John
Reply to jclarke at ae tee tee global dot net
(was jclarke at eye bee em dot net)

On Fri, 10 Oct 2003 11:27:47 GMT
(Thunder9) wrote:

On Fri, 10 Oct 2003 08:11:20 GMT, kony wrote:

On Fri, 10 Oct 2003 00:45:11 +0000, "J.Clarke"
wrote:

On Thu, 09 Oct 2003 22:41:00 GMT
kony wrote:

On 9 Oct 2003 10:12:09 -0700, (MikeW) wrote:


And what's this obsession with undervolting. I can see not
wanting to overclock, but if you run the chips as designed, you
can probably keep them cool enough without too much noise, with
intelligent case/cooling system design.

Why not undervolt? So long as it's not such a low voltage to
intruduce instability there's nothing but benefit to it... due to
the way Intel tiers their CPUs in voltage groups, almost all of 'em
but the early releases at the highest speeds (per core revision)
can run undervolted, even overclocked up to a point.

It's like overclocking I guess--some people do it because they can.
With passive coolers available for every processor currently on the
market though there's no need to do it to achieve a quiet machine
though.


There aren't truely passive coolers available for AMD or Intel
though, they require a very dedicated fan, airflow, might as well be
considered active coolers with the fan simply moved or put to take
for multiple functions as with Dell ducted systems.

It really is only for simplicities' sake that most CPUs use the
excessive voltage that they do. It would be more confusing if Intel
released CPUs each using only as much voltage as needed, instead of
same voltage per family, up until more voltage is needed at a certain
speed-grade to keep initial yields up.

Undervoltage not only decreases heat, it lessens wear on the
motherboard, uses less energy overall, including a slight bit less to
re-cool the room the system is in during the warmer months. Given
Intel's path towards even hotter CPUs I expect we'll see more and
more people undervolting their CPUs.


I would like to see a standard in which components could be set to
limit their watts by throttling down their speed or voltage or
whatever. I know some systems do this internally, but an industry
wide standard for *setting* a component via software would be cool.
Thus you could have a control panel and click on "web browsing" and
voila each component would be told to operate at a certain maximum
performance/energy ratio. Click on "Extreme Gaming" and components
will be allowed to use more energy, along with the fans speeding up to
compensate.

Well it will probably never happen because then the average consumer
might discover that they don't need a P4 3.0 Ghz + 1GB RAM to surf the
web and check on email

All the pieces to do this are in place--there are boards that allow
voltage and clock speed changes while in operation, most boards today
have SMBus connectors that allow the temperatures to be reported to an
external device, all that you need to do is put together the necessary
software and the external fan controller and you're there. Do it and
make yourself rich.

But so far thermally controlled fans seem to be doing an adequate job so
nobody seems to have seen a need for the added complexity.

Thunder9



--
--
--John
Reply to jclarke at ae tee tee global dot net
(was jclarke at eye bee em dot net)

On Fri, 10 Oct 2003 08:11:20 GMT, kony wrote:

On Fri, 10 Oct 2003 00:45:11 +0000, "J.Clarke"
wrote:

On Thu, 09 Oct 2003 22:41:00 GMT
kony wrote:

On 9 Oct 2003 10:12:09 -0700, (MikeW) wrote:


And what's this obsession with undervolting. I can see not wanting
to overclock, but if you run the chips as designed, you can probably
keep them cool enough without too much noise, with intelligent
case/cooling system design.

Why not undervolt? So long as it's not such a low voltage to
intruduce instability there's nothing but benefit to it... due to the
way Intel tiers their CPUs in voltage groups, almost all of 'em but
the early releases at the highest speeds (per core revision) can run
undervolted, even overclocked up to a point.

It's like overclocking I guess--some people do it because they can.
With passive coolers available for every processor currently on the
market though there's no need to do it to achieve a quiet machine
though.


There aren't truely passive coolers available for AMD or Intel though,
they require a very dedicated fan, airflow, might as well be
considered active coolers with the fan simply moved or put to take for
multiple functions as with Dell ducted systems.

It really is only for simplicities' sake that most CPUs use the
excessive voltage that they do. It would be more confusing if Intel
released CPUs each using only as much voltage as needed, instead of
same voltage per family, up until more voltage is needed at a certain
speed-grade to keep initial yields up.

