HardwareBanter

HardwareBanter (http://www.hardwarebanter.com/index.php)
-   Overclocking AMD Processors (http://www.hardwarebanter.com/forumdisplay.php?f=12)
-   -   Plea for sensible temperature reporting (http://www.hardwarebanter.com/showthread.php?t=12511)

Michael Brown October 18th 03 01:20 PM

Plea for sensible temperature reporting
 
OK, maybe I'm sounding a little like Wes on this, but I wish people would
stop reporting socket temps for comparisons between systems. It's useless.
If you *must* post socket temps, only do it with regard to changes you make
to your system.

Saying "changing from HSF A to HSF B leads to a drop in my socket temps, so
HSF B is better" is fine (assuming both were put on correctly, etc) and
actually has some meaningful information.

Saying "system A gets socket temperature B with HSF C, and system X gets
socket temperature Y with HSF Z (where Y B) so HSF Z is better" is totaly
devoid of any information at all, so don't even bother mentioning it.

Also, if mention any temperatures, say what they are! If they are socket,
say socket. If they are core, say core. Do *not* say "CPU" temperatures. Say
the case temp as well, and possibly the ambient temp if you're in some place
where room temperature is not somewhere around 20 deg C (eg: Antartica,
Sahara desert[*]).

I can't force you to do this (unfortunately :) ) but for the sake of
increasing the signal-to-noise ratio of this group, please be sensible when
talking about temperatures.
[*] Yes, I know the Sahara gets freakin' freezing at night, that's partially
my point :)

--
Michael Brown
www.emboss.co.nz : OOS/RSI software and more :)
Add michael@ to emboss.co.nz - My inbox is always open



\(\) |V| 3 G A October 18th 03 11:43 PM

i fully agree with you mike on this.

however some/most mobo manufactures dont use the onchip temp sensor and use
a socket based temp sensor. so untill ALL manufactures use the chip's sensor
then i think your living in a dream world.

tim
"Michael Brown" wrote in message
...
OK, maybe I'm sounding a little like Wes on this, but I wish people would
stop reporting socket temps for comparisons between systems. It's useless.
If you *must* post socket temps, only do it with regard to changes you

make
to your system.

Saying "changing from HSF A to HSF B leads to a drop in my socket temps,

so
HSF B is better" is fine (assuming both were put on correctly, etc) and
actually has some meaningful information.

Saying "system A gets socket temperature B with HSF C, and system X gets
socket temperature Y with HSF Z (where Y B) so HSF Z is better" is

totaly
devoid of any information at all, so don't even bother mentioning it.

Also, if mention any temperatures, say what they are! If they are socket,
say socket. If they are core, say core. Do *not* say "CPU" temperatures.

Say
the case temp as well, and possibly the ambient temp if you're in some

place
where room temperature is not somewhere around 20 deg C (eg: Antartica,
Sahara desert[*]).

I can't force you to do this (unfortunately :) ) but for the sake of
increasing the signal-to-noise ratio of this group, please be sensible

when
talking about temperatures.

[*] Yes, I know the Sahara gets freakin' freezing at night, that's

partially
my point :)

--
Michael Brown
www.emboss.co.nz : OOS/RSI software and more :)
Add michael@ to emboss.co.nz - My inbox is always open





Gary W. Swearingen October 19th 03 07:48 AM

"Michael Brown" writes:

the case temp as well, and possibly the ambient temp if you're in some place


And don't just say "ambient temp", because that can mean the air
surrounding the case or the air surrounding the CPU/HSF (or the air
surrounding anything else, strickly speaking). (I base this statement
on the definition of "ambient" and having recently read some Intel doc
that used "ambient" to refer to the air surrounding the CPU/HSF).

Gary W. Swearingen October 19th 03 07:48 AM

"Michael Brown" writes:

the case temp as well, and possibly the ambient temp if you're in some place


And don't just say "ambient temp", because that can mean the air
surrounding the case or the air surrounding the CPU/HSF (or the air
surrounding anything else, strickly speaking). (I base this statement
on the definition of "ambient" and having recently read some Intel doc
that used "ambient" to refer to the air surrounding the CPU/HSF).

