Laptop overheating; new option?

[orbert]

Hi:

Just joined up yesterday. I'm running the text mode client
on a laptop (WinXP, 1.8GHz, 512MB memory) and I've
noticed that the fan kicks in almost immediately after I
start and then stays on.

I know from experience that this box runs hot even without
the DF client.
I'm worried about eventually damaging the the laptop by
overheating. For now I cycle the client manually by hitting
'q' every once in a while. But I often forget to re-start it,
so it goes hours without running.

What I'd really like is a startup option in the form

-Ixx,yy

where:

-- "I" stands for "intermittent" (I know, "i" is
already in use; pick another letter if you like. Is
there a complete list of cmd line opts somewhere??)

-- "xx" stands for "time to run" (in seconds)

-- "yy" stands for "time to sleep" (in seconds)

This would let users start the client and say something
like "run for 25 seconds, sleep for 35 seconds, repeat".
Users would find the proportion of on/off time that they
(and their fans) are comfortable with.

Any tips about running on a laptop appreciated.

Regards,
orbert

[Kileran]

you might run into quite a debate over this topic :)

for the past decade or so, there has been a violent debate in the tech industry over wether or not this kind of thing causes damage or prevents it. here are the debates.

1) Running a CPU or any item constantly causes wear and will cause an early burnout. therefore you should power down whenever you dont need your machine running

2) Heating and cooling a peice of hardware causes more damage, because a hot item runs with less problems. starting up a cool machine actually causes the most amount of damage. therefore you should keep things running constantly


the 2nd group here would claim that your idea actually causes more damage, since you would be doing a "Rapid heating and cooling" as it is commonly called.

---------------------------

Now a disclaimer. i am NOT trying to cause a flamewar here. as i said, this is an ongoing VIOLENT debate. i dont want to debate it at all. i have chosen one side, and have followed it for years. i have never lost a peice of hardware due to burnout, so that's why "I" do what i do.

My recomendation to you is to get a temperature monitor program for your laptop's operating system. run DF constantly for about 15 minutes (no more) If you dont like what the program did to your processor's heat, dont run DF at all.

Kileran

[orbert]

Interesting. Thanks for the info; I didn't know it was a
sensitive issue.

I still see a useful purpose for the option I suggested,
though. It could be used with something like "1sec/1sec",
which would keep the processor inside a relatively small
temperature range. (The presumption is that the chip
would reach a fairly steady temp and stay there, since
it would not have time to cool down.) Or it could be
used with something like "15min/45min" which would
limit the number of temp cycles per day.

Also, thanks for the suggestion that I use a CPU temp
monitor; I didn't know such things existed. I found one
here:

http://www.diefer.de/i8kfan/index.html

that's GPL and that supports my laptop; how 'bout that.

Unfortunately, I will have no idea what to do with the
information the program will give me. How hot for
how long is too hot for too long? Without info from the chip/computer manufacturers on what kind of usage is
within design spec, I'm just guessing. (In my case,
uneducated guesses.)

I'm going to email the manufacturer, but I don't have much
hope of getting useful info from them.

[Brian the Fist]

You can already do what you are requesting by using the Windows Task Scheduler, and reading the Readme file thoroughly if necessary

[Paratima]

Originally posted by orbert
Interesting. Thanks for the info; I didn't know it was a sensitive issue. It's only a sensitive issue if you bring the hot laptop into contact with the wrong areas! :scared: :jester:

Seriously though, there are published specs on these things. I believe you can find what you are looking for by looking here (http://www2.amd.com/us-en/Processors/TechnicalResources/1,,30_182_739,00.html) for AMD or here (http://www.intel.com/support/processors/index.htm) for Intel.

[m0ti]

You can try dfDetect, available here (http://t2.technion.ac.il/~sm0ti/dfDetect.zip) .

I still need to do some more testing on win9x machines but for XP it works great.

You can set it up to shut down DF after X time has elapsed, wait Y time, and then start it up again. It will do this intermittently.

