A computer components & hardware forum. HardwareBanter

If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below.

Go Back   Home » HardwareBanter forum » Processors » Overclocking
Site Map Home Register Authors List Search Today's Posts Mark Forums Read Web Partners

Which temp are right.



 
 
Thread Tools Display Modes
  #1  
Old January 5th 09, 03:44 PM posted to alt.comp.hardware.overclocking
Dumbo
external usenet poster
 
Posts: 39
Default Which temp are right.

Has anyone got a sure way of monitoring the CPU temperature accurately.
When idle in BIOS my CPU displays as 45C but `Core Temp` shows it as 50C and
real temp as 36C. SANDRA shows a third different temperature and a fourth
from Everest.

How do I know which is right and are the temp sensors on the motherboard
that accurate anyway.

Robin

  #2  
Old January 5th 09, 04:41 PM posted to alt.comp.hardware.overclocking
Ed Medlin
external usenet poster
 
Posts: 601
Default Which temp are right.


"Dumbo" wrote in message
...
Has anyone got a sure way of monitoring the CPU temperature accurately.
When idle in BIOS my CPU displays as 45C but `Core Temp` shows it as 50C
and real temp as 36C. SANDRA shows a third different temperature and a
fourth from Everest.

How do I know which is right and are the temp sensors on the motherboard
that accurate anyway.

Robin


Everest, Sandra and Core Temp show the same on my i7 920 processor. At least
within a couple of degC. Bios temp should be the higher one since the hlt
command is not active outside the OS (this could change if you have a lot of
processes running). Make sure you have the latest versions of your
monitoring software.

Ed

  #3  
Old January 7th 09, 02:30 AM posted to alt.comp.hardware.overclocking
Phil Weldon[_2_]
external usenet poster
 
Posts: 131
Default Which temp are right.

'Dumbo' wrote:
Has anyone got a sure way of monitoring the CPU temperature accurately.
When idle in BIOS my CPU displays as 45C but `Core Temp` shows it as 50C
and real temp as 36C. SANDRA shows a third different temperature and a
fourth from Everest.

How do I know which is right and are the temp sensors on the motherboard
that accurate anyway.

_____

What to consider:

A. actual temperature
1. sensor somewhere in a core (some areas tend to produce more
heat that others)
2. sensor somewhere in the CPU chip (but not in a core)
3. sensor on the CPU chip in contact with the CPU heat spreader
4. sensor not part of the CPU chip, but in contact with the CPU
chip or heat spreader (yes some CPUs have not had heat spreaders)

[all of the above locations have been used in the past, but only 1.
and 2. are used for current x86 CPUs]

B. type and precision of sensor

C. convertor (A to D) used in the motherboard system monitoring chip (or
Super I/O chip) to translate sensor voltage or current signal from the
sensor to decimal digits ( precision of the reference voltages )

D. offsets (fudge factors) applied to measured temperatures to get
something closer to actual temperatures

E. monitoring program used (some monitoring programs may not recognize
the monitoring chip or Super I/O chip [ there are dozens of possibilities ]
used by your motherboard, some monitoring programs may not know the type
sensors (thermistor, diode, transistor, active devices) used other than on
CPU chip sensor

[all of the above affect the ACCURACY of the temperature reported]

F. PRECISION errors

1. the number of bits the temperature A to D convertors use; an
8 bit A to D convertor can only represent one integer steps from - 127 to +
127, with a plus or minus one error ( i.e. 45, 46, and 47 might all
represent the same analog signal )

2. if the monitoring program is set to display in Fahrenheit
then, since only integers are displayed, the reported temperature has an
even larger PRECISION error.

The monitoring program supplied by the motherboard manufacturer SHOULD be,
but is not always, the best. After two years, EVGA still has not provided a
monitoring program that will report all the values reported in the BIOS for
my 680i SLI motherboard.) MotherBoard Monitor is the most universal
motherboard monitoring program ever issued, but it hasn't been updated for 5
or more years. If you haven't already purchased your motherboard make sure
you get one that comes with a good motherboard monitoring system that works
correctly with your specific motherboard model AND revision

Then the question is, what is the most USEFUL CPU temperature? My feeling
is that the temperature most useful is that at the outer surface of the CPU
heatsink. YOU have control of heat transport from this surface outwards;
the CPU manufacturer has control of heat transport from this surface inwards
( unless you remove the heat spreader ). But to get this temperature
requires modifying your heatsink and adding a temperature sensor in contact
with the heat spreader, but insulated from the heat sink. I can suggest
three ways of doing this, one of which I have tried.

