Hello,

I'm using ubuntu on Toshiba Tecra M3. My fan is constantly working hard
and making noise. When Windows XP was installed, it controlled the fan
activity based on the CPU usage well. So I want the equivalent
performance on linux as well. I monitored the CPU activity on linux and
it is controlled well. So the problem is that even when CPU is not used
much, the fan activity does not come down. What should I do? Please
help me.

Phidippus wrote:

> Hello,
> 
> I'm using ubuntu on Toshiba Tecra M3. My fan is constantly working hard
> and making noise. When Windows XP was installed, it controlled the fan
> activity based on the CPU usage well. So I want the equivalent
> performance on linux as well. I monitored the CPU activity on linux and
> it is controlled well. So the problem is that even when CPU is not used
> much, the fan activity does not come down. What should I do? Please
> help me.

Those Tosihba models I have seen all have BIOS option for different fan
modes. You may look if you have it too.

There is a program to turn fan on/off included in toshutils (it won't work
for all models). It is possible to write a small script that will monitor
CPU temp and switch fan on/off respectively.

Finally you can ask Toshiba to publish interfaces used by Windows program;
in this case someone may be able to write program that switches fan mode on
the fly ...

=arvi=

I installed toshutils and found out something strange. I think my fan
control actually works during first 10 min (about) [without doing
anything]. But after 10 min, there is no return.

Likewise, I can manually controll the fan speed by using toshutils
during first 10 min, but after 10 min, toshutils doesn't do anything.

On 13 Jan 2006 21:03:09 -0800, "Phidippus" <phidippus@yahoo.com> wrote:

>I installed toshutils and found out something strange. I think my fan
>control actually works during first 10 min (about) [without doing
>anything]. But after 10 min, there is no return.

Sounds thermal?
>
>Likewise, I can manually controll the fan speed by using toshutils
>during first 10 min, but after 10 min, toshutils doesn't do anything.

What is CPU temp?  Perhaps fan is choked on dust bunnies?  CPU temp 
too high == fan full speed.

Grant.
-- 
Cats are smarter than dogs.  You can't make eight cats pull
a sled through the snow.

How do I find out the temp? It appears I can't use lm-sensors.

Phidippus wrote:

> How do I find out the temp? It appears I can't use lm-sensors.

look in /proc/acpi; if you did not compiled ACPI, do it (do not forget
enable Toshiba specific module).

{pts/0}% cat /proc/acpi/thermal_zone/THRM/temperature
temperature:             46 C

there may be more zones, in my case (Portege 4000) the single one is CPU.

=arvi=

Phidippus wrote:

> I installed toshutils and found out something strange. I think my fan
> control actually works during first 10 min (about) [without doing
> anything]. But after 10 min, there is no return.
>

could you rephrase it. I do not understand what "no return" means.

> Likewise, I can manually controll the fan speed by using toshutils
> during first 10 min, but after 10 min, toshutils doesn't do anything.

I thought previous paragraph was about using toshutils to control fan. Could
you please give more details what you are doing.

=arvi=
 

On Fri, 13 Jan 2006 22:19:24 -0800, Phidippus wrote:

> How do I find out the temp? It appears I can't use lm-sensors.

You have to set up lm-sensors, its all in the docs.

Dave

kermit wrote:
> Phidippus wrote:
>
> > I installed toshutils and found out something strange. I think my fan
> > control actually works during first 10 min (about) [without doing
> > anything]. But after 10 min, there is no return.
> >
>
> could you rephrase it. I do not understand what "no return" means.


"no return" I meant the fan activity never comes down even I don't use
computer, even in sleep-mode.

This message show up.

 Sorry, no chips were detected.
 Either your sensors are not supported, or they are
 connected to an I2C bus adapter that we do not support.
 See doc/FAQ, doc/lm_sensors-FAQ.html, or

http://www2.lm-sensors.nu/~lm78/cvs/lm_sensors2/doc/lm_sensors-FAQ.html
 (FAQ #4.24.3) for further information.
 If you find out what chips are on your board, see
 http://secure.netroedge.com/~lm78/newdrivers.html for driver status.

I installed ksensors, and I can see the temp now. When fan is
constantly runniung it is 57C. And even if I don't use much and
processor is running its lowest level for long time, the temp never
comes down.

If the fan activity is controlled fine, I don't uderstand why this
doesn't happen on Windows XP. Is linux using lots of hidden things that
make it hot?

After the fan start working high at 57C, this temperature never comes
down. If I reboot at this point, the temp is still 57C (of course), but
about for first 10 min, fan works at low level. But during this first
10 min, temperature is still at 57C. It appears that regardless of fan
activity, temperature is 57C.

Phidippus wrote:

> After the fan start working high at 57C, this temperature never comes
> down. If I reboot at this point, the temp is still 57C (of course), but
> about for first 10 min, fan works at low level. But during this first
> 10 min, temperature is still at 57C. It appears that regardless of fan
> activity, temperature is 57C.

what temp is under Windows? May be it the lowest level? Using PIII 750MHz I
get about 45-47C depending on CPU frequency as the lowest value; if your
system is newer, it well can have 57C as the minimum.

Fan is working to keep it from rocketing :) BTW if you starts some intensive
work, does temp go high? That is, can you believe sensor reading?

=arvi=

Yes, if I do simulations or something and use CPU at 2GHz (my max),
them will eventually go up to 77C. If I stop the simulation, it will
rather quickly come down to 57C. Also first time I bootup (after long
hours of rest), temp is about 40C, I think.

Maybe everything is working, but temerature just doesn't come down. I
wiped out Windows XP (after using it for one month), so I can't test
what's the temp under Windows. But this noise is psychologically
unhealthy enough that I will wipe out linux and reinstall Windows if
nothing can be done...

I found

cat /proc/acpi/thermal_zone/THRM/temperature
temperature:             57 C

I have no idea what this means, but if this means the computer is
aiming at this temperature, why fan doesn't come down when it is 57C?

Other information in the same directory.

cat /proc/acpi/thermal_zone/THRM/cooling_mode
<setting not supported>
cooling mode:   critical

cat /proc/acpi/thermal_zone/THRM/polling_frequency
<polling disabled>

cat /proc/acpi/thermal_zone/THRM/state
state:                   ok

cat /proc/acpi/thermal_zone/THRM/trip_points
critical (S5):           114 C

I found something strange. Ever since I installed ksensors (last night)
and started monitoring temperature, the fan control is working well.
Right not, I have been using this computer for a few hours, but fan
level is low. Each time I do some heavy computations for my work, the
fan will go up, but once the calculations are done, the fan will go
done once the temperature becomes 58C or so.

ksensors is just sensoring, right? I have no idea why it is working but
I will see if it is "really working". You never know what is going to
happen after a while... Keep fingers crossed...

On Sat, 14 Jan 2006 14:40:26 +0000, Dave Stanton <me@privacy.net> wrote:

>On Fri, 13 Jan 2006 22:19:24 -0800, Phidippus wrote:
>
>> How do I find out the temp? It appears I can't use lm-sensors.
>
>You have to set up lm-sensors, its all in the docs.

You don't need lm_sensors, you might have the info under /proc/acpi.

Grant.
-- 
Cats are smarter than dogs.  You can't make eight cats pull
a sled through the snow.

Phidippus wrote:
> I found something strange. Ever since I installed ksensors (last night)
> and started monitoring temperature, the fan control is working well.
> Right not, I have been using this computer for a few hours, but fan
> level is low. Each time I do some heavy computations for my work, the
> fan will go up, but once the calculations are done, the fan will go
> done once the temperature becomes 58C or so.
>
> ksensors is just sensoring, right? I have no idea why it is working but
> I will see if it is "really working". You never know what is going to
> happen after a while... Keep fingers crossed...


Unfortunately I was wrong. The fan started acting up again suddently
(although the appropriate controlling lasted long) when I wasn't doing
much with computer so temp was still maintained at 57C. After this
point, nothing worked such as toshset, as usual. I guess it goes crazy
suddently for some reason.

Phidippus wrote:

> Yes, if I do simulations or something and use CPU at 2GHz (my max),
> them will eventually go up to 77C. If I stop the simulation, it will
> rather quickly come down to 57C. Also first time I bootup (after long
> hours of rest), temp is about 40C, I think.
> 
> Maybe everything is working, but temerature just doesn't come down. I
> wiped out Windows XP (after using it for one month), so I can't test
> what's the temp under Windows. But this noise is psychologically
> unhealthy enough that I will wipe out linux and reinstall Windows if
> nothing can be done...

have you checked BIOS settings? May be it is set for high performance?

=arvi=

kermit wrote:
> have you checked BIOS settings? May be it is set for high performance?
>
> =arvi=


I checked BIOS and found under Battery Save Mode, Cooling Method is set
as Maximum Performance. It seems that I can select "Performance" and
"Battery Optimized" in addition to "Maximum Performance"

I just switched to "Battery Optimized" (I didn't know what to pick) but
I will not see the effect soon, I suppose. If this is the problem, why
the fan control works for some time and stop working suddently?

So far what I learned is the fan control actually works, but often
times only for short period (5-10 min) and sometimes relatively longer
time (more than 60 min). But "something" happens at some point, which
appears nothing to do with what I am doing with the computer (nor the
cpu temp), and the fan start working high continuously. After this
point, only way to bring the fan activity down is to reboot, or
hibernate and come back immediately. But my "hibernate" doesn't really
work (I'm trying to fix this now too).

The BIOS setting did not work. After about 10 min, the fan started
acting up. I was not even using the computer. The computer was doing
nothing (no screen saver either), but suddently...

Phidippus wrote:

> The BIOS setting did not work. After about 10 min, the fan started
> acting up. I was not even using the computer. The computer was doing
> nothing (no screen saver either), but suddently...

well, I am not sure. I have old Portege 4000 and so far it works the way I
expect, fan is mostly silent.

Try using CPU frequency scaling. I do not know what is supported by your
model; it may be

- acpi-cpufreq
- speedstep-centrino
- p4-clockmod
- speedstep-ich
- speedstep-smi

What chipset do you have? What 'lspci -n' show? What /proc/cpuinfo show?
What /proc/acpi/processor/CPU0/info show?

You may just try to load all of them, they usually should just barf with "no
device found". May be your system comes with detection script (Mandriva
does).

Once you have found driver for CPU frequency scaling, you need to select
governor. Userspace requires external, userspace, daemon (e.g. cpufreqd) to
control processor speed. Some drivers work only with userspace. performance
and powersave set highest resp. lowest available frequencies. ondemand and
conservative both dynamically change frequency depending on load.

I am using acpi-cpufreq (simply because this is the only one working on my
system) and ondemand when on AC and powersave on battery and have very good
results, temperature keeps quite low comparing with running always with
high frequency. Fan rarely works :)

My son has Tecra 3490CT, it is not supported by any frequency driver, but he
set variable CPU frequency in BIOS (I do not remember exact option) and has
basically the same results. When idle, CPU is slow and jumps up when used.

I guess Windows is using this feature, hence it needs fan less often. Give
it a try.

=arvi=

kermit wrote:
> Phidippus wrote:
>
> > The BIOS setting did not work. After about 10 min, the fan started
> > acting up. I was not even using the computer. The computer was doing
> > nothing (no screen saver either), but suddently...
>
> well, I am not sure. I have old Portege 4000 and so far it works the way I
> expect, fan is mostly silent.
>
> Try using CPU frequency scaling. I do not know what is supported by your
> model; it may be
>
> - acpi-cpufreq
> - speedstep-centrino
> - p4-clockmod
> - speedstep-ich
> - speedstep-smi
>
> What chipset do you have? What 'lspci -n' show? What /proc/cpuinfo show?
> What /proc/acpi/processor/CPU0/info show?

cat /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 6
model           : 13
model name      : Intel(R) Pentium(R) M processor 2.00GHz
stepping        : 8
cpu MHz         : 798.125
cache size      : 2048 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 2
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge
mca cmov pat clflush dts acpi mmx fxsr sse sse2 ss tm pbe est tm2
bogomips        : 1576.14


cat /proc/acpi/processor/CPU0/info
processor id:            0
acpi id:                 1
bus mastering control:   yes
power management:        yes
throttling control:      no
limit interface:         no


> You may just try to load all of them, they usually should just barf with "no
> device found". May be your system comes with detection script (Mandriva
> does).

How do I load them?


> Once you have found driver for CPU frequency scaling, you need to select
> governor. Userspace requires external, userspace, daemon (e.g. cpufreqd) to
> control processor speed. Some drivers work only with userspace. performance
> and powersave set highest resp. lowest available frequencies. ondemand and
> conservative both dynamically change frequency depending on load.

I don't know how to identify it but I found

/sys/devices/system/cpu/cpu0/cpufreq$ ls
affected_cpus     scaling_available_frequencies  scaling_governor
stats
cpuinfo_cur_freq  scaling_available_governors    scaling_max_freq
cpuinfo_max_freq  scaling_cur_freq               scaling_min_freq
cpuinfo_min_freq  scaling_driver                 scaling_setspeed

cat scaling_available_frequencies
2000000 1600000 1330000 1060000 800000

cat scaling_available_governors
userspace powersave ondemand conservative performance


> I am using acpi-cpufreq (simply because this is the only one working on my
> system) and ondemand when on AC and powersave on battery and have very good
> results, temperature keeps quite low comparing with running always with
> high frequency. Fan rarely works :)

I think cpu dynamics is working ok. I've been also monitoring it, but
even when the fan start going crazy, cpu is at 800 MHZ as it should be
(provided I'm not making the computer do stuffs).  CPU freq goes up
only when I do calculations, simulations, etc.


