[SOLVED] Unusual CPU Temperature Increase (CPU Scaling).

cwizardone · 01-22-2016, 12:31 PM

Since the last round of updates I've noticed an increase in CPU temperature, but
only when streaming video from Netflix using Alien Bob's Chromium browser and matching WIdevine plugin.
Streaming on Netflix has always raised the temperature but this is running it right up to the edge of the CPU's maximum operating temperature.
Stop the streaming and the temperature drops by 15 -20 degrees.
At first I thought it might be time for some early spring cleaning, but vacuuming the heat sink
did not change a thing.
Any ideas?
Mesa? Xorg?

Emerson · 01-22-2016, 12:36 PM

I just fixed a computer which was constantly overheating by replacing the dried out factory thermal paste with fresh paste.

cwizardone · 01-25-2016, 08:23 AM

Quote:

Originally Posted by Emerson

I just fixed a computer which was constantly overheating by replacing the dried out factory thermal paste with fresh paste.

Thanks for the thought, but just six months ago I removed and
cleaned the CPU and the heat sink/fan, and applied fresh thermal
paste. That should not be the problem.

WiseDraco · 01-25-2016, 01:31 PM

Quote:

Originally Posted by cwizardone

Thanks for the thought, but just six months ago I removed and
cleaned the CPU and the heat sink/fan, and applied fresh thermal
paste. That should not be the problem.

i think, you must start with thinking, analyzing, and diagnostic.
good tool there is system monitor - see, what app, process use a CPU, and how much.
then think about what you can do with it.

a brainless do vacuum cleaning, when cpu going to overwheat, and when it not help, ask for help is not very smart, as for me...

PS there are a complete solution - come to us, to Mac OS X....there, in dark side, we have a cookies

Darth Vader · 01-25-2016, 03:58 PM

Quote:

Originally Posted by cwizardone

Since the last round of updates I've noticed an increase in CPU temperature C, but
only when streaming video from Netflix using Alien Bob's Chromium browser and matching WIdevine plugin.
Streaming on Netflix has always raised the temperature but this is running it right up to the edge of the CPU's maximum operating temperature.
Stop the streaming and the temperature drops by 15 -20 degrees.
At first I thought it might be time for some early spring cleaning, but vacuuming the heat sink
did not change a thing.
Any ideas?
Mesa? Xorg?

I noticed that The Team simplified the /etc/rc.d scripts associated to the kernel modules. Not sure in which update, but, definitively true for latest 4.4.0 kernel. Those scripts used to contains a sequence which set the SCALING GOVERNOR, other hand, to reduce the CPUs speed, depending of load.

Considering that I have some monsters (AMD Buldozer with 8 cores), in mini-ATX computer cases, that thing was very important for me and used to force setting it. Now, all the cores works at full speed, and I use this excerpt from some old rc.modules to get that setup back:

Code:

#!/bin/sh

### CPU frequency scaling policies:
#
# Use the CPUFreq governor 'powersave' as default. This sets the
# frequency statically to the lowest frequency supported by the CPU.
#/sbin/modprobe cpufreq_powersave
#
# Use the CPUFreq governor 'performance' as default. This sets the
# frequency statically to the highest frequency supported by the CPU.
#/sbin/modprobe cpufreq_performance
#
# Use the CPUFreq governor 'conservative' as default. This allows you
# to get a full dynamic frequency capable system by simply loading your
# cpufreq low-level hardware driver. Be aware that not all cpufreq
# drivers support the 'conservative' governor -- the fallback governor
# will be the 'performance' governor.
#/sbin/modprobe cpufreq_conservative
#
# Use the CPUFreq governor 'ondemand' as default. This allows you to
# get a full dynamic frequency capable system by simply loading your
# cpufreq low-level hardware driver. Be aware that not all cpufreq
# drivers support the 'ondemand' governor -- the fallback governor will
# be the performance governor. This seems to be the most-recommended
# scaling policy, so rc.modules will try to load this by default.
/sbin/modprobe cpufreq_ondemand 2>/dev/null

### CPU scaling governor:
#
# Set the default scaling_governor to be used (such as userspace or ondemand)
# if there is a CPUFreq scaling policy module loaded that supports it:
SCALING_GOVERNOR=ondemand
#
# Try to enable the scaling_governor selected above:
if [ -r /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors ]; then
if grep -wq "$SCALING_GOVERNOR" /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors ; then
if [ -r /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor ]; then
for SYSCPUFILE in /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor ; do
echo "$SCALING_GOVERNOR" > $SYSCPUFILE
done
fi
fi
fi

Maybe that script can help you too.

cwizardone · 01-25-2016, 04:05 PM

Quote:

Originally Posted by WiseDraco

i think, you must start with thinking, analyzing, and diagnostic.
good tool there is system monitor - see, what app, process use a CPU, and how much. then think about what you can do with it......

Did all that before posting the original message.

Quote:

Originally Posted by Darth Vader

I noticed that The Team simplified the /etc/rc.d scripts associated to the kernel modules. Not sure in which update, but, definitively true for latest 4.4.0 kernel. Those scripts used to contains a sequence which set the SCALING GOVERNOR, other hand, to reduce the CPUs speed, depending of load.

Considering that I have some monsters (AMD Buldozer with 8 cores), in mini-ATX computer cases, that thing was very important for me and used to force setting it. Now, all the cores works at full speed, and I use this excerpt from some old rc.modules to get that setup back:

# Use the CPUFreq governor 'ondemand' as default. This allows you to
# get a full dynamic frequency capable system by simply loading your
# cpufreq low-level hardware driver. Be aware that not all cpufreq
# drivers support the 'ondemand' governor -- the fallback governor will
# be the performance governor. This seems to be the most-recommended
# scaling policy, so rc.modules will try to load this by default......

......Maybe that script can help you too.

Thank you, Darth Vader, for a thoughtful and useful reply!
Greatly appreciated.

cwizardone · 01-25-2016, 05:51 PM

@ Darth Vader,

Thanks, again. That definitely did the trick!

Richard Cranium · 01-25-2016, 05:54 PM

Looks like cpu scaling selection should be added back into the distribution somewhere. GazL talked about it, but he doesn't appear to use the scaling stuff himself.

cwizardone · 01-26-2016, 08:31 AM

Well, I've been monitoring the changes for 12+ hours (minus some sleep) and the improvements have been nothing short of amazing.
While streaming the same video, the CPU temperature dropped 10 to 14 degrees depending on the action (complexity) of the video at the moment.
What is really drawing my attention is that while simply running a browser, e.g., SeaMonkey, the CPU temperature has dropped more than 20 degrees. This, in turn, means the CPU fan runs at a reduced speed, therefore, making less noise.
The performance of the computer does not seem to be affected in anyway negative.
Just opposite.
This is the best "tweak" I've stumbled across to date.
Thanks, again, Darth Vader!

Chuck56 · 01-26-2016, 09:29 AM

The recent changes to rc.modules where discussed in this thread: http://www.linuxquestions.org/questi...on-4175483641/.

The rc.modules change will be one of many 14.2 changes that may catch many Slackers off guard as they upgrade. Hopefully Pat and team will call out this change in the upgrade documentation. So many improvements to Slackware over the current upgrade cycle!

stormtracknole · 01-26-2016, 09:33 AM

Quote:

Originally Posted by cwizardone

Well, I've been monitoring the changes for 12+ hours (minus some sleep) and the improvements have been nothing short of amazing.
While streaming the same video, the CPU temperature dropped 10 to 14 degrees depending on the action (complexity) of the video at the moment.
What is really drawing my attention is that while simply running a browser, e.g., SeaMonkey, the CPU temperature has dropped more than 20 degrees. This, in turn, means the CPU fan runs at a reduced speed, therefore, making less noise.
The performance of the computer does not seem to be affected in anyway negative.
Just opposite.
This is the best "tweak" I've stumbled across to date.
Thanks, again, Darth Vader!

Do you mind sharing the specific command that you used? I don't seem to have access to the old rc.modules script.

bassmadrigal · 01-26-2016, 09:49 AM

You can just do (assuming 4 cores)

Code:

echo ondemand > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
echo ondemand > /sys/devices/system/cpu/cpu1/cpufreq/scaling_governor
echo ondemand > /sys/devices/system/cpu/cpu2/cpufreq/scaling_governor
echo ondemand > /sys/devices/system/cpu/cpu3/cpufreq/scaling_governor

Or you can use the portion that Darth Vader posted that will do it for all available CPUs.

Code:

### CPU scaling governor:
#
# Set the default scaling_governor to be used (such as userspace or ondemand)
# if there is a CPUFreq scaling policy module loaded that supports it:
SCALING_GOVERNOR=ondemand
#
# Try to enable the scaling_governor selected above:
if [ -r /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors ]; then
  if grep -wq "$SCALING_GOVERNOR" /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors ; then
    if [ -r /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor ]; then
      for SYSCPUFILE in /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor ; do
        echo "$SCALING_GOVERNOR" > $SYSCPUFILE
      done
    fi
  fi
fi

cwizardone · 01-26-2016, 10:02 AM

Quote:

Originally Posted by stormtracknole

Do you mind sharing the specific command that you used? I don't seem to have access to the old rc.modules script.

Not at all.

I simply copied and pasted what Darth Vader was kind enough to provide;

Quote:

From Darth Vader:

.......Now, all the cores works at full speed, and I use this excerpt from some old rc.modules to get that setup back:

#!/bin/sh

### CPU frequency scaling policies:
#
# Use the CPUFreq governor 'powersave' as default. This sets the
# frequency statically to the lowest frequency supported by the CPU.
#/sbin/modprobe cpufreq_powersave
#
# Use the CPUFreq governor 'performance' as default. This sets the
# frequency statically to the highest frequency supported by the CPU.
#/sbin/modprobe cpufreq_performance
#
# Use the CPUFreq governor 'conservative' as default. This allows you
# to get a full dynamic frequency capable system by simply loading your
# cpufreq low-level hardware driver. Be aware that not all cpufreq
# drivers support the 'conservative' governor -- the fallback governor
# will be the 'performance' governor.
#/sbin/modprobe cpufreq_conservative
#
# Use the CPUFreq governor 'ondemand' as default. This allows you to
# get a full dynamic frequency capable system by simply loading your
# cpufreq low-level hardware driver. Be aware that not all cpufreq
# drivers support the 'ondemand' governor -- the fallback governor will
# be the performance governor. This seems to be the most-recommended
# scaling policy, so rc.modules will try to load this by default.
/sbin/modprobe cpufreq_ondemand 2>/dev/null

### CPU scaling governor:
#
# Set the default scaling_governor to be used (such as userspace or ondemand)
# if there is a CPUFreq scaling policy module loaded that supports it:
SCALING_GOVERNOR=ondemand
#
# Try to enable the scaling_governor selected above:
if [ -r /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors ]; then
if grep -wq "$SCALING_GOVERNOR" /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors ; then
if [ -r /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor ]; then
for SYSCPUFILE in /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor ; do
echo "$SCALING_GOVERNOR" > $SYSCPUFILE
done
fi
fi
fi

stormtracknole · 01-26-2016, 10:38 AM

Thank you both!!

stormtracknole · 01-26-2016, 10:39 AM

Quote:

Originally Posted by bassmadrigal

You can just do (assuming 4 cores)

Code:

echo ondemand > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
echo ondemand > /sys/devices/system/cpu/cpu1/cpufreq/scaling_governor
echo ondemand > /sys/devices/system/cpu/cpu2/cpufreq/scaling_governor
echo ondemand > /sys/devices/system/cpu/cpu3/cpufreq/scaling_governor

Or you can use the portion that Darth Vader posted that will do it for all available CPUs.

Code:

### CPU scaling governor:
#
# Set the default scaling_governor to be used (such as userspace or ondemand)
# if there is a CPUFreq scaling policy module loaded that supports it:
SCALING_GOVERNOR=ondemand
#
# Try to enable the scaling_governor selected above:
if [ -r /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors ]; then
  if grep -wq "$SCALING_GOVERNOR" /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors ; then
    if [ -r /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor ]; then
      for SYSCPUFILE in /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor ; do
        echo "$SCALING_GOVERNOR" > $SYSCPUFILE
      done
    fi
  fi
fi

Interesting, I get:

Code:

-su: echo: write error: Invalid argument

When I try to echo ondemand manually.