[SOLVED] Tem.CPU

Skywalker · 2011-05-01 14:31:08

My Arch system has the 2.6.38 ARCH kernel and is running ~ 10º C hotter than an Ubuntu 10.10 with 2.6.35-28 generic kernel in the same PC.
With no charge Arch is ~ 38º while Ubuntu is ~ 28º.
Is a kernel's issue?
Regards

Last edited by Skywalker (2011-05-04 19:54:32)

cybertorture · 2011-05-01 15:21:40

did u try cpufreq ?
https://wiki.archlinux.org/index.php/CP … cy_Scaling

ewaller · 2011-05-01 16:27:07

Skywalker wrote:

My Arch system has the 2.6.38 ARCH kernel and is running ~ 10º C hotter than an Ubuntu 10.10 with 2.6.35-28 generic kernel in the same PC.
With no charge Arch is ~ 38º while Ubuntu is ~ 28º.
Is a kernel's issue?
Regards

I don't believe either of those numbers.

Do you think a 5 degree temperature rise over room ambient as reported by Ubuntu makes sense? It should be much greater than that.
Even the 15 degree rise reported by Arch is not reasonable. This HP laptop, running nothing but Openbox and a browser is running at 50 degrees.

As to the 10 degree delta -- cybertorture is correct. Look into cpufreq and the other power management tools. The Arch philosophy does not assume that everyone wants or needs the tools, so they are not included by default. The wiki is your friend.

Skywalker · 2011-05-01 19:50:48

ewaller:

My room is at 20º C and I'm running the PC without covers. With the covers on the temperature raise 7º ~ 8 º.

The temperature I indicated is doing nothing. For example, playing chess XBoard the temperature is ~ 45º.

My question about the kernel is because in Ubuntu time ago (half a year or more) after one update (updating the kernel) the temperature reached the bios limit, stoping the PC while converting video. This issue was fixed after a posterior update involving also the kernel.

Excuse my typing errors, english not my language.

To all, thanks for your help.

Leonid.I · 2011-05-02 17:17:35

CPU temp. is NOT measured with a thermometer. It is measured by essentially Joule's law : current*voltage. This gives power, which is then somehow converted to temperature. Therefore, the number "xxx C", which you get is pretty meaningless, because it is an instantaneous value. What is important, is the variation of this number under different loads.

Put another way: locally the chip can be ~100 C, but just 1cm outside it is ~50C. I saw cases with nvidia GPUs when this local heat-up caused the contacts to soften and circuit to break down. Another "joke" is vastly different temperatures on some multicore Intel chips: core 1 -- 45C, core 2 -- 55C. How can this be, if the chip size is ~ 0.5cm? All these reading go through BIOS, which is broken in some way or another on 99% of prebuilt machines...

ewaller · 2011-05-02 21:36:46

http://en.wikipedia.org/wiki/Thermal_re … lectronics It is all about the RθJC

Leonid.I wrote:

It is measured by essentially Joule's law : current*voltage. This gives power, which is then somehow converted to temperature.

Not really. It is either measured with a thermocouple in which the voltage at the junction at the dissimilar metals is a direct function of temperature, or, in the case of silicon based sensors used in instrumenting temperatures on the cases of computer components, or for CPUs and GPUs, an internal junction that is dedicated to measuring temperature uses the fact that the voltage across a P-N junction is proportional to exp (kT/q) -- http://en.wikipedia.org/wiki/P-n_junction

Last edited by ewaller (2011-05-02 21:39:16)

ewaller · 2011-05-02 21:50:29

Here is a good paper on how Intel handles it : http://hal.archives-ouvertes.fr/docs/00 … /TMI23.pdf

Skywalker · 2011-05-02 22:10:20

I instaled cpufrequtils and with the governor "ondemand" the frequency has dropped from maximum of 2,5 GHz to 1GHz with no charge.

Now the readed temp. is ~ 30º C, the same than in Ubuntu.

cybertorture, ewaller thanks.

Pennyless · 2011-05-13 21:47:11

I hate to revive an old thread, but I have the same issue and setting the ondemand governor through cpufrequtils didn't fix it for me. Here are some details:

In Archlinux, my output for sensors is:

acpitz-virtual-0
Adapter: Virtual device
temp1:        +60.0°C  (crit = +100.0°C)

coretemp-isa-0000
Adapter: ISA adapter
Core 0:       +58.0°C  (high = +90.0°C, crit = +90.0°C)

coretemp-isa-0001
Adapter: ISA adapter
Core 1:       +57.0°C  (high = +90.0°C, crit = +90.0°C)

In Ubuntu my output for sensors is:

acpitz-virtual-0
Adapter: Virtual device
temp1:        +39.0°C  (crit = +100.0°C)

coretemp-isa-0000
Adapter: ISA adapter
Core 0:       +32.0°C  (high = +90.0°C, crit = +90.0°C)

coretemp-isa-0001
Adapter: ISA adapter
Core 1:       +35.0°C  (high = +90.0°C, crit = +90.0°C)

Windows 7 also hovers in the mid 30s.