Michael Brown October 19th 03 08:00 AM

Gary W. Swearingen wrote:
"Michael Brown" writes:

the case temp as well, and possibly the ambient temp if you're in
some place


And don't just say "ambient temp", because that can mean the air
surrounding the case or the air surrounding the CPU/HSF (or the air
surrounding anything else, strickly speaking). (I base this statement
on the definition of "ambient" and having recently read some Intel doc
that used "ambient" to refer to the air surrounding the CPU/HSF).


(Presuming your definition of ambient temp is the temperature outside the
system, thermodynamically speaking)

True, I hadn't actually thought of that :) Ambient for the HSF/CPU system
would be the case temperature, and for the computer as a whole would be the
room temperature. So I suppose strictly speaking, if the post was only
concerned about the HSF (as in the Intel document I presume) then ambient
would be case, and if it mentioned things like case fans, etc then it would
be the room temperature. So I agree, it's best to explicitly say case or
room temperatures.

--
Michael Brown
www.emboss.co.nz : OOS/RSI software and more :)
Add michael@ to emboss.co.nz - My inbox is always open



CoCo October 19th 03 10:12 AM

hi

one more thing! whitch bios the mb use. yes bios. i have a water setup and
my mobo use internal temp diod only but dependent on the bios installed i
get different temp readouts. vendor of mobo have changed the algorithm!

gigabyte ga-7n440 pro2 rev2 with a barton 2500+ @ 2100mhz
f1 bios (came with the mb) level out at 36°c. (start temp idle 19°c with bus
dissconect fea.)
f5 bios (latest) level out at 49°c. (start temp idle 32°c with bus
dissconect fea.)
load temps with prime after 5hours with ambient room temp of 20°c
(testmate).
meassured (testmate) case temp 25°c (internat mb case temp probe report 30°c
f5 bios).

coco

"Michael Brown" skrev i meddelandet
...
Gary W. Swearingen wrote:
"Michael Brown" writes:

the case temp as well, and possibly the ambient temp if you're in
some place


And don't just say "ambient temp", because that can mean the air
surrounding the case or the air surrounding the CPU/HSF (or the air
surrounding anything else, strickly speaking). (I base this statement
on the definition of "ambient" and having recently read some Intel doc
that used "ambient" to refer to the air surrounding the CPU/HSF).


(Presuming your definition of ambient temp is the temperature outside the
system, thermodynamically speaking)

True, I hadn't actually thought of that :) Ambient for the HSF/CPU system
would be the case temperature, and for the computer as a whole would be

the
room temperature. So I suppose strictly speaking, if the post was only
concerned about the HSF (as in the Intel document I presume) then ambient
would be case, and if it mentioned things like case fans, etc then it

would
be the room temperature. So I agree, it's best to explicitly say case or
room temperatures.

--
Michael Brown
www.emboss.co.nz : OOS/RSI software and more :)
Add michael@ to emboss.co.nz - My inbox is always open





Michael Brown October 19th 03 11:31 AM

CoCo wrote:
"Michael Brown" skrev i meddelandet
...
Gary W. Swearingen wrote:
"Michael Brown" writes:

the case temp as well, and possibly the ambient temp if you're in
some place

And don't just say "ambient temp", because that can mean the air
surrounding the case or the air surrounding the CPU/HSF (or the air
surrounding anything else, strickly speaking). (I base this
statement on the definition of "ambient" and having recently read
some Intel doc that used "ambient" to refer to the air surrounding
the CPU/HSF).


(Presuming your definition of ambient temp is the temperature
outside the system, thermodynamically speaking)

True, I hadn't actually thought of that :) Ambient for the HSF/CPU
system would be the case temperature, and for the computer as a
whole would be the room temperature. So I suppose strictly speaking,
if the post was only concerned about the HSF (as in the Intel
document I presume) then ambient would be case, and if it mentioned
things like case fans, etc then it would be the room temperature. So
I agree, it's best to explicitly say case or room temperatures.


hi


Hello!

one more thing! whitch bios the mb use. yes bios. i have a water
setup and my mobo use internal temp diod only but dependent on the
bios installed i get different temp readouts. vendor of mobo have
changed the algorithm!

gigabyte ga-7n440 pro2 rev2 with a barton 2500+ @ 2100mhz


Weird ... the Gigabyte site doesn't acknowledge the existance of the Pro2 on
its products page, but has a page on the site ...

f1 bios (came with the mb) level out at 36°c. (start temp idle 19°c
with bus dissconect fea.)


