Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)
"George Macdonald" <fammacd=!SPAM^nothanks@tellurian.com> wrote in message news:a73fn0lfma9j4nfkvpkpmklvknmq4b4u1s@4ax.com...
> On Wed, 20 Oct 2004 19:56:33 GMT, "AJ" <ng@newsgroups.net> wrote:
>
>>
>>"George Macdonald" <fammacd=!SPAM^nothanks@tellurian.com> wrote in message
>>news:2ri8n0t5ol1naldood6ev0ddlqovdo9bk4@4ax.com...
>>> On Mon, 18 Oct 2004 19:30:03 GMT, "AJ" <ng@newsgroups.net> wrote:
>
>>>>The 3 motherboard temp sensors on my PC as I type this are 28, 26, 26 C.
>>>>(I plan on slowing down or replacing my case fans though. They are 92 mm
>>>>Zalmans that have the resistor inline causing them to turn at 1600 rpm, but
>>>>I can still hear them so I'm going to try to find even slower fans, perhaps
>>>>PWM ones and a controller). That's why I like Northwoods over Prescotts:
>>>>I can get much closer to silent computing.
>>>
>>> I think you're placing far too much faith in temp readings from your mbrd's
>>> BIOS.
>>
>>Why? I'm not doing a scientific comparison for a magazine.
>
> You're quoting temperature numbers as though they are actually accurate
> temperature readings - they're *not*.
No I wasn't. If you need more info on why mine runs at 28 C, my system specs
for example, fine. Of course there's a lot of variables. But the numbers pull off
of say Intel Active Monitor and the numbers someone else pulls off of Intel
Active Monitor with the same components, are going to be relatively comparable.
If someone says their P4C runs at 38 C at idle and I say mine runs at 28 C, then
it begs the question what is different between the systems. I don't reject those
relative numbers just because they weren't measured with a highly accurate
thermocouple and data acquisition equipment in a controlled environment. It's
not necessary and the info IS useful. That starts the dialog and further info is
gathered as need (what's the ambient temp.. etc).
And just FYI, my system is running at 28 C right now (as taken from Active
Monitor). If you wanted further qualification of where that number came from
originally, all you had to do is ask me. If you don't ask, then I'll assume
you were fine with the limited info.
>
>>> The mbrd mfrs can calibrate them to read anything you want
>>
>>They seem to be consistent across different boards.
>
> They're not necessarily so - they can vary across different BIOS versions.
So what? If you need that info, you'll get it upon further investigation and ask for
it. Some things can be assumed or derived you know: like I just built MY system,
so it's probably got a fairly recent BIOS. Every unlikely detail is not likely to be
encountered all of the time. No one was trying to make an absolutely irrefutable
statement. Take it with a grain of salt and not so absolutely literally. If you need
more info, ask for it.
>
>>>- there
>>> have been several cases where they have responded to user concerns of high
>>> reported CPU temps by lowering them, in a later BIOS, to the point they
>>> read lower than the mbrd "system temp".
>>
>>As long as they are measured the same way, it's fine. The readings are
>>hardly useless. Quite useful actually.
>
> I did *not* say they are useless... I hope you're not suggesting such!
I thought you were trying make such or similar a point. Mine was that most
times you don't need to be so diligent/scientific to arrive at conclusion from
info/data even if it is "fuzzy" info/data.
>
>>> The fact is that such readings are not useful as an absolute measure of
>>> temperature - the only use they really have is for detecting changes in
>>> general system/CPU thermal behavior.
>>
>>I think most people report those readings here though. Most people don't
>>have scientifically thermocoupled systems! The relative readings were
>>relevant for this thread (pretty much measured the same way and are
>>probably fairly consistent from board to board. I've built a few (5) of them
>>and they all turn out the same (except for the +10C Prescott).
>
> As I said, they are useful as relative changes for any given mbrd+CPU... to
> detect, e.g., loss of cooling efficiency or a failing fan... for that
> system.
Well as long as you mean for any same PN and not just any given standalone
system. If you and me buy the same model Dell PC configured the same, the
temps are gonna read pretty much the same in the same conditions. Unless
something is different. It's not like there's a 5 or 10 degree reporting difference.
When a 10 degree idle difference between systems has been noted, then one
can investigate further if there is cause for concern or if they wanna do anything
about it. It goes without saying that there are a number of variables, but I know
relative accuracy is not one of them (based upon the ones I just built. And I
know I just didn't get lucky with the consistency of parts over the last few months
in that they all operate pretty much the same). (The Dell example is of course
one of the special cases because they use their own proprietary cooling. They
run hot to keep fan noise down in at least some of their systems. They accept
that trade off.)
Even outside of specific PNs and brands, one could use one's own judgement
with a little bit of fact gathering and determine with good probability of correct
analysis. I've seen many sources give the 30-ish C number for the 2.X GHz P4Cs.
And I have no reason to doubt it. Similarly, I can count on a Prescott being almost
10 C hotter (40-ish). Just loosely given numbers meant to be taken as useful where
appropriate.