1. Back in the days of the Pentium Celeron 333a I drilled a hole through
the center of the heatsink bottom surface and glued it in place with
Silicone which held the sensor in contact with the heat spreader, but
insulated from the heat sink. Since this was a temporary setup I used a
DVOM ( 4 1/2 digit ) to measure the thermistor resistance and manually
converted to temperature. I tested extensively with the following
variables; CPU voltage, CPU clock speed, and thermal compound. The
temperatures reported by this sensor was stable, and tracked the temperature
reported by MBM using the on-CPU-die thermal diode. This temporary system
could be made permanent and more convenient but replacing the thermistor
with an active device that produces a pulse width modulated signal that can
be digitized and presented to a serial port. The sensor device and
digitizer/RS-232 Serial interface with a monitoring program are available
off-the-shelf.

2. Fiber optic bundle with non-contact digital temperature readout, less
than $10 US for both.

3. Small diameter heatpipe with non-contact digital temperature readout,
less than $10 US for both if you get a fee sample heatpipe swizzle stick.

On the other hand, why worry about your CPU temperature? The on-CPU-die
thermal protection device will prevent heat damage (this is a fail-safe
device that can't be bypassed ). Assuming you want the best possible
reliable performance and not just bragging rights for the fastest possible
CPU without regards to stability, then
1. check for proper thermal compound application ( the brand and type
make almost no differences )
2. check for proper heatsink installation
3. check or proper case ventilation
4. check room ambient temperature ( each degree drop in room ambient
temperature will result in almost a degree drop in CPU temperature )
After the above checks, consider if a better cooling solution is worth the
expense and how that expense compares to the expense of spreader plate
temperature monitoring.

Phil Weldon



"Dumbo" wrote in message
...
Has anyone got a sure way of monitoring the CPU temperature accurately.
When idle in BIOS my CPU displays as 45C but `Core Temp` shows it as 50C
and real temp as 36C. SANDRA shows a third different temperature and a
fourth from Everest.

How do I know which is right and are the temp sensors on the motherboard
that accurate anyway.

Robin


  #4  
Old January 7th 09, 04:58 PM posted to alt.comp.hardware.overclocking
Ed Medlin
external usenet poster
 
Posts: 601
Default Which temp are right.


"Phil Weldon" wrote in message
news
'Dumbo' wrote:
Has anyone got a sure way of monitoring the CPU temperature accurately.
When idle in BIOS my CPU displays as 45C but `Core Temp` shows it as 50C
and real temp as 36C. SANDRA shows a third different temperature and a
fourth from Everest.

How do I know which is right and are the temp sensors on the motherboard
that accurate anyway.

_____

What to consider:

A. actual temperature
1. sensor somewhere in a core (some areas tend to produce more
heat that others)
2. sensor somewhere in the CPU chip (but not in a core)
3. sensor on the CPU chip in contact with the CPU heat spreader
4. sensor not part of the CPU chip, but in contact with the CPU
chip or heat spreader (yes some CPUs have not had heat spreaders)

[all of the above locations have been used in the past, but only 1.
and 2. are used for current x86 CPUs]

B. type and precision of sensor

C. convertor (A to D) used in the motherboard system monitoring chip
(or Super I/O chip) to translate sensor voltage or current signal from the
sensor to decimal digits ( precision of the reference voltages )

D. offsets (fudge factors) applied to measured temperatures to get
something closer to actual temperatures

E. monitoring program used (some monitoring programs may not recognize
the monitoring chip or Super I/O chip [ there are dozens of
possibilities ] used by your motherboard, some monitoring programs may not
know the type sensors (thermistor, diode, transistor, active devices) used
other than on CPU chip sensor

[all of the above affect the ACCURACY of the temperature reported]

F. PRECISION errors

1. the number of bits the temperature A to D convertors use;
an 8 bit A to D convertor can only represent one integer steps from - 127
to + 127, with a plus or minus one error ( i.e. 45, 46, and 47 might all
represent the same analog signal )

2. if the monitoring program is set to display in Fahrenheit
then, since only integers are displayed, the reported temperature has an
even larger PRECISION error.