Undervoltage not only decreases heat, it lessens wear on the
motherboard, uses less energy overall, including a slight bit less to
re-cool the room the system is in during the warmer months. Given
Intel's path towards even hotter CPUs I expect we'll see more and more
people undervolting their CPUs.


I would like to see a standard in which components could be set to
limit their watts by throttling down their speed or voltage or
whatever. I know some systems do this internally, but an industry
wide standard for *setting* a component via software would be cool.
Thus you could have a control panel and click on "web browsing" and
voila each component would be told to operate at a certain maximum
performance/energy ratio. Click on "Extreme Gaming" and components
will be allowed to use more energy, along with the fans speeding up to
compensate.

Well it will probably never happen because then the average consumer
might discover that they don't need a P4 3.0 Ghz + 1GB RAM to surf the
web and check on email

Thunder9

On Fri, 10 Oct 2003 18:44:53 GMT
kony wrote:

On Fri, 10 Oct 2003 08:58:01 +0000, "J.Clarke"
wrote:


There aren't truely passive coolers available for AMD or Intel
though, they require a very dedicated fan, airflow, might as well be
considered active coolers with the fan simply moved or put to take
for multiple functions as with Dell ducted systems.

Well, actually passive coolers have been constructed for AMD CPUs.
And since "everybody knows" that "Intel runs cooler" there should be
no problem doing the same for an Intel. Not off-the-shelf items but
it has been done.

Please link a passive cooler that doesn't require a dedicated
fan/airflow as I described.

http://www.silentpcreview.com/module...e=Sections&fil
e=index&req=viewarticle&artid=114&page=1

Nobody knows that Intel runs cooler, it's been untrue from the point
when AMD stopped making Palominos.

Lighten up, OK?

Regardless, I've been running a Thunderbird 1400, which is one of the
hardest CPUs to keep cool, for several years now using the quietest
fan that Papst makes and it's been working fine. And there are
bigger and better heat sinks available now than there were then.

OK, and that is near the ceiling speed of the given core, which as I
wrote just below this, could need that stock voltage for AMD to
likewise get a good yield.

Uh, you miss the point--you are not going to get anything to run quieter
than that by undervolting.

It really is only for simplicities' sake that most CPUs use the
excessive voltage that they do. It would be more confusing if
Intel released CPUs each using only as much voltage as needed,
instead of same voltage per family, up until more voltage is needed
at a certain speed-grade to keep initial yields up.

Uh-huh. Generally speaking I prefer stable systems, and to me that
means running everything within spec and not second guessing
engineers who have vastly more resources at their disposal than I do.
But to each
his own. If you are certain that all CPUs run at "excessive voltage"
I would like to see your sources, and don't show me some overclocker
site, show me the calculations that demonstrate this and the test
results that confirm them.

Reread what I wrote just above this paragraph. Engineers do NOT
choose the voltage that particular CPU needs to run,

Please provide a link to a statement by an engineer working for Intel
that that is the case for the P4.

they set the
voltage for a family, subgoup of CPUs, you are not second-guessing
them when the spec'd operating voltage is offset like this.

If you are using a voltage other than the one that they tell you to use
then you are indeed second-guessing them.

Some CPUs
DO need the spec'd voltage but others don't. Knowledge and/or
experience with the given CPU can be needed.

And where do you get that "knowledge"?

This is common knowledge for many years. I'll let you do your own
Google search instead of re-educating you myself.

Nope, doesn't work that way, you made the assertion it's up to you to
back it up.

The primary reason
people don't as often undervolt is that they're looking to INCREASE
performance, not looking for cooler operation at the same performance
or lowering performance a little for even more heat reduction.

The primary reason that "people don't as often undervolt" is that there
is no percieved need for it.

There's more info needed to bring you "up to speed" than a newsgroup
post of a couple of links could provide. HERE you have already seen
evidence of it, with NO evidence that it isn't possible.

I have seen no "evidence". I have seen you make some unsupported
assertions.

As for
"reasonable", that depends on the user... for some people it's
reasonable to have a system sounding like a leaf-blower, but for
others it's unfortunate when they have to replace their failed P2
system.

I see. So the alternatives are to undervolt or to "sound like a leaf
blower"? Sorry, but now you're engaging in hyperbole. There are a
number of heat sinks on the market which can be used to cool any
processor currently in production using the quietest fans currently in
production, without operating the processor at a voltage level outside
the specified range.

