Why all the concern over heat?

On Sat, 30 Oct 2004 15:39:36 GMT, "jakesnake66"
wrote:

After reading thousands of posts here and elsewhere regarding overclocking
amd cpus, it seems to me that most overclockers (and pc builder/enthusiasts)
are too concerned about heat - or at least their concerns about particular
heat ranges are off the mark. Mind you, I'm by no means an expert, but I've
overclocked many cpu/mobos and graphics cards, and I often run cpus in heat
ranges that a lot of people seem to consider undesirable or even dangerous.
Is there any real danger running an athlon at 60+ degrees C under load?
Doesn't AMD consider anything under 80-90C "safe?" It seems like the
consensus is that desirable temp ranges are 40-50C, with the upper end of
that giving a lot of people concern, and causing many to go to greater
lengths for cooling. I've seen people brag about temps under 40C, as if
they've really accomplished something. Have they? We're only going to use
these cpus a couple years (at most) anyway, so any supposed
shortening/lenthening of life seems irrelevant to me. On this particular
pc, I'm running a mobile 2500 at 2.3ghz using a TR2M1. I have two 80mm case
fans that I leave off most of the time because of the noise. If I can
believe Sandra, I idle at about 54C, and top out around 64C after 30 minutes
of gaming. My 9800Pro g-card is slightly oc'ed and puts off a bit of heat
itself. The computer runs great, and I see no evidence that my temp range
causes any negative consequences.
Just looking for a little discussion.
Thank you,

jakesnake




I think there's a few things that aren't so obvious, but I may be
wrong. First, I would tend to believe that a CPU would be able to
handle heat better at low speeds than high speeds. A tempature that is
fine for stock speed may not be so healthy for the same CPU when it's
OC'd. Second, I think that heat also shortens the life of the CPU, so
the cooler it runs, the longer it lasts. Lastly, I imagine that the
reccomended maximum temps for a CPU is actually higher than is healthy
for it, but CPU makers list that temp so their processors are used in
more common but cheaper systems. These cheap systems often have cheap
cases, with cheap PSUs which drive up the heat in the CPU as opposed
to a good case and psu, each with good airflow.

These are not based on any specific knowledge, just more or less how I
see things.
----------------------------------------------------------------------

You should start drinking prune juice and KY jelly cocktails right now,
that will make things a lot smoother.
-Felatio Love

Jakesnake: I agree wholeheartedly! I have said similar things on posts to
this newsgroup in the past: many overclockers spend so much time and
effort and worry over heat when it has been shown that there is a huge
variability between motherboards, reporting software, heat sensors, heat
sensor postions, BIOS reporting, etc, - in other words there is no
ACCURATE basis to compare heat reports between various installations.

Secondly, it appears most overclockers are also keen techos - and thus
they change their systems frequently as you said - so why bother worrying
about your precious CPU when you'll probably replace it in 2-3 years!

If I were to be cynical I would propose that perhaps the whole heat issue
has been blown out of all proportion by the marketing machines of the
computer cooling industry! Are we being sold a "bill of goods" to keep our
economies ticking over?

I have always maintained that many overclockers also do not pay enough
attention to their ambient room temperature issue first - before going out
and spending a ****load of money on fans and even fancier and pricier
cooling solutions. Secondly, they also often have a tendency to put too
many fans in their boxes, thus unnecessarily increasing airflow turbulence
inside the case (rather than creating a nice easy continuous unturbulent
flow of air from the front lower case up through the back upper part of
their case).

I would suggest that any CPU temperature from 35-65 or perhaps even 70
(for shorter periods of time) is not an issue for overclockers.


As you mention in the second last paragraph, its all relative. CPU temps
35-45C above 20C room temps in winter are OK but when summer hits and the
outside temps are in the mid to high 30s C then 35-45C on top of a room temp
in the low to mid 30s C can be a cause for concern, and crashes!

Today, with a room temp of about 27C, my CPU hit 57C - no crashes but
because this is 30C above room temp I AM concerned as I have not seen it
that much above room temp.

I suspect the exhaust fan on the back is on the way out, its pitch hase been
varying and it made a funny startup noise yesterday. It happens when you
recycle old fans from other machines.

The 30C CPU temp above room temp will have me more concerned if it is not
the exhaust fan and every thing else is OK. My room temp is in the low to
mid 30s C during summer and that is coming up next - spring here at the
moment.

Xp2000+ Tbred B (12.5*133, 1.6V) @ 13x166, 1.85V with 2x256MB PC2100 @
166FSB + 1x512MB PC3200 @ 166FSB on an ABIT NF7-S V2 that I got about a week
or two ago.