The monitoring program supplied by the motherboard manufacturer SHOULD be,
but is not always, the best. After two years, EVGA still has not provided
a monitoring program that will report all the values reported in the BIOS
for my 680i SLI motherboard.) MotherBoard Monitor is the most universal
motherboard monitoring program ever issued, but it hasn't been updated for
5 or more years. If you haven't already purchased your motherboard make
sure you get one that comes with a good motherboard monitoring system that
works correctly with your specific motherboard model AND revision

Then the question is, what is the most USEFUL CPU temperature? My feeling
is that the temperature most useful is that at the outer surface of the
CPU heatsink. YOU have control of heat transport from this surface
outwards; the CPU manufacturer has control of heat transport from this
surface inwards ( unless you remove the heat spreader ). But to get this
temperature requires modifying your heatsink and adding a temperature
sensor in contact with the heat spreader, but insulated from the heat
sink. I can suggest three ways of doing this, one of which I have tried.

1. Back in the days of the Pentium Celeron 333a I drilled a hole through
the center of the heatsink bottom surface and glued it in place with
Silicone which held the sensor in contact with the heat spreader, but
insulated from the heat sink. Since this was a temporary setup I used a
DVOM ( 4 1/2 digit ) to measure the thermistor resistance and manually
converted to temperature. I tested extensively with the following
variables; CPU voltage, CPU clock speed, and thermal compound. The
temperatures reported by this sensor was stable, and tracked the
temperature reported by MBM using the on-CPU-die thermal diode. This
temporary system could be made permanent and more convenient but replacing
the thermistor with an active device that produces a pulse width modulated
signal that can be digitized and presented to a serial port. The sensor
device and digitizer/RS-232 Serial interface with a monitoring program
are available off-the-shelf.

2. Fiber optic bundle with non-contact digital temperature readout, less
than $10 US for both.

3. Small diameter heatpipe with non-contact digital temperature readout,
less than $10 US for both if you get a fee sample heatpipe swizzle stick.

On the other hand, why worry about your CPU temperature? The on-CPU-die
thermal protection device will prevent heat damage (this is a fail-safe
device that can't be bypassed ). Assuming you want the best possible
reliable performance and not just bragging rights for the fastest possible
CPU without regards to stability, then
1. check for proper thermal compound application ( the brand and type
make almost no differences )
2. check for proper heatsink installation
3. check or proper case ventilation
4. check room ambient temperature ( each degree drop in room ambient
temperature will result in almost a degree drop in CPU temperature )
After the above checks, consider if a better cooling solution is worth the
expense and how that expense compares to the expense of spreader plate
temperature monitoring.

Phil Weldon

On the new i7 (Nehalem) processors, I am not sure about a few things in
regard to temperatures. The tjmax is 100C. That is about where the processor
begins to throttle down. I backed down my a very stable 3.7Ghz OC just
because of my older views on what is too hot.....:-). I don't know if the
fact that the memory controller is on chip make these run hotter, but there
is a significant rise in the temperatures compared to what I have become
used to over the years. If these temps are actually the max temps inside
each core, then I might not be quite as concerned since that would be about
what I would expect since previous sensors usually just read the surface
temps of each core. I have ran the i7 920 at over 85C for hours on end and
it stays perfectly stable, but I don't........:-)


Ed

OT--- BTW, how is your astronomy project going Phil?


  #5  
Old January 7th 09, 09:40 PM posted to alt.comp.hardware.overclocking
Phil Weldon[_2_]
external usenet poster
 
Posts: 131
Default Which temp are right.