> My son has Tecra 3490CT, it is not supported by any frequency driver, but he
> set variable CPU frequency in BIOS (I do not remember exact option) and has
> basically the same results. When idle, CPU is slow and jumps up when used.

I'm beginning to think that the difference between Windows and linux is
the cutoff temperatures. I think my computer's regular CPU temperature
is something around 57 to 58 C. If I can somehow make the computer fan
to start acting up only temperature goes up 59 to 60 degree (with a
script), I may get a similar results with Windows. I don't know how
sensitive this temperature sensor is, but if it is detecting a very
small difference --- for example, when it goes up to 58.6 C, the fan
start working at high while if the temp is 58.5 fan is at low. This
explains why the fan goes crazy even when my temperature reading is not
changing (ksensors only show the temp by 1C). And the fan activity
never comes down because the upper 58 C is the standard temperature of
this CPU. (This fan is just not capable to bring the temperature down
below 58.6 even when the computer is working at 800 MHz). ----- This is
my hypothesis.

I looked for such a script to specify the temperature for fan activity
online and played with it, but my limited knowledge won't allow it to
be possible. I thought about changing to a different distribution, but
if my hypothesis is correct, I guess it's not going to help...

I also found a thing called pwmconfig which allows one to set the fan
activity based on temperature manually, but it requires lm-sensors,
which apparently is not working on my computer...

Phidippus wrote:

> cat scaling_available_frequencies
> 2000000 1600000 1330000 1060000 800000
>

So you have 2GHz CPU. Well, 57C is not as bad :)
 
> 
> I'm beginning to think that the difference between Windows and linux is
> the cutoff temperatures. I think my computer's regular CPU temperature
> is something around 57 to 58 C. If I can somehow make the computer fan
> to start acting up only temperature goes up 59 to 60 degree (with a
> script), I may get a similar results with Windows. 

Show contents of all files under /proc/acpi/thermal_zone (you may have
several subdirectories there). Do you have anything
under /proc/acpi/embedded_controller? If yes, show contents of files too.

=arvi=

kermit wrote:

> Show contents of all files under /proc/acpi/thermal_zone (you may have
> several subdirectories there). Do you have anything
> under /proc/acpi/embedded_controller? If yes, show contents of files too.
>
> =arvi=

I have

/proc/acpi$ ls
ac_adapter  dsdt                 fan             processor     toshiba
alarm       embedded_controller  hotkey          sleep         video
battery     event                info            sony          wakeup
button      fadt                 power_resource  thermal_zone  wmi

but embedded_controller is empty. In thermal_zome, there is one
directory THRM, and within it, contents are

cat cooling_mode
<setting not supported>
cooling mode:   critical

cat polling_frequency
<polling disabled>

cat state
state:                   ok

cat temperature
temperature:             57 C

cat trip_points
critical (S5):           114 C

Hello,

I ended up wiping out ubuntu and installed another linux distrinution
(Vine Linux) beofer going back to WIN. For some reason I don't know,
the fan is working well under Vine. Although file structure is very
different and there is no thermal_zone and I don't know how to monitor
cpu temp at moment. I guess the problem was something to do with
ubuntu, perhaps.............. The reason I decided to leave ubuntu was
because I had other things not working, either, and their help forum
was not very helpful (usually no response at all) although it's
probably the most active linux forum.

Well, I guess I can keep using linux!

Thanks bunch.

Le Sun, 15 Jan 2006 21:42:47 +0300, kermit a �crit�:

> Phidippus wrote:
> 
>> cat scaling_available_frequencies
>> 2000000 1600000 1330000 1060000 800000
>>
> 
> So you have 2GHz CPU. Well, 57C is not as bad :)

May I ask something? I have a Toshiba Qosmio F20 and (according to
gkrellm) the CPU temperature is mostly 64�C (fan is off) to 73�C (fan is
on). 

Are these temperatures too high? I ask Toshiba hot-line: they answer that
the "normal temterature should be 45�C (but they are not so sure...!!!)

CPU is a pentium M 1.7GHz.

If these temperatures are too high, how can I get them lower? 

Thank you.
-- 
Fran�ois Patte
UFR de math�matiques et informatique
Universit� Ren� Descartes
http://www.math-info.univ-paris5.fr/~patte

Fran?ois Patte wrote:

> 
> If these temperatures are too high, how can I get them lower?
>

1. are you using CPU frequency scaling? If no, try to find driver that
supports your system, it may gain you 10 - 15C in average, may be more on
mostly idle system.

2. If your system is not supported by any CPU frequency scaling driver,
check BIOS if it offers some automatic frequency adjustment mode. My son is
using it on old Tecra 3940CT.

=arvi=