Right now, the room temp is 24-25C and the CPU temp is 33C as I write this.

Dave

"Michael Brown" wrote in message
...
Wes Newell wrote:
[...]
If it's running fine, then don't worry about it. The same setup on
another board may read the temps as 45C and 55C, and even lower on
yet another MB. It's all relative, and without knowing the base
ambient temp, and the case temp, there's not much one can say one way
or the other. The cooler you have is good cooler so I wouldn't worry
about it. Now if the case temps are more than 10C above room temp,
then you don't have very good case cooling.

Damn, you took the words right out of my mouth


Mine to!

Except with the addendum of the position of the case/board temp TR. On my
VIA chipped MSI Kt2 Combo, the TR was between the back of the board and the
back AGP slot, near one ot the Voltage Regulators with passive Heat Sink on
it so it read about 5-8C high most of the time and got hotter when the comp
was worked.

Dave

Today, with a room temp of about 27C, my CPU hit 57C - no crashes but
because this is 30C above room temp I AM concerned as I have not seen it
that much above room temp.

I'm not an expert, as many in this group are, but I really don't follow the
line of thinking regarding the relationship of cpu temps to room/case temp,
other than as an indicator of what's causing the heat. And your comment is
precisely what I was getting at with my original post. Again, I'm not being
critical; I'm trying to understand. What's the relevance of your cpu
hitting 57C, especially given that in your own words, "no crashes but
because this is 30C above room temp I AM concerned as I have not seen it
that much above room temp?" Where did you get that standard of measurement?
Why is 30C above room temp a relevant figure? The way I see it, your cpu is
functioning perfectly well at that temp. End of story, other than - like I
said - as an indication that something else is going wrong.

What got me thinking about all this was a review I read on Newegg the other
day. A guy had written a negative review of a cpu cooler, because his temps
were running 45C-50C, and he, to quote, just is "not comfortable with that."
He then said, "I sleep better at night when my temps run in the 30s." First
of all, the dude needs other things in his life if cpu temps are what
determines the quality of his sleep. Secondly, why would one care?

As for the issue of room temp to cpu temp, are we saying that if I turn on
the air conditioner in my office and lower my room temp by 10 degrees, then
the 55C cpu temp is suddenly unacceptable or a reason for concern? Why does
it matter if 55C is perfectly acceptable for stability? What does the cpu
care if the room is cooler?

Just wanting to know.

jakesnake

"jakesnake66" wrote in message
...

Today, with a room temp of about 27C, my CPU hit 57C - no crashes but
because this is 30C above room temp I AM concerned as I have not seen it
that much above room temp.

I'm not an expert, as many in this group are, but I really don't follow
the
line of thinking regarding the relationship of cpu temps to room/case
temp,
other than as an indicator of what's causing the heat. And your comment
is
precisely what I was getting at with my original post. Again, I'm not
being
critical; I'm trying to understand. What's the relevance of your cpu
hitting 57C, especially given that in your own words, "no crashes but
because this is 30C above room temp I AM concerned as I have not seen it
that much above room temp?" Where did you get that standard of
measurement?
Why is 30C above room temp a relevant figure? The way I see it, your cpu
is
functioning perfectly well at that temp. End of story, other than - like
I
said - as an indication that something else is going wrong.

It is spring at the moment where I am and 27C room temp is about average for
that time of year in the room where the computer is used. When summer hits,
it not uncommon for the room temp to be over 35C on a regular basis - no air
conditioning. There have a few times when the room temp was nearly 40C.
With my CPU temp being 30C above room temp during hard work, which is
frequently, the final temp of the CPU could be as high as 70C - in the CPU
socket. The core of the CPU would be much higher, puting the CPU on the
limit of its heat capacity, possibly leading to its destruction. So I have
to watch the CPU temps. My CPU is overclocked and various voltages have
been increased to keep its performance stable. The extra voltages cause
extra heat in the system because the current (Amps) is also increased in
proportion to the voltage increase, so the total wattage increase is also
increased by the square of either the volt or amps increase. Increase the
volts by 10% would result in about 21% extra watts used.

What got me thinking about all this was a review I read on Newegg the
other
day. A guy had written a negative review of a cpu cooler, because his
temps
were running 45C-50C, and he, to quote, just is "not comfortable with
that."
He then said, "I sleep better at night when my temps run in the 30s."
First
of all, the dude needs other things in his life if cpu temps are what
determines the quality of his sleep. Secondly, why would one care?

