[ltp] Kick up the Fan.

[Bret Comstock Waldow]

On Fri, 05 Aug 2005 07:58, honey@gneek.com wrote:

> > Has yours ever been serviced?  (Based on cars - no one puts them together
> > like the factory can.)
>
> Yes, lots - agreed, it's like going to the dentist - all your teeth
> fall out a week later.  Having said that, it'd have to be some very
> specific engineer mistake - the fan works for sure, and other on this
> list seemed to be getting the same temps as me...  I'd expect a
> service engineer to leave gaps (ventilation!) more than close them
> up.

In fact, I'd suggest otherwise, although probably not an issue here, and 
unlikely given IBM's quality standards.

A laptop is a rather constrained environment, both physically and thermally - 
while there are many possible solutions, including several successful ones, 
the difference between success and... "less than optimum" is likely more 
critical.  And if anyone is going to do a good job at this, IBM probably 
does.

The possible airflow is quite limited - so it will be purposely directed 
where it can do the most good.  I haven't looked at the hardware manuals yet, 
but I wouldn't expect an air-cooled heat sink on the CPU, I'd expect a heat 
pipe, transferring the heat to the best air stream.

As air flow is limited, any subtraction will have a big effect.  As well, I'd 
say the thermal paste layer on the CPU is also quite important for coupling 
to the heat pipe's input sink.

A loosely assembled case would allow air to flow to more places - a tight one 
keeps it under control.  Probably the case would have to be loose enough to 
alarm you when you picked it up in order to significantly affect air flow, 
but any internal blockage would be significant - a misrouted cable, a part 
that isn't fully seated sticking into the (already quite modest) air route.

Sockets are usually made of synthetic material - bad for heat conduction as 
well as being points of failure due to vibration, shock, and thermal 
expansion-contraction - so I'd expect the CPU to be a soldered surface mount 
component and I doubt a technician would replace one separate from replacing 
the entire motherboard, or at least a CPU daughtercard.  But the heat pipe 
input sink might come off, and if it's not coupled well with thermal paste on 
replacement, that would be an issue.

Overall, I doubt these are real concerns - IBM is good at what it does, and 
you mention others are getting temperatures like yours.

But gaps would be bad - that limited air needs to all go to the right place 
and not be wasted cooling the seams of the case.


Is it on a hard flat surface?  Soft surfaces insulate half the case...

I have mine on a hard desktop at work, or an overturned baking tray (metal - 
even better) on top of a box (to get the display up for ergonomics) at home.


> > Refined breeding, superior education, rugged good looks...
>
> And stunning modesty :)

How perceptive.


> Yes.  Or maybe something genuinely isn't working as hard as it
> should?  Might be interesting to benchmark, but I don't have the same
> model.  Or time :)

I checked the implementation of the CPU governor - it does indeed flip between 
600000 & 1700000 on demand.  I'd expect it to remain at the high end when 
demand remains pegged at 100%.

There are some graphics chip optimizations I read about on this list, and I'm 
not sure if I'm using them now.  I am not using the ATI drivers.  I made my 
first configuration off this page:
http://www.jenny-and-jp.org/index.php?page=thinkpadr51

...as the hardware matched mine.  Now I am using a 2.6.12-rc3 kernel with 
Gentoo's patches.  I've enabled many of the kernel settings as listed in the 
page above, but I'm certain I'm not using the swsusp2 patches - they wouldn't 
fit with the Hauppauge WINTV-PVR-USB2 driver I wanted.

ibm-acpi is compiled as a module external to the kernel.

I'd wonder about the thermal paste if anything.  Or I'd just expect the T40s 
run hotter than the T42s.

Cheers,
Bret