ROF,L. Any motherboard you buy today, unless something is defective,
will still be running fine long after it is so far obsolete that you
won't be able to give it away.

"Obsolete" is relative to the users needs, but also the overall speed
of the system, and the capabilities of the OS it can effectively run.
Many people need a minimal level of basic function that requires
faster than a couple hundred MHz CPU, but moving forward to today, a
3GHz system could be useful to them for much longer than it'll last.

So how many users _are_ still running 1975-vintage CP/M machines? And
yet those motherboards, running at full specified voltage, are still
going strong.

Go read capacitor manufacturer's spec sheets if you think it doesn't
matter. I don't need to provide a link because you can pick any
manufacturer you wish, go to their website... for example the caps on
your best motherboard.

In other words you don't have a link that supports your assertion.

uses less energy overall, including a slight bit less to
re-cool the room the system is in during the warmer months.

How much less?

Energy consumption of the CPU is proportional to the square of the
voltage. It's not a huge difference, but IS a significant difference.
Do the math yourself, see Intel's spec sheet and calculate out how
much power reduction results from a given voltage drop. Typical
results can be from 5-25 percent reduction, but depend on the CPU
core, speed it's running. A 3.2GHz P4 isn't going to tolerate as much
voltage drop as a 2.5GHz for example.

Please be kind enough to provide a link to this spec sheet.

Given
Intel's path towards even hotter CPUs I expect we'll see more and
more people undervolting their CPUs.

Why? We don't see many AMD people undervolting, and they've got more
incentive than Intel people.

Current Intel CPUs run hotter at full load. Read the spec sheets,
they're on the respective manufacturer's websites. At idle a P4 "can"
run cooler, but the system must run at up to full-load also.

I see. Intel CPUs runn hotter at full load? So an Intel CPU with a
Vapochill attached will run hotter than an Athlon with no heat sink?

Now, would you care to use a term that is more precise than "run
hotter"?

I used to have an XP1600 running at 1.6V, o'c to 1.6 GHz. The main
reason you don't see many people undervolting, is that typically the
knowledge gained towards understanding the frequency/voltage
relationship, is gained as a result of a desire to o'c the CPU as much
as reasonably possible.

You're repeating yourself.

Running a CPU or any other component out of spec is something you get
away with, not correction of an error on the part of the designers.
Trying to sell it as anything else does nobody a service.

It's not an "error" of the designer, the designers know this. It is
for practicality, and in some cases, initially for yields. This
makes it easier to implement the CPU in a broad range of motherboards
and providing a large stability margin to ensure the entire line of
CPUs will work, even in poor systm environments, including those that
might not work at a lower voltage because of poor yields earlier in
the production. Nobody knows what exact gear user "X" is going to
have except for user X.

It's a choice... nobody is focing you to undervolt your CPU. It works
fine if you know what you're doing.

Yes, it works fine if you _know_ _what_ _you_ _are_ _doing_, which means
that you are an electrical engineer with IC design experience,
an intimate familiarity with the particular device under
consideration, and you know what constitute the worst cases
that need to be tested to confirm reliable operation.

Skepticism could be due to lack
of information... at one point in history there were plenty of
skeptics that thought the world was flat.

So provide a source for this information. Skepticism can also be due to
having seen more than one cocky young kid standing on his weenie after
his wild assertions proved to be somewhat in error.

Dave

--
--
--John
Reply to jclarke at ae tee tee global dot net
(was jclarke at eye bee em dot net)

On Fri, 10 Oct 2003 08:58:01 +0000, "J.Clarke"
wrote:


There aren't truely passive coolers available for AMD or Intel though,
they require a very dedicated fan, airflow, might as well be
considered active coolers with the fan simply moved or put to take for
multiple functions as with Dell ducted systems.

Well, actually passive coolers have been constructed for AMD CPUs. And
since "everybody knows" that "Intel runs cooler" there should be no
problem doing the same for an Intel. Not off-the-shelf items but it has
been done.

Please link a passive cooler that doesn't require a dedicated
fan/airflow as I described.

Nobody knows that Intel runs cooler, it's been untrue from the point
when AMD stopped making Palominos.

Regardless, I've been running a Thunderbird 1400, which is one of the
hardest CPUs to keep cool, for several years now using the quietest fan
that Papst makes and it's been working fine. And there are bigger and
better heat sinks available now than there were then.

OK, and that is near the ceiling speed of the given core, which as I
wrote just below this, could need that stock voltage for AMD to
likewise get a good yield.