'Ed Medlin' wrote:
On the new i7 (Nehalem) processors, I am not sure about a few things in
regard to temperatures. The tjmax is 100C. That is about where the
processor begins to throttle down. I backed down my a very stable 3.7Ghz
OC just because of my older views on what is too hot.....:-). I don't know
if the fact that the memory controller is on chip make these run hotter,
but there is a significant rise in the temperatures compared to what I
have become used to over the years. If these temps are actually the max
temps inside each core, then I might not be quite as concerned since that
would be about what I would expect since previous sensors usually just
read the surface temps of each core. I have ran the i7 920 at over 85C for
hours on end and it stays perfectly stable, but I don't........:-)

_____

Well, the astronomy project is progressing but I won't receive the repaired
telescope until sometimes this week ( I hope.) If you remember, UPS dropped
the package ( and the UPS store did not package it adequately, $25 worth of
packing material would have likely averted the damage which included
shattered mirrors and corrector lens, broken aluminum castings and damaged
GPS receiver and circuit boards [ the mount includes 5 microprocessors ] )
I returned the telescope to Meade; the repair charge was $2150 US, including
return shipping. On the upside, for that I get essentially a new telescope,
and UPS will cover the repair costs.

The only computer related purchases I need to make are

an RS232 to USB adapter ( to enable my notebook to use a signal from a
USB output guide camera to correct tracking errors of the telescope mount
via RS232 [ my notebook doesn't have RS232 inputs ] )

a red filter for my notebook screen ( to preserve night vision )

eventually, perhaps, a better image processing application.

Despite the wait, I am still pleased with my eBay purchase; the seller has
kept in touch throughout the insurance settlement and repair process. The
new cost of the items for which I paid $3500 US was about $8800, and only
one of the items has been improved since the original purchase.

One interesting temperature related item; the focal length of the telescope
is so long ( 3048 mm ) that the aluminum tube ( ~ 22 " between the front and
rear elements ) in which the mirrors and corrector lens are mounted can
change length enough to throw the focus out with a temperature change of a
few degrees. This can be a problem with long exposures ( which can be an
hour or more ). One of the accessories included is a temperature
compensating focuser that uses a sensor to measure the aluminum tube
temperature and a stepping motor to adjust the film plane by 0.00008" per
step. This makes me wonder about warping of the aluminum or copper castings
in CPU heatsinks. Is a 50 C ( or more ) temperature change enough to affect
the flatness of the heatsink / CPU heat spreader interface? Would it be
helpful to lap the heatsink to flatness at the CPU heat spreader
temperature? I'd guess you could check this possible problem by just
heating a bare heatsink to 50 C and measuring any departure from flatness.
Perhaps only a few hundredths of a millimeter could make a difference.

Phil Weldon




"Ed Medlin" wrote in message
...

"Phil Weldon" wrote in message
news
'Dumbo' wrote:
Has anyone got a sure way of monitoring the CPU temperature accurately.
When idle in BIOS my CPU displays as 45C but `Core Temp` shows it as 50C
and real temp as 36C. SANDRA shows a third different temperature and a
fourth from Everest.

How do I know which is right and are the temp sensors on the motherboard
that accurate anyway.

_____

What to consider:

A. actual temperature
1. sensor somewhere in a core (some areas tend to produce more
heat that others)
2. sensor somewhere in the CPU chip (but not in a core)
3. sensor on the CPU chip in contact with the CPU heat
spreader
4. sensor not part of the CPU chip, but in contact with the
CPU chip or heat spreader (yes some CPUs have not had heat spreaders)

[all of the above locations have been used in the past, but only
1. and 2. are used for current x86 CPUs]

B. type and precision of sensor

C. convertor (A to D) used in the motherboard system monitoring chip
(or Super I/O chip) to translate sensor voltage or current signal from
the sensor to decimal digits ( precision of the reference voltages )

D. offsets (fudge factors) applied to measured temperatures to get
something closer to actual temperatures

E. monitoring program used (some monitoring programs may not
recognize the monitoring chip or Super I/O chip [ there are dozens of
possibilities ] used by your motherboard, some monitoring programs may
not know the type sensors (thermistor, diode, transistor, active devices)
used other than on CPU chip sensor

[all of the above affect the ACCURACY of the temperature reported]

F. PRECISION errors

1. the number of bits the temperature A to D convertors use;
an 8 bit A to D convertor can only represent one integer steps from - 127
to + 127, with a plus or minus one error ( i.e. 45, 46, and 47 might all
represent the same analog signal )

2. if the monitoring program is set to display in Fahrenheit
then, since only integers are displayed, the reported temperature has an
even larger PRECISION error.

