What are normal operating temperatures for a T500?

[rhombus]

I have a T500 with ATI switchable graphics that I picked up used from a recycler. I've had it for four years now and have been running Gentoo Linux on it.

Recently it has been running hotter. It would idle about 80C and when I compiled on both cores one particular sensor would read as hot as 108C. It has shut down once due to overheating, which seems low considering how hot that one sensor gets (I'm assuming this is the CPU temperature sensor, since I wasn't running any demanding graphics).

I decided to renew the thermal paste and clean the fan with canned air. I also wanted to lubricate the fan and properly clean the blades, but the whole thing is taped shut. I briefly considered peeling back the tape but decided against fiddling with the HSF assembly, since it basically works.

The new thermal paste did seem to work initially -- the "hot" sensor was idling at 60C yesterday at home. Now I am at work it's idling at 71C (although I have a different work surface). I ran a compile and it gets hot as 89C.

That is a big improvement, but it still seems much hotter than it should be. The fan is being run by the EC and is on auto. Typically, it never runs faster than 3000 RPM. That seems odd.

So, here are my questions:

1. What are "normal" or expected operating temperatures for the T500?
2. Is there any point in opening the HSF up to clean and lubricate the fan or to clean the heat sink fins with a fine brush?
3. My fan works, I am just wondering if it needs to be lubricated. If I risk that, though, I might break it -- in which case I would want to have a replacement on hand.I see fans for the T500 advertised on Amazon and Ebay. Just the fan. I have also read a lot of complaints from people who have bought these fans saying they either don't work or don't fit or both. Are these worth bothering with?
4. Should I be using software fan control? The fan never breaks 3100 RPM. That's quiet, which is nice, but not so much when it's very hot.
5. What sensors go to what components? Specifically, nobody seems to be quite sure what lmsensors/thinkpad-isa-0000/temp4 and temp5 are for. In my case, it's temp4 that is running very hot.

[Cigarguy]

71 deg C idle is way too hot. Something is wrong with the cooling system. My T500 idle around 41 deg C, 75ish with some load, and no higher than low 90 when stressed with Prime95 at 20-25 deg C room temp. I use HWMonitor to monitor temp in addition to other useful items. Most Intel CPUs start to throttle down around 90 and will crash at temps not that much higher.

[brchan]

Although uncommon, a failed heatsink can be to blame. Heatsink pipes contain a vaporizing/condensing liquid that helps transfer heat between the ends. If there is a crack or hole somwhere, the liquid can dissipate and make the heatsink ineffective.

[rhombus]

Although uncommon, a failed heatsink can be to blame. Heatsink pipes contain a vaporizing/condensing liquid that helps transfer heat between the ends. If there is a crack or hole somwhere, the liquid can dissipate and make the heatsink ineffective.

Is there any way to confirm if this is the cause? If I buy a used HSF assembly, how can I be sure it's any good?

What are other potential causes for operating temperatures like the ones I'm observing?

[rhombus]

71 deg C idle is way too hot. Something is wrong with the cooling system. My T500 idle around 41 deg C, 75ish with some load, and no higher than low 90 when stressed with Prime95 at 20-25 deg C room temp. I use HWMonitor to monitor temp in addition to other useful items. Most Intel CPUs start to throttle down around 90 and will crash at temps not that much higher.

I agree that it sounds too hot, but if you are right, the system should be throttling down when I compile, and yet it isn't. Also, at full load (compile on both cores) it never exceeds 89 C, which is better than you are reporting. I'm wondering if this particular sensor is even polled by HWMonitor; my wife's T500 is running Windows, reports lower temperatures, but has shut down more than once due to overheating.

If this were your thinkpad, what would you do?

[Hans Gruber]

Use your switchable graphics and toggle over to integrated graphics. While you troubleshoot your temperature problem, your temps will be much lower with integrated graphics.

[rhombus]

Use your switchable graphics and toggle over to integrated graphics. While you troubleshoot your temperature problem, your temps will be much lower with integrated graphics.

I haven't been able to get switchable graphics working reliably in Linux, because I had trouble with it (especially when working with the docking station -- I would have video on my laptop display with the laptop closed while my external monitor would be dark; sometimes it would spontaneously switch at boot-time.) I run with ATI Radeon graphics almost all the time.