It really is only for simplicities' sake that most CPUs use the
excessive voltage that they do. It would be more confusing if Intel
released CPUs each using only as much voltage as needed, instead of
same voltage per family, up until more voltage is needed at a certain
speed-grade to keep initial yields up.

Uh-huh. Generally speaking I prefer stable systems, and to me that
means running everything within spec and not second guessing engineers
who have vastly more resources at their disposal than I do. But to each
his own. If you are certain that all CPUs run at "excessive voltage" I
would like to see your sources, and don't show me some overclocker site,
show me the calculations that demonstrate this and the test results that
confirm them.

Reread what I wrote just above this paragraph. Engineers do NOT
choose the voltage that particular CPU needs to run, they set the
voltage for a family, subgoup of CPUs, you are not second-guessing
them when the spec'd operating voltage is offset like this. Some CPUs
DO need the spec'd voltage but others don't. Knowledge and/or
experience with the given CPU can be needed.

This is common knowledge for many years. I'll let you do your own
Google search instead of re-educating you myself. The primary reason
people don't as often undervolt is that they're looking to INCREASE
performance, not looking for cooler operation at the same performance
or lowering performance a little for even more heat reduction.

There's more info needed to bring you "up to speed" than a newsgroup
post of a couple of links could provide. HERE you have already seen
evidence of it, with NO evidence that it isn't possible. As for
"reasonable", that depends on the user... for some people it's
reasonable to have a system sounding like a leaf-blower, but for
others it's unfortunate when they have to replace their failed P2
system.


ROF,L. Any motherboard you buy today, unless something is defective,
will still be running fine long after it is so far obsolete that you
won't be able to give it away.

"Obsolete" is relative to the users needs, but also the overall speed
of the system, and the capabilities of the OS it can effectively run.
Many people need a minimal level of basic function that requires
faster than a couple hundred MHz CPU, but moving forward to today, a
3GHz system could be useful to them for much longer than it'll last.

Go read capacitor manufacturer's spec sheets if you think it doesn't
matter. I don't need to provide a link because you can pick any
manufacturer you wish, go to their website... for example the caps on
your best motherboard.

uses less energy overall, including a slight bit less to
re-cool the room the system is in during the warmer months.

How much less?

Energy consumption of the CPU is proportional to the square of the
voltage. It's not a huge difference, but IS a significant difference.
Do the math yourself, see Intel's spec sheet and calculate out how
much power reduction results from a given voltage drop. Typical
results can be from 5-25 percent reduction, but depend on the CPU
core, speed it's running. A 3.2GHz P4 isn't going to tolerate as much
voltage drop as a 2.5GHz for example.


Given
Intel's path towards even hotter CPUs I expect we'll see more and more
people undervolting their CPUs.

Why? We don't see many AMD people undervolting, and they've got more
incentive than Intel people.

Current Intel CPUs run hotter at full load. Read the spec sheets,
they're on the respective manufacturer's websites. At idle a P4 "can"
run cooler, but the system must run at up to full-load also.

I used to have an XP1600 running at 1.6V, o'c to 1.6 GHz. The main
reason you don't see many people undervolting, is that typically the
knowledge gained towards understanding the frequency/voltage
relationship, is gained as a result of a desire to o'c the CPU as much
as reasonably possible.



Running a CPU or any other component out of spec is something you get
away with, not correction of an error on the part of the designers.
Trying to sell it as anything else does nobody a service.

It's not an "error" of the designer, the designers know this. It is
for practicality, and in some cases, initially for yields. This
makes it easier to implement the CPU in a broad range of motherboards
and providing a large stability margin to ensure the entire line of
CPUs will work, even in poor systm environments, including those that
might not work at a lower voltage because of poor yields earlier in
the production. Nobody knows what exact gear user "X" is going to
have except for user X.

It's a choice... nobody is focing you to undervolt your CPU. It works
fine if you know what you're doing. Skepticism could be due to lack
of information... at one point in history there were plenty of
skeptics that thought the world was flat.


Dave

On Fri, 10 Oct 2003 11:27:47 GMT, (Thunder9)
wrote:


I would like to see a standard in which components could be set to
limit their watts by throttling down their speed or voltage or
whatever. I know some systems do this internally, but an industry
wide standard for *setting* a component via software would be cool.
Thus you could have a control panel and click on "web browsing" and
voila each component would be told to operate at a certain maximum
performance/energy ratio. Click on "Extreme Gaming" and components
will be allowed to use more energy, along with the fans speeding up to
compensate.