The monitoring program supplied by the motherboard manufacturer SHOULD
be, but is not always, the best. After two years, EVGA still has not
provided a monitoring program that will report all the values reported in
the BIOS for my 680i SLI motherboard.) MotherBoard Monitor is the most
universal motherboard monitoring program ever issued, but it hasn't been
updated for 5 or more years. If you haven't already purchased your
motherboard make sure you get one that comes with a good motherboard
monitoring system that works correctly with your specific motherboard
model AND revision

Then the question is, what is the most USEFUL CPU temperature? My
feeling is that the temperature most useful is that at the outer surface
of the CPU heatsink. YOU have control of heat transport from this
surface outwards; the CPU manufacturer has control of heat transport from
this surface inwards ( unless you remove the heat spreader ). But to get
this temperature requires modifying your heatsink and adding a
temperature sensor in contact with the heat spreader, but insulated from
the heat sink. I can suggest three ways of doing this, one of which I
have tried.

1. Back in the days of the Pentium Celeron 333a I drilled a hole through
the center of the heatsink bottom surface and glued it in place with
Silicone which held the sensor in contact with the heat spreader, but
insulated from the heat sink. Since this was a temporary setup I used a
DVOM ( 4 1/2 digit ) to measure the thermistor resistance and manually
converted to temperature. I tested extensively with the following
variables; CPU voltage, CPU clock speed, and thermal compound. The
temperatures reported by this sensor was stable, and tracked the
temperature reported by MBM using the on-CPU-die thermal diode. This
temporary system could be made permanent and more convenient but
replacing the thermistor with an active device that produces a pulse
width modulated signal that can be digitized and presented to a serial
port. The sensor device and digitizer/RS-232 Serial interface with a
monitoring program are available off-the-shelf.

2. Fiber optic bundle with non-contact digital temperature readout, less
than $10 US for both.

3. Small diameter heatpipe with non-contact digital temperature readout,
less than $10 US for both if you get a fee sample heatpipe swizzle stick.

On the other hand, why worry about your CPU temperature? The on-CPU-die
thermal protection device will prevent heat damage (this is a fail-safe
device that can't be bypassed ). Assuming you want the best possible
reliable performance and not just bragging rights for the fastest
possible CPU without regards to stability, then
1. check for proper thermal compound application ( the brand and type
make almost no differences )
2. check for proper heatsink installation
3. check or proper case ventilation
4. check room ambient temperature ( each degree drop in room ambient
temperature will result in almost a degree drop in CPU temperature )
After the above checks, consider if a better cooling solution is worth
the expense and how that expense compares to the expense of spreader
plate temperature monitoring.

Phil Weldon

On the new i7 (Nehalem) processors, I am not sure about a few things in
regard to temperatures. The tjmax is 100C. That is about where the
processor begins to throttle down. I backed down my a very stable 3.7Ghz
OC just because of my older views on what is too hot.....:-). I don't know
if the fact that the memory controller is on chip make these run hotter,
but there is a significant rise in the temperatures compared to what I
have become used to over the years. If these temps are actually the max
temps inside each core, then I might not be quite as concerned since that
would be about what I would expect since previous sensors usually just
read the surface temps of each core. I have ran the i7 920 at over 85C for
hours on end and it stays perfectly stable, but I don't........:-)


Ed

OT--- BTW, how is your astronomy project going Phil?


  #6  
Old January 8th 09, 06:07 AM posted to alt.comp.hardware.overclocking
Gorby[_2_]
external usenet poster
 
Posts: 40
Default Which temp are right.

Ed Medlin wrote:
"Phil Weldon" wrote in message
news
'Dumbo' wrote:
Has anyone got a sure way of monitoring the CPU temperature accurately.
When idle in BIOS my CPU displays as 45C but `Core Temp` shows it as 50C
and real temp as 36C. SANDRA shows a third different temperature and a
fourth from Everest.