Here are a couple of screenshots from the KDE system monitor while at idle but running ATI Radeon 3650 graphics with KDE4.x with the compositing window manager and effects enabled. Here is the CPU load history at the point where I took the temperature readings:
http://ibin.co/2LGA4ZzaGM8H

Here are the temperatures explicitly designated "core",
http://ibin.co/2LGB0eUUTbEA

Here are some other temperature sensors and the fan. The first two sensors are lmsensors/aciptz-virtual-0/temp1 and aciptz-virtual-0/temp2 (I'm assuming the "tz" refers to the "temperature zones", whatever that means exactly) and the second two are lmsensors/thinkpad-isa-0000/temp3 and lmsensors/thinkpad-isa-0000/temp4. Note the last temperature sensor (lmsensors/thinkpad-isa-0000/temp4), that's the "hot" sensor I referred to above. I thought perhaps it might be the GPU, although it also goes way up when I start loading the machine, which doesn't entirely make sense:
http://ibin.co/2LGAdEg1oMAs

You can see on the right all the temperature sensors that the kernel and drivers make available, though not all of those are producing any output.

I checked the thermal settings in the BIOS, and the system is running in "Maximum Performance" mode. Perhaps that's the difference between Cigarguy's T500 and mine?


MOD warning: please read the Forum Rules, especially Section 5. Such huge pictures are not allowed here!

[TPFanatic]

My T500 idles between 55 and 60 C with the fan off and low usage (ie. typing this post). Temperatures go up to 65-75 when I'm running CPU-intensive processes (3D Digital Designer, Movie Maker, Gimp), and I think I've gotten the ATI card to 90, maybe a little over when gaming.

All the Thinkpads I've used don't seem to spin the fan faster than 3100-3200~ RPM, except at the startup initialization screen, without a third party controller. I do use Thinkpad Fan Control, since it uses the faster 3600 RPM "Fan 7" and if necessary 4900 RPM "Fan 64" speeds at higher temperatures. The default "BIOS" Thinkpad fan speed does work, but faster fan speeds do seem to cool off the computer faster.

[Hans Gruber]

Use your switchable graphics and toggle over to integrated graphics. While you troubleshoot your temperature problem, your temps will be much lower with integrated graphics.

I haven't been able to get switchable graphics working reliably in Linux, because I had trouble with it (especially when working with the docking station -- I would have video on my laptop display with the laptop closed while my external monitor would be dark; sometimes it would spontaneously switch at boot-time.) I run with ATI Radeon graphics almost all the time.

Here are a couple of screenshots from the KDE system monitor while at idle but running ATI Radeon 3650 graphics with KDE4.x with the compositing window manager and effects enabled. Here is the CPU load history at the point where I took the temperature readings:
http://ibin.co/2LGA4ZzaGM8H

Here are the temperatures explicitly designated "core",
http://ibin.co/2LGB0eUUTbEA

Here are some other temperature sensors and the fan. The first two sensors are lmsensors/aciptz-virtual-0/temp1 and aciptz-virtual-0/temp2 (I'm assuming the "tz" refers to the "temperature zones", whatever that means exactly) and the second two are lmsensors/thinkpad-isa-0000/temp3 and lmsensors/thinkpad-isa-0000/temp4. Note the last temperature sensor (lmsensors/thinkpad-isa-0000/temp4), that's the "hot" sensor I referred to above. I thought perhaps it might be the GPU, although it also goes way up when I start loading the machine, which doesn't entirely make sense:
http://ibin.co/2LGAdEg1oMAs

You can see on the right all the temperature sensors that the kernel and drivers make available, though not all of those are producing any output.

I checked the thermal settings in the BIOS, and the system is running in "Maximum Performance" mode. Perhaps that's the difference between Cigarguy's T500 and mine?


MOD warning: please read the Forum Rules, especially Section 5. Such huge pictures are not allowed here!

I didn't realize you were running Linux. Go into your bios and set your graphics to integrated graphics and see if that solves your temperature problems. I have one T500 that runs hot as blazes with discrete ATI graphics and cool with integrated graphics. I have another T500 that runs relatively cool with both. With ATI graphics on the cool running T500, it's about 20C higher with ATI graphics vs integrated graphics. All of my temps come from TPFanControl sensors.

The default fan setting on TPFanControl is Smart 1. Change the settings to Smart Fan 2, it runs the fan RPM's over 4000RPM. This is the better alternative rather than running the manual settings to get maximum fan RPM's. On my T500 that runs hot, I have had to use the manual fan control to cool the T500 at times.