Right now I'm going through the wiki looking for possible solutions, but any advice from the forum would be appreciated. The machine in question is a MSI A6000, and just in case it matters I'm running the latest KDE with compositing turned on. The temperature doesn't change with compositing disabled.

cybertorture · 2011-05-13 22:34:33

post some more info like cpufreq-info , see powertop to see how is it runing , check loadaverages etc

also check this --> click

Pennyless · 2011-05-13 23:09:10

I get the same output for cpufreq-info for Arch and Ubuntu:

cpufreq-info 
cpufrequtils 008: cpufreq-info (C) Dominik Brodowski 2004-2009
Report errors and bugs to cpufreq@vger.kernel.org, please.
analyzing CPU 0:
  driver: acpi-cpufreq
  CPUs which run at the same hardware frequency: 0
  CPUs which need to have their frequency coordinated by software: 0
  maximum transition latency: 10.0 us.
  hardware limits: 1.20 GHz - 2.20 GHz
  available frequency steps: 2.20 GHz, 1.60 GHz, 1.20 GHz
  available cpufreq governors: userspace, powersave, conservative, ondemand, performance
  current policy: frequency should be within 1.20 GHz and 2.20 GHz.
                  The governor "ondemand" may decide which speed to use
                  within this range.
  current CPU frequency is 1.20 GHz.
analyzing CPU 1:
  driver: acpi-cpufreq
  CPUs which run at the same hardware frequency: 1
  CPUs which need to have their frequency coordinated by software: 1
  maximum transition latency: 10.0 us.
  hardware limits: 1.20 GHz - 2.20 GHz
  available frequency steps: 2.20 GHz, 1.60 GHz, 1.20 GHz
  available cpufreq governors: userspace, powersave, conservative, ondemand, performance
  current policy: frequency should be within 1.20 GHz and 2.20 GHz.
                  The governor "ondemand" may decide which speed to use
                  within this range.
  current CPU frequency is 1.20 GHz.

Here's my output for powertop in Arch:

Cn                Avg residency       P-states (frequencies)
C0 (cpu running)        (10.3%)         2.21 Ghz    11.0%
polling           0.0ms ( 0.0%)         1.60 Ghz     0.1%
C1 mwait          0.0ms ( 0.0%)         1200 Mhz    88.9%
C2 mwait          2.1ms (80.9%)
C4 mwait          3.5ms ( 8.9%)

Wakeups-from-idle per second : 405.6    interval: 10.0s                                   
no ACPI power usage estimate available

Top causes for wakeups:
  20.2% ( 69.3)   [ehci_hcd:usb1, nvidia] <interrupt>
  13.7% ( 46.9)   [hda_intel] <interrupt>
  10.7% ( 36.6)   [kernel scheduler] Load balancing tick
   9.3% ( 32.0)   USB device  2-1 : Comfort Optical Mouse 1000 (Microsoft)
   8.6% ( 29.4)   kworker/0:0
   7.3% ( 24.9)   [ohci_hcd:usb2] <interrupt>
   5.9% ( 20.1)   knotify4

Here's powertop in Ubuntu:

Cn                Avg residency       P-states (frequencies)
C0 (cpu running)        ( 5.9%)         2.21 Ghz     0.9%
polling           0.0ms ( 0.0%)         1.60 Ghz     0.4%
C1 mwait          0.0ms ( 0.0%)         1200 Mhz    98.7%
C2 mwait          1.0ms (74.1%)
C4 mwait          1.2ms (20.0%)


Wakeups-from-idle per second : 933.0    interval: 10.0s
no ACPI power usage estimate available

Top causes for wakeups:
  83.4% (1272.5)   [eth0, nvidia] <interrupt>
   5.5% ( 83.6)   [kernel core] sk_reset_timer (tcp_delack_timer)
   4.6% ( 70.6)   [kernel scheduler] Load balancing tick
   0.9% ( 14.4)D  transmission-gt
   0.0% (  0.2)D  flush-8:0
   1.2% ( 17.6)   [ahci] <interrupt>

toad · 2011-05-14 07:04:27

Note the differences:

Top causes for wakeups:
20.2% ( 69.3) [ehci_hcd:usb1, nvidia] <interrupt>
13.7% ( 46.9) [hda_intel] <interrupt>
10.7% ( 36.6) [kernel scheduler] Load balancing tick
9.3% ( 32.0) USB device 2-1 : Comfort Optical Mouse 1000 (Microsoft)

cybertorture · 2011-05-14 12:41:29

Well that is not so important difference, but any chanse u post nvidia temperatures also, 'cos i realy suspect this, also any notable "with toutch" difference in temperatures (sensors output may not show real temperatures) .
And one more hint kill Xorg and see after 1-2 minutes what happens.

Arch Linux

#1 2011-05-01 14:31:08

[SOLVED] Tem.CPU

#2 2011-05-01 15:21:40

Re: [SOLVED] Tem.CPU

#3 2011-05-01 16:27:07

Re: [SOLVED] Tem.CPU

#4 2011-05-01 19:50:48

Re: [SOLVED] Tem.CPU

#5 2011-05-02 17:17:35

Re: [SOLVED] Tem.CPU

#6 2011-05-02 21:36:46

Re: [SOLVED] Tem.CPU

#7 2011-05-02 21:50:29

Re: [SOLVED] Tem.CPU

#8 2011-05-02 22:10:20

Re: [SOLVED] Tem.CPU

#9 2011-05-13 21:47:11

Re: [SOLVED] Tem.CPU

#10 2011-05-13 22:34:33

Re: [SOLVED] Tem.CPU

#11 2011-05-13 23:09:10

Re: [SOLVED] Tem.CPU

#12 2011-05-14 07:04:27

Re: [SOLVED] Tem.CPU

#13 2011-05-14 12:41:29

Re: [SOLVED] Tem.CPU

Board footer