How do I know which is right and are the temp sensors on the motherboard
that accurate anyway.

_____

What to consider:

A. actual temperature
1. sensor somewhere in a core (some areas tend to produce more
heat that others)
2. sensor somewhere in the CPU chip (but not in a core)
3. sensor on the CPU chip in contact with the CPU heat spreader
4. sensor not part of the CPU chip, but in contact with the CPU
chip or heat spreader (yes some CPUs have not had heat spreaders)

[all of the above locations have been used in the past, but only 1.
and 2. are used for current x86 CPUs]

B. type and precision of sensor

C. convertor (A to D) used in the motherboard system monitoring chip
(or Super I/O chip) to translate sensor voltage or current signal from the
sensor to decimal digits ( precision of the reference voltages )

D. offsets (fudge factors) applied to measured temperatures to get
something closer to actual temperatures

E. monitoring program used (some monitoring programs may not recognize
the monitoring chip or Super I/O chip [ there are dozens of
possibilities ] used by your motherboard, some monitoring programs may not
know the type sensors (thermistor, diode, transistor, active devices) used
other than on CPU chip sensor

[all of the above affect the ACCURACY of the temperature reported]

F. PRECISION errors

1. the number of bits the temperature A to D convertors use;
an 8 bit A to D convertor can only represent one integer steps from - 127
to + 127, with a plus or minus one error ( i.e. 45, 46, and 47 might all
represent the same analog signal )

2. if the monitoring program is set to display in Fahrenheit
then, since only integers are displayed, the reported temperature has an
even larger PRECISION error.

The monitoring program supplied by the motherboard manufacturer SHOULD be,
but is not always, the best. After two years, EVGA still has not provided
a monitoring program that will report all the values reported in the BIOS
for my 680i SLI motherboard.) MotherBoard Monitor is the most universal
motherboard monitoring program ever issued, but it hasn't been updated for
5 or more years. If you haven't already purchased your motherboard make
sure you get one that comes with a good motherboard monitoring system that
works correctly with your specific motherboard model AND revision

Then the question is, what is the most USEFUL CPU temperature? My feeling
is that the temperature most useful is that at the outer surface of the
CPU heatsink. YOU have control of heat transport from this surface
outwards; the CPU manufacturer has control of heat transport from this
surface inwards ( unless you remove the heat spreader ). But to get this
temperature requires modifying your heatsink and adding a temperature
sensor in contact with the heat spreader, but insulated from the heat
sink. I can suggest three ways of doing this, one of which I have tried.

1. Back in the days of the Pentium Celeron 333a I drilled a hole through
the center of the heatsink bottom surface and glued it in place with
Silicone which held the sensor in contact with the heat spreader, but
insulated from the heat sink. Since this was a temporary setup I used a
DVOM ( 4 1/2 digit ) to measure the thermistor resistance and manually
converted to temperature. I tested extensively with the following
variables; CPU voltage, CPU clock speed, and thermal compound. The
temperatures reported by this sensor was stable, and tracked the
temperature reported by MBM using the on-CPU-die thermal diode. This
temporary system could be made permanent and more convenient but replacing
the thermistor with an active device that produces a pulse width modulated
signal that can be digitized and presented to a serial port. The sensor
device and digitizer/RS-232 Serial interface with a monitoring program
are available off-the-shelf.

2. Fiber optic bundle with non-contact digital temperature readout, less
than $10 US for both.

3. Small diameter heatpipe with non-contact digital temperature readout,
less than $10 US for both if you get a fee sample heatpipe swizzle stick.

On the other hand, why worry about your CPU temperature? The on-CPU-die
thermal protection device will prevent heat damage (this is a fail-safe
device that can't be bypassed ). Assuming you want the best possible
reliable performance and not just bragging rights for the fastest possible
CPU without regards to stability, then
1. check for proper thermal compound application ( the brand and type
make almost no differences )
2. check for proper heatsink installation
3. check or proper case ventilation
4. check room ambient temperature ( each degree drop in room ambient
temperature will result in almost a degree drop in CPU temperature )
After the above checks, consider if a better cooling solution is worth the
expense and how that expense compares to the expense of spreader plate
temperature monitoring.