[jaspen-meyer]

If it were my machine I'd try throwing a little money at the problem.

Here's what I'd try:

1. disable radeon in bios
2. underclock radeon from command line
3. write a little script to watch cpu frequency to be sure frequency scaling was working. it's easy to screw this up in bios, or by not adding something like 'cpufreqd'. would watch cpu freq with this

Code: Select all

while true;do grep MHz /proc/cpuinfo|tail -1;sleep 1;done

4. replace the, I'm guessing Txxxx (35W TDP), processor with a P9600 (25W TDP)
5. reapply thermal paste to graphics card and cpu

here is the output of ~$ sensors on my T400, which is set to turn on the fan at 70 C:
http://termbin.com/7c7f

the cpufreqd scaling files are here:
/sys/devices/system/cpu/cpu*/cpufreq

[rhombus]

Very interesting post, jaspen. Thanks. My comments and questions:

If it were my machine I'd try throwing a little money at the problem.

Here's what I'd try:

1. disable radeon in bios
2. underclock radeon from command line

How do I underclock the radeon?

3. write a little script to watch cpu frequency to be sure frequency scaling was working. it's easy to screw this up in bios, or by not adding something like 'cpufreqd'. would watch cpu freq with this

Code: Select all

while true;do grep MHz /proc/cpuinfo|tail -1;sleep 1;done

That can be done even easier -- I just created a tab in ksysguard and added CPU frequency and temp4

Here's a screenshot:

http://ibin.co/2M3ktj8MG5eL

It seems to be working fine.

4. replace the, I'm guessing Txxxx (35W TDP), processor with a P9600 (25W TDP)

My processor is a T9400 (2.53 GHz). Is the only difference between the T-series and P-series processors the power consumption? That sounds attractive, I'd even squeeze a few more cycles per second out of that upgrade.

5. reapply thermal paste to graphics card and cpu

Ok -- I've already replaced the thermal paste (with Arctic MX-4). As I mentioned above, that knocked about 20 C off the median operating temperature.

I had a look at my wife's T500 (which is running Windows 7) the other day and installed CPUID's Hardware Monitor. The CPU core temperatures were in the low 40s (41 C to 43 C). However, I observed that the processor frequency was constantly 800 MHz; I also think she is running with integrated graphics. The Lenovo chipset drivers have a habit of turning on switchable graphics in the BIOS, even if this has been explicitly turned off.

On my T500, I am using the ondemand governor policy -- the CPU frequency is freely selected by the governor according to processor load. That is one big difference. I am running KDE4 with compositing graphics and all the window effects turned on. So even if I am "idle", just having nice visuals means I put this machine to work

I notice also that you do not have a temp4 sensor. I'm assuming also that you have radeon graphics turned off. This is leading me more and more to suspect that temp4 is in fact the GPU temperature sensor. There are two temperature sensors on my system that are explicitly labelled as processor core sensors, and those temperatures are usually in the 40s and occasionally in the 50s. Those would be totally acceptable temperatures for me. If I am right about temp4, the bulk of the excess heat is coming from the GPU.

here is the output of ~$ sensors on my T400, which is set to turn on the fan at 70 C:
http://termbin.com/7c7f

Here is my sensors output:
http://pastebin.ca/3246069

I see your fan speed is 0 rpm, but is your fan really off? I can't even imagine running without any fan at all. Maybe I should run a fan daemon (thinkfan?) just to blow off extra heat from time to time. My machine never runs the fan above 3100 rpm. I can barely hear it, it's very quiet.

I'd really rather keep going with radeon graphics, if possible, the difference in performance is big. Maybe underclocking the radeon chip would be an option.

[jaspen-meyer]

How do I underclock the radeon?

The utility I used with my T41 was:

Code: Select all

rovclock - utility to control frequency rates of your Radeon card

My T400 doesn't have switchable graphics. Has only

Code: Select all

        description: Display controller
             product: Mobile 4 Series Chipset Integrated Graphics Controller

Here's a screenshot:
http://ibin.co/2M3ktj8MG5eL

It seems to be working fine.
...
My processor is a T9400 (2.53 GHz). Is the only difference between the T-series and P-series processors the power consumption? That sounds attractive, I'd even squeeze a few more cycles per second out of that upgrade.

