T4x owners who went for the Atheros or 2915ABG, look here!

[ricerocket]

Wanna do a quick favor for me?

Download the Fan Control Utility here and tell me the values you get for the PCI (0xc1 sensor) vs. the rest of the sensors? I guess the system should be on for at least an hour before measuring to let the temps "stabilize".

My 2200BG is consistently 5-10 degrees (celsius) above any other sensor in my T43, and I was wondering if upgrading to the Atheros or 2915ABG chipset would lower the overall system temperature. Thanks in advance.

Here's my values for reference (CPU on adaptive on a T43, 750, x300, 2200BG set at the lowest performance setting):

CPU 44°C (0x78)
APS 44°C (0x79)
PCM 32°C (0x7a)
GPU 47°C (0x7b)
BAT 32°C (0x7c)
BAT 25°C (0x7e)
BUS 45°C (0xc0)
PCI 54°C (0xc1) <= ugh
PWR 44°C (0xc2)


CPU & 2200BG on lowest:

CPU 39°C (0x78)
APS 39°C (0x79)
PCM 29°C (0x7a)
GPU 40°C (0x7b)
BAT 28°C (0x7c)
BAT 25°C (0x7e)
BUS 37°C (0xc0)
PCI 48°C (0xc1) <= you again?
PWR 39°C (0xc2)

[christopher_wolf]

Well, here are my values at full throttle

CPU 53°C (0x78)
APS 45°C (0x79)
PCM 32°C (0x7a)
GPU 51°C (0x7b)
BAT 32°C (0x7c)
BAT 28°C (0x7e)
BUS 45°C (0xc0)
PCI 47°C (0xc1) <--- This is the sensor you are refering to; I have the 2915ABG Intel Wireless Card
PWR 46°C (0xc2)

These tend to change at a power save level though; sometimes it equals the CPU temperature at 39 or 41 degrees.

HTH

[DIGITALgimpus]

I've been thinking the EXACT same thing... nobody has been willing to provide data thus far.

[GomJabbar]

One observation. My DSL modem stays quite warm all of the time. Probably just the nature of the beast.

[dinotopia]

Hi, I have the Thinkpad wireless card, and my T43 model is 2668-G7U

These are my stats running at full tilt:

CPU 71°C (0x78)
APS 53°C (0x79)
PCM 41°C (0x7a)
GPU 64°C (0x7b)
BUS 52°C (0xc0)
PCI 53°C (0xc1)
PWR 53°C (0xc2)

That's with the fan BIOS controlled, of course. It gets hotter sometimes, I've seen it go up to 78 deg.

[DIGITALgimpus]

What I'd really love to have is the following:

Idle load temps
High load temps
High Rez Internal Mobo photo pinpointing sensors.


for a loud T43, and a quiet one. Each with a IBM wireless, and Intel Wireless card.

And compare based on that.

[leif]

CPU 45 (0x78)
APS 44 (0x79)
PCM 35 (0x7a)
GPU 47 (0x7b)
BUS 45 (0xc0)
PCI 49 (0xc1)
PWR 47 (0xc2)

here's mine. I exactly don't know which wireless card in my machine.

[fullauto]

Here are mine for 2668-B1U (1.86MHz, 7200rpm 60GB HD):

CPU 43°C (0x78)
APS 41°C (0x79)
PCM 34°C (0x7a)
GPU 47°C (0x7b)
BAT 30°C (0x7c) <= Taken with T43 running on AC
BAT 26°C (0x7e)
BUS 38°C (0xc0)
PCI 45°C (0xc1)
PWR 40°C (0xc2)

Note: I am using the "Thinkpad default" profile in the power manager, and the FanControl app (which has the fan turned completely turned off).

I am using a Thinkpad-BARTLETT 802.11A/B/G WIRELESS ADAPTER II card with uses the Atheros driver.


My woes with noise + heating problems were profoundly improved once I backed off from "maximum performance" in my power profiles.

Anyone know how much of hit you take in performance by using a "dumbed down" profile in the Power Manager?

[fullauto]

BTW, I bought a USB jump fan this afternoon, and have been playing around with "extra" ventilation around my T43. (I am trying to figure out a way to get back to the "Maximum Performance" profile in the Power Manager).

Guess what? The fan doesn't seem to make a [censored] bit of difference where ever I aim it...

Geesh, methinks the heat removal system of the T43 really sux.

I notice the T60 has a different design.

[christopher_wolf]

Well, the cooling system of the t43 is almost identical to that of the T42; so I don' think that is the problem. It seems, rather, to be the GPU and/or the Wireless cards that are cranking up the temperature; the controller notices and then spools up the fan to quickly compensate. The cooling system, mechanically, is probably not to blame; it is most likely the controller logic. My guess is that they over-compensate with PD control and are sensitive to very minor variations in the rate of change of the temperature; so they flutter around steady-state and can immediately spool the fan up to maximum in a matter of seconds if any big change is deemed to require correction.

[DIGITALgimpus]

I think part of is also design: The wifi card and southbridge (2 chips that can generate a fair amount of heat) are a bit close together.

[fullauto]

Yeah but Chris, don't you think the air inlet vents are too small for the heat load?

Or, perhaps the fan blades need more pitch?

Overall the air flow just seems way too low to me.

BTW, I am assuming the inlet vents are on the bottom of the machine; you know right where all the freak'n heat is...

This is all, of course, in the context of trying to run a T43 at maximum performance (i.e. no battery conservation, using AC power), which is the way I need to run when I am in a R&D phase in my office running Mathematica or C++ compilers balls out.

Ever play with undervolting? Wow...

[christopher_wolf]

There are no vents at the bottom of the T43; there are vents distributed around the Thinkpad, a big one is the PCMCIA bay as far as I can tell, one is near the HDD as well. You could increase the pitch on the fan blades, but that has other concequences as well and would change the characteristic of the airflow. I have also done crude bench makring; I ran MSVC++ 6.0 and Visucal Fortran at the same time with no limits on the processor. Got it to 63 Degrees, but that was it for a 55K line Monte Carlo Simulator and OpenGL displaying it in realtime. Same with MATLAB, Simulink, and Mathematica. I can only get above the low 60s when running them alongside Prime95, BOINC, and cloning instances of the same app over and over again. Only then the fan went to full and stayed there for an hour until it came back down; I noticed the same behavior with the T42 I did this to as well as a T40.

[fullauto]

Chris, what profile are you using in the Power Manager?

When you ran the simulations were you on AC or battery power?

You weren't in a tent in Nome, Alaska by any chance?

[christopher_wolf]

The profile was at Maximum Performance; the ambient temperature was 20-19 Degrees. I tried it both on AC and Battery, no observable difference was noted; I also checked to make sure that the profile did not automatically switch between Battery and AC.

[fullauto]

Well Chris, all I can tell you is when I switch over to Maximum Performance (using AC power) my T43 turns into a blowtorch, even at idle...

Arrrgh.

[DIGITALgimpus]

I've found Access Connections 4.x can spike every several seconds, causing the CPU to cycle up, and it's definately generating more heat as a result.

[dr_st]

The fan control program wouldn't show the PCI temperature on my T42. Could it be that the PCI sensor is present on the T43 only?