Phil Weldon

On the new i7 (Nehalem) processors, I am not sure about a few things in
regard to temperatures. The tjmax is 100C. That is about where the processor
begins to throttle down. I backed down my a very stable 3.7Ghz OC just
because of my older views on what is too hot.....:-). I don't know if the
fact that the memory controller is on chip make these run hotter, but there
is a significant rise in the temperatures compared to what I have become
used to over the years. If these temps are actually the max temps inside
each core, then I might not be quite as concerned since that would be about
what I would expect since previous sensors usually just read the surface
temps of each core. I have ran the i7 920 at over 85C for hours on end and
it stays perfectly stable, but I don't........:-)


Ed

OT--- BTW, how is your astronomy project going Phil?


Might this be part of the industry move allowing for hotter computing
rooms? There is a move to do away with costly air conditioners in server
rooms. Going to use more ambient air; collect the cooler night air, etc.
But also getting the electronics to run at higher temperatures for long
periods of time, without harm.
  #7  
Old January 8th 09, 04:22 PM posted to alt.comp.hardware.overclocking
Ed Medlin
external usenet poster
 
Posts: 601
Default Which temp are right.

On the new i7 (Nehalem) processors, I am not sure about a few things in
regard to temperatures. The tjmax is 100C. That is about where the
processor begins to throttle down. I backed down my a very stable 3.7Ghz
OC just because of my older views on what is too hot.....:-). I don't
know if the fact that the memory controller is on chip make these run
hotter, but there is a significant rise in the temperatures compared to
what I have become used to over the years. If these temps are actually
the max temps inside each core, then I might not be quite as concerned
since that would be about what I would expect since previous sensors
usually just read the surface temps of each core. I have ran the i7 920
at over 85C for hours on end and it stays perfectly stable, but I
don't........:-)


Ed

OT--- BTW, how is your astronomy project going Phil?

Might this be part of the industry move allowing for hotter computing
rooms? There is a move to do away with costly air conditioners in server
rooms. Going to use more ambient air; collect the cooler night air, etc.
But also getting the electronics to run at higher temperatures for long
periods of time, without harm.


I really think it is mostly due to the complete redesigning of the i7 from
the previous Intel processors. They may be measuring the temps at a
different point too since I have seen some overclocking reviews where even
with liquid cooling temps are only 3-4 degC lower and sometimes not even
that. If temps were measured at the hottest point in the core of even a C2D
I would bet it would be much higher than what we see with peripheral core
readings. The only times that those extremely high temperatures are seen are
with "burn-in" apps that max out all the cores to the extreme. Even during
intense gaming the temperatures are nowhere near 85C and stay in the 60-65C
range which is acceptable for any processor. That is 35-40C under the tjmax
of 100C of the processor so that should be just fine.


Ed


  #8  
Old January 8th 09, 09:31 PM posted to alt.comp.hardware.overclocking
Dumbo
external usenet poster
 
Posts: 39
Default Which temp are right.

At POST my BIOS reported 39C for CPU and 34C for the motherboard yet booting
into Vista even on idle the temps range from 46C-49C with Core Temp and
31C-34C with Real Temp but I just want to know which one is most accurate so
I can do a bit of overclocking without the worry of burning anything out.
Real Temp gives a table with a `Distance to TJ Max on it but 100C seems to
boiling temp and only asking for bother.
With a bit of gaming and returning to Vista I get a temp of around 69C. Is
this a safe enough temp or do I need to cool things a bit more (lower speed
or increased airflow).
I am trying to figure out the best most stable overclock but this seems
inherently hot! I try to determine what is causing the BSOD or crash. Does
the memory, CPU, NB or SB need more voltage and how hot is it going to get.
I have a reasonably stable 3GHz Q6600 B3 but still get crashes but recently
I have found the same crashes at default speeds of 2.4GHz so I assume its a
bit of bad data on the drive or graphics card problem even though I haven't
tried the card until I get 100% stable CPU.
Thanks for your input.

Robin

  #9  
Old January 8th 09, 11:44 PM posted to alt.comp.hardware.overclocking
[email protected]
external usenet poster
 
Posts: 14
Default Which temp are right.