Your cpu doesn't have an 'idle' state. The P9600, and your wife's processor (probably a P8400 or P8700), do have an idle state and at that state cpu freq is 800 mHz.

The cpu cycles are about the same, and the cache is 6 MB for each, though I'm not sure how much L1 cache you have compared to the P9600. Here is a comparison from Intel:
http://ark.intel.com/compare/37266,35562

P9600 output from 'lscpu':
http://termbin.com/1l5w

What is your output for 'lscpu' - can upload the output to termbin.com with

Code: Select all

lscpu | nc termbin.com 9999

The Lenovo chipset drivers have a habit of turning on switchable graphics in the BIOS, even if this has been explicitly turned off.

I don't have a machine to test on, or a need to find a work around. The T9400's 35W TDP is the root of your problem.

I see your fan speed is 0 rpm, but is your fan really off? I can't even imagine running without any fan at all.

Yes, my fan's off. In a typical day of computing the fan never turns on. I just ran a test, putting one of the 2 cores at 100%. The fan turned on after 37 seconds, at a speed of 2500 rpm. I'm using a program to control fanspeed but I'm not certain whether it's thinkfan. The service I use is started with "service fancontrol start".

[rhombus]

The utility I used with my T41 was:

Code: Select all

rovclock - utility to control frequency rates of your Radeon card

Unfortunately, this does not work for me, even following the directions in the README:
http://pastebin.ca/3246339

My T400 doesn't have switchable graphics. Has only

Code: Select all

        description: Display controller
             product: Mobile 4 Series Chipset Integrated Graphics Controller

My processor also has 32k L1 cache. I am on an institutional network right now and outbound connections to port 9999 seem to be blocked by the firewall, so I have to use pastebin. Here's my lscpu output: http://pastebin.ca/3246331

The Lenovo chipset drivers have a habit of turning on switchable graphics in the BIOS, even if this has been explicitly turned off.

I don't have a machine to test on, or a need to find a work around. The T9400's 35W TDP is the root of your problem.

The switch back to switchable graphics occurs only when running Windows. I will look into getting a P9600, but I still think that the GPU is contributing (and note again - it's temp4 that is high, and that is not tied to the processor )

I'm going to try forcing integrated graphics and see how that affects thermal performance before I speculate any further

I see your fan speed is 0 rpm, but is your fan really off? I can't even imagine running without any fan at all.

Yes, my fan's off. In a typical day of computing the fan never turns on. I just ran a test, putting one of the 2 cores at 100%. The fan turned on after 37 seconds, at a speed of 2500 rpm. I'm using a program to control fanspeed but I'm not certain whether it's thinkfan. The service I use is started with "service fancontrol start".

I've just learned that fancontrol is part of lm_sensors, which I have installed. If it works for you, then I guess it must work with Thinkpads. I'll try it out. Did you make a custom configuration for it?

[rhombus]

Okay - now I have confirmed it. The excess heat is coming from the ATI Radeon GPU.

I restarted the computer and switched from discrete graphics to integrated graphics in the BIOS.

Code: Select all

$ /usr/sbin/lspci | grep VGA
00:02.0 VGA compatible controller: Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller (rev 07)

Here was the sensors output before, with ATI Radeon graphics activated:
http://pastebin.ca/3246069

Now my sensors look like this:
http://pastebin.ca/3246372

Note temp4 -- it is now showing a negative value. This is the temperature sensor for the ATI Radeon 3650HD. The whole laptop is perceptibly cooler now. My keyboard used to be very warm, now it is cool to the touch.

The thing to keep in mind is that the CPU and GPU share the same heatsink-fan assembly, so if the GPU is running the total heat flux through the HSFA will increase and that will result in reduced heat dissipation from and higher temperatures in the CPU, even if the CPU is not producing much heat itself.

The Radeon mobility chipsets running the open source driver have a history of power and heat management problems (see https://bugs.freedesktop.org/show_bug.cgi?id=36438; this bug has been fixed, but the power management can be configured: https://wiki.gentoo.org/wiki/Radeon#Power_management) I'll give this a try when I have a bit of time to spare, it should result in lower operating temperatures. (I'll also look at a P9600 and tweaking my fan controls.)