AJ
"George Macdonald" <fammacd=!SPAM^nothanks@tellurian.com> wrote in message news:a73fn0lfma9j4nfkvpkpmklvknmq4b4u1s@4ax.com...
> On Wed, 20 Oct 2004 19:56:33 GMT, "AJ" <ng@newsgroups.net> wrote:
>
>>
>>"George Macdonald" <fammacd=!SPAM^nothanks@tellurian.com> wrote in message
>>news:2ri8n0t5ol1naldood6ev0ddlqovdo9bk4@4ax.com...
>>> On Mon, 18 Oct 2004 19:30:03 GMT, "AJ" <ng@newsgroups.net> wrote:
>
>>>>The 3 motherboard temp sensors on my PC as I type this are 28, 26, 26 C.
>>>>(I plan on slowing down or replacing my case fans though. They are 92 mm
>>>>Zalmans that have the resistor inline causing them to turn at 1600 rpm, but
>>>>I can still hear them so I'm going to try to find even slower fans, perhaps
>>>>PWM ones and a controller). That's why I like Northwoods over Prescotts:
>>>>I can get much closer to silent computing.
>>>
>>> I think you're placing far too much faith in temp readings from your mbrd's
>>> BIOS.
>>
>>Why? I'm not doing a scientific comparison for a magazine.
>
> You're quoting temperature numbers as though they are actually accurate
> temperature readings - they're *not*.
No I wasn't. If you need more info on why mine runs at 28 C, my system specs
for example, fine. Of course there's a lot of variables. But the numbers pull off
of say Intel Active Monitor and the numbers someone else pulls off of Intel
Active Monitor with the same components, are going to be relatively comparable.
If someone says their P4C runs at 38 C at idle and I say mine runs at 28 C, then
it begs the question what is different between the systems. I don't reject those
relative numbers just because they weren't measured with a highly accurate
thermocouple and data acquisition equipment in a controlled environment. It's
not necessary and the info IS useful. That starts the dialog and further info is
gathered as need (what's the ambient temp.. etc).
And just FYI, my system is running at 28 C right now (as taken from Active
Monitor). If you wanted further qualification of where that number came from
originally, all you had to do is ask me. If you don't ask, then I'll assume
you were fine with the limited info.
>
>>> The mbrd mfrs can calibrate them to read anything you want
>>
>>They seem to be consistent across different boards.
>
> They're not necessarily so - they can vary across different BIOS versions.
So what? If you need that info, you'll get it upon further investigation and ask for
it. Some things can be assumed or derived you know: like I just built MY system,
so it's probably got a fairly recent BIOS. Every unlikely detail is not likely to be
encountered all of the time. No one was trying to make an absolutely irrefutable
statement. Take it with a grain of salt and not so absolutely literally. If you need
more info, ask for it.
>
>>>- there
>>> have been several cases where they have responded to user concerns of high
>>> reported CPU temps by lowering them, in a later BIOS, to the point they
>>> read lower than the mbrd "system temp".
>>
>>As long as they are measured the same way, it's fine. The readings are
>>hardly useless. Quite useful actually.
>
> I did *not* say they are useless... I hope you're not suggesting such!
I thought you were trying make such or similar a point. Mine was that most
times you don't need to be so diligent/scientific to arrive at conclusion from
info/data even if it is "fuzzy" info/data.
>
>>> The fact is that such readings are not useful as an absolute measure of
>>> temperature - the only use they really have is for detecting changes in
>>> general system/CPU thermal behavior.
>>
>>I think most people report those readings here though. Most people don't
>>have scientifically thermocoupled systems! The relative readings were
>>relevant for this thread (pretty much measured the same way and are
>>probably fairly consistent from board to board. I've built a few (5) of them
>>and they all turn out the same (except for the +10C Prescott).
>
> As I said, they are useful as relative changes for any given mbrd+CPU... to
> detect, e.g., loss of cooling efficiency or a failing fan... for that
> system.
Well as long as you mean for any same PN and not just any given standalone
system. If you and me buy the same model Dell PC configured the same, the
temps are gonna read pretty much the same in the same conditions. Unless
something is different. It's not like there's a 5 or 10 degree reporting difference.
When a 10 degree idle difference between systems has been noted, then one
can investigate further if there is cause for concern or if they wanna do anything
about it. It goes without saying that there are a number of variables, but I know
relative accuracy is not one of them (based upon the ones I just built. And I
know I just didn't get lucky with the consistency of parts over the last few months
in that they all operate pretty much the same). (The Dell example is of course
one of the special cases because they use their own proprietary cooling. They
run hot to keep fan noise down in at least some of their systems. They accept
that trade off.)
Even outside of specific PNs and brands, one could use one's own judgement
with a little bit of fact gathering and determine with good probability of correct
analysis. I've seen many sources give the 30-ish C number for the 2.X GHz P4Cs.
And I have no reason to doubt it. Similarly, I can count on a Prescott being almost
10 C hotter (40-ish). Just loosely given numbers meant to be taken as useful where
appropriate.
AJ
Facebook
Twitter
Instagram