Its all relative, temperature wise. The 45-50C CPu temp above could be good
or bad, depending what the CPU was doing - at idle or working at max and the
variation of room temp in which it is working and if the system is
overclocked. So, without stating the room temp and how hard the CPU was
working, those temps are just meaningless drivel, as could be the reviewer's
comments you quoted. As to why someone would care, to it put into
perspective, re-read my previous paragraph.

As for the issue of room temp to cpu temp, are we saying that if I turn on
the air conditioner in my office and lower my room temp by 10 degrees,
then
the 55C cpu temp is suddenly unacceptable or a reason for concern? Why
does
it matter if 55C is perfectly acceptable for stability? What does the cpu
care if the room is cooler?

If your CPU was running at 55C before the air-con was on and the work it was
doing is unchanged, switching on the air-con and dropping the room temp 10C
would see the CPU also drop about 10C temp, with a bit of a time lag.

The main concern with CPU temps is:- will it reach its maximum; how does it
perform as it reaches its maximum rated temp - does it crash or give errors;
does the heat output affect or damage other components - causing other
probems, etc.

Some CPUs give lots of errors well before they reach their rated max temp
and others don't.

Thus, by keeping the CPU and other temps down on a heavily stressed system,
they avoid crashes and errors when the system is pushed hard and highly
overclocked.

The final thing for some is, of course, bragging rights. For others, like
me, they try to be helpful and offer advice to newcomers so they don't
suffer the frustration of a poor performing or bad system set-up.



Dave

Brain Washing!! Say it often enough and it becomes a fact.CPU's are designed to
be able to handle up to 85C. A normal system rarely gets too hot to run unless
there is either some hardware installed incorrectly,like the heatsink or it's
being overclocked without adequate care,or even clogged or dead fans. Most
people aren't even aware of the temps of there systems unless they are
overclocking.

"YanquiDawg" wrote in message
...
Brain Washing!! Say it often enough and it becomes a fact.CPU's are
designed to
be able to handle up to 85C. A normal system rarely gets too hot to run
unless
there is either some hardware installed incorrectly,like the heatsink or
it's
being overclocked without adequate care,or even clogged or dead fans. Most
people aren't even aware of the temps of there systems unless they are
overclocking.

Only 70C for Athlon 64.
Over 60C is risking errors. Corruption. Have you run prime95 torture test
for hours at, say, 70C with no errors? It's possible, of course, but a
risk.


--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at

Thanks, robots.

Ed Light wrote:
[...]
Over 60C is risking errors. Corruption. Have you run prime95 torture
test for hours at, say, 70C with no errors? It's possible, of
course, but a risk.

On my K7D with two Barton 2500's (before I did the ducting, and in the
middle of summer), it happily did a 48 hour Prime95 where the temperature
was in the high 60's, low 70's (depending on the time of day). With
identical settings and configuration, the same CPUs read mid to low 50's in
my Soltek 75DRV5. Who knows which board is right. Both boards are reading
the core diode. I've even seen variation between identical boards (though I
don't recall if I checked the BIOS versions on each). That is why I say the
absolute temperature is meaningless, and the only use it has is to see if a
certain change increased or decreased the temperature.

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

Thus spake Ed Light:

Have you run prime95 torture test
for hours at, say, 70C with no errors? It's possible, of course, but a
risk.

Actually, I ran P95 on my 1800+ Palomino with a die temp in the mid
90s. Of course that was stock speed - my normal overclock require that
it be under 35 C to pass.

--
sls

"Michael Brown" wrote in message
...
Ed Light wrote:
[...]
Over 60C is risking errors. Corruption. Have you run prime95 torture
test for hours at, say, 70C with no errors? It's possible, of
course, but a risk.

On my K7D with two Barton 2500's (before I did the ducting, and in the
middle of summer), it happily did a 48 hour Prime95 where the temperature
was in the high 60's, low 70's (depending on the time of day). With
identical settings and configuration, the same CPUs read mid to low 50's
in
my Soltek 75DRV5. Who knows which board is right. Both boards are reading
the core diode. I've even seen variation between identical boards (though
I
don't recall if I checked the BIOS versions on each). That is why I say
the
absolute temperature is meaningless, and the only use it has is to see if
a
certain change increased or decreased the temperature.

That's _very_ interesting.
It brings to my mind that since the 2500's can be run at the faster bus
speed as 3200's or thereabouts, maybe at their stock speed they have alot of
headroom that, say, a 3200, which could be the same thing with different
bridge cuts, wouldn't have.
I swapped motherboards from one that reads the cpu to one that reads under
the socket, and the temps went down 5C, so they could be pretty close -- but
I guess they could both be way off.


--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at

Thanks, robots.