Well it will probably never happen because then the average consumer
might discover that they don't need a P4 3.0 Ghz + 1GB RAM to surf the
web and check on email

Thunder9


It would be nice, and I think we're moving in that direction. For a
long time now the HALT instructions would reduce heat, power usage,
but of course the actual frequency needs dropped for the most sizeable
voltage decrease. I assume we'll see the most work on this put into
notebooks initially, but as an automatic function, not something the
user would have to manually change.


Dave

On Fri, 10 Oct 2003 21:34:49 GMT
kony wrote:

On Fri, 10 Oct 2003 15:49:59 +0000, "J.Clarke"
wrote:

Please link a passive cooler that doesn't require a dedicated
fan/airflow as I described.

http://www.silentpcreview.com/module...name=Sections&
fil e=index&req=viewarticle&artid=114&page=1

Ok, but the expense and unique system layout will prohibit it's use in
most systems. Unfortunately cost dictates most design decisions the
most.

You did not read the article. It works in any system as long as you
know how to design the thing and order the pieces.

Nobody knows that Intel runs cooler, it's been untrue from the
point when AMD stopped making Palominos.

Lighten up, OK?

Hmmm. Sometimes my writing seems easy to misinterpret... That's the
problem with usenet, you can't see the expressions on the faces of
those doing the writing... large amounts of info are condensed into a
few words to reduce post size and it comes out reading less friendly
when it wasn't meant to.



OK, and that is near the ceiling speed of the given core, which as
I wrote just below this, could need that stock voltage for AMD to
likewise get a good yield.

Uh, you miss the point--you are not going to get anything to run
quieter than that by undervolting.

You mean quieter than that special, unusal, expensive heatsink you
linked, right?

No I do not. My point, is that if I can adequately cool a Thunderbird
1400 using a Papst 8412NGL, then it is highly unlikely that you are
going to get a machine to run any quieter by undervolting.

I'm not talking theory here, I'm talking practice...
in practice I've undervolted MANY different CPUs to cheaply and easily
get the needed performance per the specific appliction. Granted these
were usually special-purpose systems.

And I'm talking practice here too--in practice I have never needed to
undervolt a CPU to get a machine to run quietly.

Remember something about that passive cooler you linked- It is quite
likely to make the motherboard run hotter, have a shorter lifespan.

So how long will the motherboard last?

Products are designed to make $, not necessarily to benefit users,
only give the IMPRESSION of benefit to users. Sometimes that
impression is larger than the real benefit, or ignores the detriments.

And mairzy doats and dozey doats. Which has about as much relevance as
that statement.

CPUs can run pretty hot with no problems, so long as stable, but
motherboards benefit from the reduced power supply to CPU, reduced
heat from CPU exhaust, and cooler resulting air. There are two most
common failure points I see on modern systems (excluding defect and
power surges)- Capacitors wearing out and fan failure. I have STACKS
of systems that have failed in this way. Sometimes I'll replace the
caps and have a board as good as new again. These are not any of
those defective caps from Taiwan often mentioned.

Statistics please.

Reread what I wrote just above this paragraph. Engineers do NOT
choose the voltage that particular CPU needs to run,

Please provide a link to a statement by an engineer working for Intel
that that is the case for the P4.

Please provide a link to a statement by an engineer working for Intel
that the P4 can only run stabily at the spec'd voltage.

You're the one making the assertion that chips can run undervolted.
It's up to you to support it.

The spec'd voltage is that at which Intel guarantees, warranties the
part to run at spec'd speed. Even using a different cooler voids
Intel's guarantee, but you seem to think a different cooler is OK.

Using a different cooler from what?

I don't need theories or statements, I've observed it myself many
times, and seen others have success in addition to plenty of reports
of it all over the 'net. I don't seek Intel's statements because I
have DIRECT evidence of it already.

So provide this evidence.

Systems I've set up undervolted
have run completely stable, including months, even over a year (maybe
two by now) of continuous uptime... Systems that have no instability
in actual operation or testing.

And you would stake your own life on the accuracy of the calculations
performed by those systems? Hint--there are malfunctions that do not
cause instability.

they set the
voltage for a family, subgoup of CPUs, you are not second-guessing
them when the spec'd operating voltage is offset like this.