[jaspen-meyer]

apt-get install fancontrol lm-sensors
echo 'options thinkpad_acpi fan_control=1' >> sudo tee /etc/modprobe.d/options
echo 'options thinkpad_acpi fan_control=1' >> sudo tee /etc/modprobe.d/thinkpad_acpi.conf
shutdown -P now
sudo service fancontrol stop
sudo pwmconfig
sudo service fancontrol start

I configure the fan to turn on (70 PWM) at 70 C and run nearly full speed (210) at 80 C.

[rhombus]

apt-get install fancontrol lm-sensors
echo 'options thinkpad_acpi fan_control=1' >> sudo tee /etc/modprobe.d/options
echo 'options thinkpad_acpi fan_control=1' >> sudo tee /etc/modprobe.d/thinkpad_acpi.conf
shutdown -P now
sudo service fancontrol stop
sudo pwmconfig
sudo service fancontrol start

I configure the fan to turn on (70 PWM) at 70 C and run nearly full speed (210) at 80 C.

Hmn, no thinkpad_acpi in my modules, perhaps this is built into my kernel. Here's my lsmod:

Code: Select all

# lsmod
Module                  Size  Used by
hdaps                   6896  1
vboxnetadp             17670  0
vboxnetflt             15506  0
vboxdrv               320943  2 vboxnetadp,vboxnetflt
tp_smapi               16688  0
thinkpad_ec             3967  2 hdaps,tp_smapi
cdc_wdm                 9348  0
cdc_ether               4749  0
cdc_acm                18274  0
usbnet                 19191  1 cdc_ether
iwldvm                120612  0
iwlwifi                81059  1 iwldvm

[jaspen-meyer]

try 'pwmconfig'

it will complain if it doesn't find what it needs.

[rhombus]

Success! The power management features of the Radeon GPU work! Following the instructions here: https://wiki.gentoo.org/wiki/Radeon#Power_management

(Good news, too, as docking station video does *not* work with integrated graphics.)

Here's the default configuration:

Code: Select all

# cat /sys/class/drm/card0/device/power_profile 
default
# cat /sys/kernel/debug/dri/0/radeon_pm_info
default engine clock: 600000 kHz
current engine clock: 594000 kHz
default memory clock: 700000 kHz
current memory clock: 693000 kHz
voltage: 1100 mV
PCIE lanes: 16

The "mid" level:

Code: Select all

# echo "mid" > /sys/class/drm/card0/device/power_profile
# cat /sys/kernel/debug/dri/0/radeon_pm_info
default engine clock: 600000 kHz
current engine clock: 297000 kHz
default memory clock: 700000 kHz
current memory clock: 405000 kHz
voltage: 900 mV
PCIE lanes: 16

Switching just from default to mid results in a dramatic drop in operating temperature without an immediately apparent loss of performance. Here's a screenshot of ksysguard: http://ibin.co/2MCM2PLZsDWP. That's a 7 degree C drop in about as many seconds.

The "low" level:

Code: Select all

# echo "low" > /sys/class/drm/card0/device/power_profile
# cat /sys/kernel/debug/dri/0/radeon_pm_info
default engine clock: 600000 kHz
current engine clock: 109680 kHz
default memory clock: 700000 kHz
current memory clock: 405000 kHz
voltage: 900 mV
PCIE lanes: 16

This level results in perceptible loss of performance when drawing more demanding video effects (desktop cube, window collapsing, etc.) Movies work as usual.
After about a minute, the GPU temp sensor (temp4) is down at 52 C:

Code: Select all

thinkpad-isa-0000
Adapter: ISA adapter
fan1:        3041 RPM
temp1:        +49.0°C  
temp2:        +47.0°C  
temp3:        +38.0°C  
temp4:        +52.0°C  
temp5:        +38.0°C  
temp6:            N/A  
temp7:        +34.0°C  
temp8:            N/A  
temp9:        +39.0°C  
temp10:       +49.0°C  
temp11:       +49.0°C  
temp12:           N/A  
temp13:           N/A  
temp14:           N/A  
temp15:           N/A  
temp16:           N/A 

Very nice!

EDIT: These power management features should probably used with caution. After switching between the modes a few times, I ultimately ended up with corrupted video from which I could not recover (not even restarting X did the trick; I had to restart the computer completely).

[rhombus]

try 'pwmconfig'

it will complain if it doesn't find what it needs.

Thanks for the tip - it detects a fan controller. I will test this when I have a quiet moment. Thanks!