Heh, this is obviously broken ... idling at below the room temperature
violates the second law of thermodynamics :)

f5 bios (latest) level out at 49°c. (start temp idle 32°c with bus
dissconect fea.)
load temps with prime after 5hours with ambient room temp of 20°c
(testmate).
meassured (testmate) case temp 25°c (internat mb case temp probe
report 30°c f5 bios).


Hmm, interesting ... I presume this is with MBM5 (or equavalent), not the
BIOS viewing screen? Also, are you sure it's the diode temperature, ie: does
it spike up pretty much instantly when load is applied (set MBM5 to update
every second to see this clearly)? If it is the diode temperature then
(besides it being impressive :) at ~0.36 deg C/W) this is quite worrying.
There should be no need for motherboard manufacturers to adjust the reading
of the core temperature, unless there is a bug in the measuring. If I get
the time I might send off an email to Gigabyte inquiring about the changes
made, though I doubt I'll get a sensible response.

--
Michael Brown
www.emboss.co.nz : OOS/RSI software and more :)
Add michael@ to emboss.co.nz - My inbox is always open



CoCo October 20th 03 10:38 AM

hi

temp taken with aida and mbm and it is the diod, temp raise very fast with
load (1-2 seconds).
(dont even saw any termistor in the socket, none mentioned in mb book)

gigabyte have put a statement about the temp with the f2 bios! i have
desided that f1 show to low and f5 is to high!
the copper plate just besides the core is not even warm at full load (33°c).

my cooling: swifttech 8500 water kit with a gallon water cooled by hardware
labs extreme II with 4 120mm fans. (external)
right now writing this: 34/21/21/24°c cpu/mb/room/water (after 3hours
gaming,idle for about 20min)

coco

"Michael Brown" skrev i meddelandet
...
CoCo wrote:
"Michael Brown" skrev i meddelandet
...
Gary W. Swearingen wrote:
"Michael Brown" writes:

the case temp as well, and possibly the ambient temp if you're in
some place

And don't just say "ambient temp", because that can mean the air
surrounding the case or the air surrounding the CPU/HSF (or the air
surrounding anything else, strickly speaking). (I base this
statement on the definition of "ambient" and having recently read
some Intel doc that used "ambient" to refer to the air surrounding
the CPU/HSF).

(Presuming your definition of ambient temp is the temperature
outside the system, thermodynamically speaking)

True, I hadn't actually thought of that :) Ambient for the HSF/CPU
system would be the case temperature, and for the computer as a
whole would be the room temperature. So I suppose strictly speaking,
if the post was only concerned about the HSF (as in the Intel
document I presume) then ambient would be case, and if it mentioned
things like case fans, etc then it would be the room temperature. So
I agree, it's best to explicitly say case or room temperatures.


hi


Hello!

one more thing! whitch bios the mb use. yes bios. i have a water
setup and my mobo use internal temp diod only but dependent on the
bios installed i get different temp readouts. vendor of mobo have
changed the algorithm!

gigabyte ga-7n440 pro2 rev2 with a barton 2500+ @ 2100mhz


Weird ... the Gigabyte site doesn't acknowledge the existance of the Pro2

on
its products page, but has a page on the site ...

f1 bios (came with the mb) level out at 36°c. (start temp idle 19°c
with bus dissconect fea.)


Heh, this is obviously broken ... idling at below the room temperature
violates the second law of thermodynamics :)

f5 bios (latest) level out at 49°c. (start temp idle 32°c with bus
dissconect fea.)
load temps with prime after 5hours with ambient room temp of 20°c
(testmate).
meassured (testmate) case temp 25°c (internat mb case temp probe
report 30°c f5 bios).


Hmm, interesting ... I presume this is with MBM5 (or equavalent), not the
BIOS viewing screen? Also, are you sure it's the diode temperature, ie:

does
it spike up pretty much instantly when load is applied (set MBM5 to update
every second to see this clearly)? If it is the diode temperature then
(besides it being impressive :) at ~0.36 deg C/W) this is quite worrying.
There should be no need for motherboard manufacturers to adjust the

reading
of the core temperature, unless there is a bug in the measuring. If I get
the time I might send off an email to Gigabyte inquiring about the changes
made, though I doubt I'll get a sensible response.

--
Michael Brown
www.emboss.co.nz : OOS/RSI software and more :)
Add michael@ to emboss.co.nz - My inbox is always open






All times are GMT +1. The time now is 02:09 PM.

Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
HardwareBanter.com