If you are using a voltage other than the one that they tell you to
use then you are indeed second-guessing them.

No, a guess is only a guess until you have plenty of evidence. I have
machines in my own home running undervolted right now.

So? You are still second-guessing the engineers.

I will repeat that Intel's engineers are NOT specifing the minimum
voltage possible for stable operation, in their own minds. They do
not test each CPU for undervoltage operation, only spec a voltage for
entire groups of CPUs.

Now, you have just made an assertion. So can you provide a verifiable
statement made by one of those engineers that supports your assertion or
did you just pull it out of your ass?

Some CPUs
DO need the spec'd voltage but others don't. Knowledge and/or
experience with the given CPU can be needed.

And where do you get that "knowledge"?

My own testing, other's testing, all over the place... Experience...
like everything else in life.

And what evidence do you have that those "tests" actually evaluated the
worst case conditions?

This is common knowledge for many years. I'll let you do your own
Google search instead of re-educating you myself.

Nope, doesn't work that way, you made the assertion it's up to you to
back it up.

No, I stated what I know to be fact.

You are asserting that it is a fact. If it is such a well-known fact
then you should have little trouble providing the evidence to support
your assertion.

Constantly reproving what is fact is a wasteful endevor.

Your stating that something is fact does not make is so.

I could similarly aruge that when you disagree, you must provide
evidence that it is required to operate the CPU at the exact spec'd
voltage.

Nope. Doesn't work that way.

It is not my burden to educate someone. I already know it's
a fact because I've proven it time and time again, by actually doing
it, as have others.

So you claim. Since you can't present any evidence I am forced to
conclude that you are a liar.

The primary reason
people don't as often undervolt is that they're looking to INCREASE
performance, not looking for cooler operation at the same
performance or lowering performance a little for even more heat
reduction.

The primary reason that "people don't as often undervolt" is that
there is no percieved need for it.

Perception is subjective. Nobody is forced to undervolt. Nobody is
forced to build their own computer either.


I have seen no "evidence". I have seen you make some unsupported
assertions.

Again, it is not my burden to educate, it is a choice, and I choose
not to do so because you can find this info with Google search or
asking in many hardware forums.

In other words you'd rather be thought a liar than support your claims.

I am perfectly content leaving you to disbelieve while my undervolted
systems continue to work great, as do other's systems.

You are assuming it can't be done, without doing any research, even
after you've (so far) only received feedback that it CAN be done.

No, I've received your unsupported assertion that it can be done.
And since you are providing much bluster and no evidence your
credibility is rapidly diminishing.

And the fact that something_can_ be done doesn't mean it's a
good idea.

As for
"reasonable", that depends on the user... for some people it's
reasonable to have a system sounding like a leaf-blower, but for
others it's unfortunate when they have to replace their failed P2
system.

I see. So the alternatives are to undervolt or to "sound like a leaf
blower"? Sorry, but now you're engaging in hyperbole.

No, it was an example of what some consider reasonable.

I see. So what relevance does it have to the making of quiet machines?

There are a
number of heat sinks on the market which can be used to cool any
processor currently in production using the quietest fans currently
in production, without operating the processor at a voltage level
outside the specified range.

Yes, again these are usually rather expensive, not always fitting on
the motherboard. A better heatsink or quieter fan IS one alternative,
but not the only alternative.

Well, actually, a quieter fan is the only alternative since no matter
how much you undervolt if you don't put in a quieter fan the noise level
doesn't change.

"Obsolete" is relative to the users needs, but also the overall
speed of the system, and the capabilities of the OS it can
effectively run. Many people need a minimal level of basic function
that requires faster than a couple hundred MHz CPU, but moving
forward to today, a 3GHz system could be useful to them for much
longer than it'll last.

So how many users _are_ still running 1975-vintage CP/M machines?
And yet those motherboards, running at full specified voltage, are
still going strong.


Reread what I wrote. I specifically mentioned a minimum level of
performance needed, which would be exceeded by huge margin with a 3GHz
system.

The issue was motherboards becoming obsolete before they fail. I
remember a time when a 12 MHz system exceeded by a huge margin the
"minimum level of performance needed". Now those 12 MHz systems are
still running fine, but for what purpose are they useful today?

Low-powered systems do not need voltage reduction as much, the
power-density of the system is lower... see the commonality here? By
reducing the voltage, you're coming nearer to a low-powered system.