[orbert]

Hey howdy hey!

Brian, thanks for the suggestion that I use the XP task
scheduler. I've got a "once an hour for 15 min" task set
up; it will be interesting to see how it works out (if data
is uploaded on exit, etc.). We'll see.

m0ti, thaks for dfDetect. I'll take a look at it. It may be
easier to use or better than the task scheduler in some
way.

Paratima, special thanks for the links to the AMD/Intel specs.
I'm running on a Pentium 4, and guess what? The durn chip
does almost exactly what I'm asking for ... in hardware!

I found what I think is the datasheet for the
version of the P4 that I have; it's here:

http://developer.intel.com/design/pentium4/datashts/249198.htm

The doc in question is 24919805.pdf. Section 7.3 (p.78)
talks about the "Thermal Monitor." It says, in part:

Thermal Monitor controls the processor
temperature by modulating the internal
processor core clocks. ...
When TCC is enabled, and a high temperature
situation exists (i.e. TCC is active), the clocks
will be modulated by alternately turning the
clocks off and on at a a 50% duty cycle.
How 'bout that. Just about exactly the "1 sec on/1 sec
off" philosophy.

Of course, it still does not solve our problem, since what
it does is keep the CPU at max temp, where we want
to keep it in a range under max temp. But it's interesting.

Now, at this point I have:

-- the spec for the max operating temp for this processor,
-- a tool that can monitor the temperature, and
-- a way to tun the app on and off.

One could say that I have the tools I need to do what I
want to do: I could experiment to see what on/off
cycle times keep the temp stable within a range that
I like (say, between 70% and 90% of the max temp).

I probably won't.

Mostly because it's more time than I want to invest in
running a distributed computing client. But also because I
would not feel conifdent in a solution based on this method
and these tools.

The goals are:

1. Keep the processor away from its max temp.
2. Prevent temp cycling outside a user-specified range.

(OK, I guess the second goal subsumes the first,
assuming appropriate high value for the range.)
A method that would work is:

1. Monitor temp
2. When temp is near top of range, reduce load by
2.1 down-clocking the processor, or
2.2 pausing or terminating one or more of a
set of apps that have been identified as low-priority.
3. When temp is near bottom of range, reverse the
action taken in step 2.

Now you could argue that the place for this facility is
not in a d.c. client, and I'd have to agree. Still, it seems
odd that this problem has not been solved. I guess I'm
in the minority, in that my last three main machines
have been laptops. (I ran SETI for a while, and I forgot
why I stopped. Now I remember: it was the same issue
on an older machine.) Anyway, it's beyond my capability
to write the monitor described above.

Thanks to all for the input; I've certainly learned
a lot in this discussion.

Warm regards, ;)
orbert

[tpdooley]

I've run overnight testing on a few dell laptops at work and noticed the floppy disk was hot.. the wooden bench underneath the corner of the laptop with the cpu was hot.. and the keyboard was rather warm.

With desktop cpus, such as AMD Athlon and XP series, they post a maximum temperature of 85-90 degrees C. And we generally try to keep our cpus under 60 degrees C, while not creating a hurricane with shrill 20k rpm fans that annoy neighbors.

Most of the thermal diode solutions are just to protect the cpu from burning up. If the cpu gets to 90 degrees C (or 85 with newer XPs) then the cpu gets throttled back so it doesn't produce so much heat.

Run DF overnight, and then while it's still running, pull up the motherboard monitor software and see where the temperature is. Is it staying at 70% (or less) of the maximum for the cpu?

Most of the laptops with internal fans blow the exhaust air out the rear or from the side near the rear. To help with cooling, you could put the laptop on a stand at least an inch off the surface of the desk/counter. leave most of the bottom of the laptop exposed. Have the rear of the laptop a bit higher than the front (so heat doesn't pool beneat the laptop). And put a quiet fan in front of the laptop to blow air under the laptop and over the keyboard to help cool it down.

Good luck with keeping it in the temperature range you want..


:cheers: