temperature nvidia with Bumblebee in normal use is arround 40C

bOlaE · 2017-10-14 05:24:38

Hi! I recently install archlinux on my new laptop DELL inspiron 15 7000 gaming with nvidia GeForce GTX 1050 Ti. I use Bumblebee with the nvidia driver to automatically switch between Intel integrated GPU and discrete NVIDIA GPU, all seems to be good but temperature at the start is arround 27C - 30C and after 10 minutes temperature arrive arround 40C and stay arround 40C.
When I use windows in normal use ( with example Microsoft Word for example ), temperature is arround 27C - 30C.
What should I do for having temperature arround 27C - 30C?

Thanks

Last edited by bOlaE (2017-10-14 05:26:35)

mrunion · 2017-10-15 00:39:21

Use Windows?

Seriously, though -- 40C is not something to complain about as far as temperature goes. Is there a reason you are wanting it lower?

bOlaE · 2017-10-15 09:19:37

mrunion wrote:

Use Windows?

Windows was the default OS with the laptop so I compare the temperature in windows with my archlinux.

mrunion wrote:

Seriously, though -- 40C is not something to complain about as far as temperature goes. Is there a reason you are wanting it lower?

The reason is that the default OS with the laptop ( windows 10 ) run the nvidia with a temperature arround 27C and in my archlinux it's arround 40C.

Is there any way to reduce or limit the temperature of the nvidia card?

V1del · 2017-10-15 09:44:37

The point of bumblebee is to have the nvidia card disabled unless used though (if bbswitch is installed). But is it really disabled? What is the output of

cat /proc/acpi/bbswitch
dmesg | grep bbswitch

bOlaE · 2017-10-15 10:03:02

The output of cat /proc/acpi/bbswitch is

0000:01:00.0 OFF

And the output of dmesg | grep bbswitch is

[    6.268624] bbswitch: version 0.8
[    6.268627] bbswitch: Found integrated VGA device 0000:00:02.0: \_SB_.PCI0.GFX0
[    6.268631] bbswitch: Found discrete VGA device 0000:01:00.0: \_SB_.PCI0.PEG0.PEGP
[    6.268702] bbswitch: detected an Optimus _DSM function
[    6.268783] bbswitch: Succesfully loaded. Discrete card 0000:01:00.0 is on
[    6.269940] bbswitch: disabling discrete graphics
[   47.497184] bbswitch: enabling discrete graphics
[   55.001164] bbswitch: disabling discrete graphics

seth · 2017-10-15 14:49:37

How do you measure the temperature?
What about the other temperatures in the book? (If the global temerature is at 40°C, say eg. due to CPU load, it's no magic that the GPU takes this temperature as well)

bOlaE · 2017-10-15 17:53:47

seth wrote:

How do you measure the temperature?

I use the nvidia-settings to measure the temperature

seth wrote:

What about the other temperatures in the book? (If the global temerature is at 40°C, say eg. due to CPU load, it's no magic that the GPU takes this temperature as well)

As I said, in Windows OS, temperature is at 27°C... that is why I am concerned

ewaller · 2017-10-15 18:23:30

It is likely that the tool in Windows is incorrect. 27°C is only a couple degrees above room ambient. GPUs draw several Watts even when quiescent. I would expect a not less than a 20 degree rise above ambient

seth · 2017-10-15 18:42:38

"More ambitious fan control" - also not everybody lives in southern California

@bOlaE - you somehow missed that the core of my question was about the other system temperatures.
Also (while we're at it) check the fan behavior on windows and linux.
Since the GPU seems turned off (though nvidia-settings should immediately re-activate it) this is probably not the cause of the temperature. A loaded CPU will easily rise to 60°C and could lead to a case temperature of 40°C - which would be picked up by the GPU.

ewaller · 2017-10-15 19:03:35

Well, the biggest fans won't get you below room ambient. There is probably a thermal resistance of a couple °C/W from the GPU core to the heatsink no matter how much the wind blows.

seth · 2017-10-15 19:34:26

Sure.
Yet my GPU actually is at 27°C (as is my HDD) and the case is at 30°C - room temperature should be around 18°C.

So the OPs claim is realistic - outside southern California where everbody would probably get a cold at those conditions

ewaller · 2017-10-15 20:01:45

18°C seems a bit cool, but fair enough. Funny part is here we virtually do not heat our homes except for about a month. 18C is a comfortable sleeping temperature here, but 20's are normal in the day. In colder parts of the country (Boston, I'm looking at you) the buildings are like saunas.
Never-the-less, the question is really are we burning energy in the GPU needlessly? Or is the rise normal? And if the rise is normal, and there is no danger of running at 45C, is it worth spending extra energy to keep it incrementally cooler. So, is the assertion that the GPU is operating at a lower efficiency than one wold expect?

mrunion · 2017-10-16 14:42:08

FWIW: My nVidia 1070 GTX in my laptop right now sitting on my desk, only running a VM with Windows in it and doing nothing else, the GPU is at 43C. If I turn on the fan in the laptop at it's high speed setting -- I have a keyboard shortcut for this the manufacturer put in -- it drops to about 41C in 20-30 seconds. I didn't wait longer than that, but it might get down to the high 30's at a minimum.

bOlaE · 2017-10-17 10:16:26

seth wrote:

you somehow missed that the core of my question was about the other system temperatures.