If you want a low powered system then just buy one.

Go read capacitor manufacturer's spec sheets if you think it
doesn't matter. I don't need to provide a link because you can pick
any manufacturer you wish, go to their website... for example the
caps on your best motherboard.

In other words you don't have a link that supports your assertion.

I don't need to prove it to anyone because I have/had, dozens of
systems running fine undervolted.

Liar.

Again, it doesn't matter to me if you agree, because at this very
monent plenty of people use undervolted systems.

If I wrote that there is a sun in the sky, and you disagreed, would I
really need to prove it? Instead I tell you to "look up". Likewise,
here I tell you to "look IT up".

No, because it is generally accepted that there is a sun in the sky.
Your assertion that processors may be run reliably under conditions at
variance with the manufacturer's specification is akin to the assertion
that there is no sun.

s?

Energy consumption of the CPU is proportional to the square of the
voltage. It's not a huge difference, but IS a significant
difference. Do the math yourself, see Intel's spec sheet and
calculate out how much power reduction results from a given voltage
drop. Typical results can be from 5-25 percent reduction, but
depend on the CPU core, speed it's running. A 3.2GHz P4 isn't going
to tolerate as much voltage drop as a 2.5GHz for example.

Please be kind enough to provide a link to this spec sheet.

http://www.intel.com

All I get there is the Intel home page. So where is the spec sheet?

Current Intel CPUs run hotter at full load. Read the spec sheets,
they're on the respective manufacturer's websites. At idle a P4
"can" run cooler, but the system must run at up to full-load also.

I see. Intel CPUs runn hotter at full load? So an Intel CPU with a
Vapochill attached will run hotter than an Athlon with no heat sink?

Since you aren't interested in meaningful comparisions, what's the
point?

So provide a "meaningful comparison".

Now, would you care to use a term that is more precise than "run
hotter"?

"Temperature"

So an Intel system running at -25 C is running hotter than an AMD system
at 50C? Again _temperature_ depends on the efficacy of the cooling
system. Other factors determine just how efficacious that system must
be in order to provide a specified temperature.

As I've already mentioned in the prevous post, an entire education
about undervolting CPUs is beyond the scope of a newsgroup post or
even a few links. I wasn't trying to be rude, just stating that the
volume of information is large and make take anyone some time to
digest. Both Intel and AMD provide specs on their processors... go
to their websites and read the spec sheets.

If you have read them then surely you have links to them. Now would
you care to provide a quotation from one of those specification sheets
that supports your argument?


I used to have an XP1600 running at 1.6V, o'c to 1.6 GHz. The
main reason you don't see many people undervolting, is that
typically the knowledge gained towards understanding the
frequency/voltage relationship, is gained as a result of a desire to
o'c the CPU as much as reasonably possible.

You're repeating yourself.

Sometimes it's needed, when someone doesn't seem to realize what's
already be stated... Which is greater- Merely having a theory based
on a spec sheet, narrowly interpreted, or actually having many working
demonstrations of it? I have many working systems undervolted, as do
others. They run fine so long as the voltage isn't decreased TOO
much, as verified by extensive testing and use over several years.

Sez you.

Yes, it works fine if you _know_ _what_ _you_ _are_ _doing_, which
means that you are an electrical engineer with IC design experience,
an intimate familiarity with the particular device under
consideration, and you know what constitute the worst cases
that need to be tested to confirm reliable operation.

No, the device is designed to perform a function. If this device
continues to perform the function properly, it is not necessary to
have designed it yourself.

And of course you are qualified to determine that the device performs
properly. So what methodology do you use to make this determination and
in what manner did you validate that methodology?

So provide a source for this information. Skepticism can also be due
to having seen more than one cocky young kid standing on his weenie
after his wild assertions proved to be somewhat in error.

You seem to be a little confused. I AM a source of this information,
as are others all over the 'net.

I see. And we should believe you why?

You have NO sources claiming it
can't be done, and have so far still argued against both this source
(and further extended, the original poster) without any evidence that
it can't be done.

I will repeat myself- You have evidence that it can be done, but no
evidence that it can't. Your argument is made in madness. If you
choose to disbelieve, that's your choice to make. I know it works.

No. I have your unsupported claim that it can be done. Your claim does
not constitute "evidence".

--
--
--John
Reply to jclarke at ae tee tee global dot net
(was jclarke at eye bee em dot net)