On Thu, 8 Jan 2009 20:31:38 -0000, "Dumbo"
wrote:

At POST my BIOS reported 39C for CPU and 34C for the motherboard yet booting
into Vista even on idle the temps range from 46C-49C with Core Temp and
31C-34C with Real Temp but I just want to know which one is most accurate so
I can do a bit of overclocking without the worry of burning anything out.
Real Temp gives a table with a `Distance to TJ Max on it but 100C seems to
boiling temp and only asking for bother.
With a bit of gaming and returning to Vista I get a temp of around 69C. Is
this a safe enough temp or do I need to cool things a bit more (lower speed
or increased airflow).
I am trying to figure out the best most stable overclock but this seems
inherently hot! I try to determine what is causing the BSOD or crash. Does
the memory, CPU, NB or SB need more voltage and how hot is it going to get.
I have a reasonably stable 3GHz Q6600 B3 but still get crashes but recently
I have found the same crashes at default speeds of 2.4GHz so I assume its a
bit of bad data on the drive or graphics card problem even though I haven't
tried the card until I get 100% stable CPU.
Thanks for your input.

Robin


I get identical temps from Core Temp and Real Temp. I have noticed
that most of the people showing screen shots of values on the
overclocking forums seem to use Real Temp. Similarly, most of them use
Memtest86+2.11 rather than Memtest86 3.?. They also use Everest, but
that's to show other parameters rather than the temperatures. I don't
know that that's a recommendation, but in any event, start by checking
your memory for at least four passes or 8 hours. You may need to
detune your CPU back to 2.4 GHz until you are sure your memory is
stable.

Jim
  #10  
Old January 9th 09, 02:21 AM posted to alt.comp.hardware.overclocking
Phil Weldon[_2_]
external usenet poster
 
Posts: 131
Default Which temp are right.

'Dumbo' wrote, in part:
... I just want to know which one is most accurate so I can do a bit of
overclocking without the worry of burning anything out.

_____

It isn't possible to "burn anything out" by overclocking. As has been
pointed out, the CPU will lock up and cool off long before any damage is
done. Moreover, the 'fail-safe' on-CPU-die thermal diode is always
functional. Raising the CPU core voltage too high can cause immediate
destruction of the CPU, but damage from overheating just isn't possible.

Phil Weldon


"Dumbo" wrote in message
...
At POST my BIOS reported 39C for CPU and 34C for the motherboard yet
booting into Vista even on idle the temps range from 46C-49C with Core
Temp and 31C-34C with Real Temp but I just want to know which one is most
accurate so I can do a bit of overclocking without the worry of burning
anything out. Real Temp gives a table with a `Distance to TJ Max on it but
100C seems to boiling temp and only asking for bother.
With a bit of gaming and returning to Vista I get a temp of around 69C.
Is this a safe enough temp or do I need to cool things a bit more (lower
speed or increased airflow).
I am trying to figure out the best most stable overclock but this seems
inherently hot! I try to determine what is causing the BSOD or crash.
Does the memory, CPU, NB or SB need more voltage and how hot is it going
to get.
I have a reasonably stable 3GHz Q6600 B3 but still get crashes but
recently I have found the same crashes at default speeds of 2.4GHz so I
assume its a bit of bad data on the drive or graphics card problem even
though I haven't tried the card until I get 100% stable CPU.
Thanks for your input.

Robin


 




Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Mobo "case" temp sensor vs actual case temp Lee M. Homebuilt PC's 5 May 10th 08 03:36 PM
Max temp & normal running temp (Dual Core)? Graham Naylor Intel 9 April 12th 07 05:21 AM
Case Temp vs Die Temp Rob Stow General 0 August 28th 04 03:10 AM
Mobo temp Vs CPU temp rays Homebuilt PC's 1 May 6th 04 11:10 PM
CPU Temp N/A on P2B ? Stephan Grossklass Asus Motherboards 1 February 22nd 04 06:34 PM


All times are GMT +1. The time now is 10:11 AM.


Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright ©2004-2024 HardwareBanter.
The comments are property of their posters.