The other system temperatures seem to be good because it's only in the part of the materiel where the nvida card is placed that I feel that it's hot

seth wrote:

Also (while we're at it) check the fan behavior on windows and linux.

And the fan seems to be working well because when I launch the command optirun glxspheres64 for example, the fan runs after some second

bOlaE · 2017-10-17 11:06:38

DELL website wrote:

Your options where usually to turn either the integrated or discrete GPU off and use one of them for everything or try and configure Bumblebee on your system. (This could be difficult if you didn't have the option to use Nvidia only in your BIOS.)

Maybe the best solution is to Fully Power Down Discrete GPU?

seth · 2017-10-17 13:13:35

Errrr... "feel"?
Read! => https://www.archlinux.org/packages/extr … m_sensors/
(You'll "feel" the best heat distribution which probably is a copper plate near the GPU ...)

You can try to power it down (if you can) to see whether that has impact on the system temperature.

bOlaE · 2017-10-17 17:07:55

@seth thanks for the tools
the result of the command sensors

iwlwifi-virtual-0
Adapter: Virtual device
temp1: +42.0°C
coretemp-isa-0000
Adapter: ISA adapter
Package id 0: +45.0°C (high = +100.0°C, crit = +100.0°C)
Core 0: +43.0°C (high = +100.0°C, crit = +100.0°C)
Core 1: +44.0°C (high = +100.0°C, crit = +100.0°C)
Core 2: +45.0°C (high = +100.0°C, crit = +100.0°C)
Core 3: +43.0°C (high = +100.0°C, crit = +100.0°C)
acpitz-virtual-0
Adapter: Virtual device
temp1: +25.0°C (crit = +107.0°C)
pch_skylake-virtual-0
Adapter: Virtual device
temp1: +46.0°C
dell_smm-virtual-0
Adapter: Virtual device
Processor Fan: 0 RPM
Video Fan: 0 RPM
CPU: +44.0°C
Ambient: +39.0°C
SODIMM: +39.0°C
Other: +43.0°C

seth · 2017-10-17 18:43:13

So your CPU is > 40°C and the ambient temperature is 39°C, fans are off.
No idea which is thermal zone 0 (25°C, maybe HDD/SDD area?) but even your WiFi is at 42°C, so all in all, that's just the general temperature in the case.

To see whether the GPU is loaded, use nvidia-smi, for the CPUs, just use top (or htop, atop or so) and consult the wiki on fan control if you want to steer things down, but at this point it's no way clear that the GPU picked its temperature by its own load.
Also those temperatures are not critical (and as long as your fans work in general and keep things below like 50° or so everything's absolutely fine - as ewaller pointed out, the fan will also reduce battery time)

V1del · 2017-10-17 19:03:07

I've read through the man page of nvidia-smi and it does mention that the readings might be inaccurate/simply reflecting CPU temp for fanless GPUs and I assume that most optimus cards fall into this category.

bOlaE · 2017-10-17 20:08:31

I have removed the nvidia driver (and blacklist nouveau) and the bumblebee to try booting the OS without using discrete card but the temperature become higher

iwlwifi-virtual-0
Adapter: Virtual device
temp1: +43.0°C
dell_smm-virtual-0
Adapter: Virtual device
Processor Fan: 0 RPM
Video Fan: 0 RPM
CPU: +46.0°C
Ambient: +39.0°C
SODIMM: +39.0°C
Other: +43.0°C
pch_skylake-virtual-0
Adapter: Virtual device
temp1: +46.5°C
coretemp-isa-0000
Adapter: ISA adapter
Package id 0: +60.0°C (high = +100.0°C, crit = +100.0°C)
Core 0: +60.0°C (high = +100.0°C, crit = +100.0°C)
Core 1: +53.0°C (high = +100.0°C, crit = +100.0°C)
Core 2: +55.0°C (high = +100.0°C, crit = +100.0°C)
Core 3: +54.0°C (high = +100.0°C, crit = +100.0°C)
acpitz-virtual-0
Adapter: Virtual device
temp1: +25.0°C (crit = +107.0°C)

Last edited by bOlaE (2017-10-17 20:22:16)

seth · 2017-10-17 20:13:24

This doesn't deactivate the GPU but you're now using the nouveau driver ... or in doubt software emulated OpenGL (desktop environment dependent)

bOlaE · 2017-10-17 20:23:49

So what should I do to fully power down the discrete GPU?
PS: I have blacklisted nouveau but I forgot to mention this

Last edited by bOlaE (2017-10-17 20:25:56)

seth · 2017-10-17 20:34:12

blacklisting will probably only work by "install nouveau /bin/false" - "lsmod | grep nouveau" to check.
The reliable way to power down the GPU is via BIOS/UEFI - otherwise you're actually better of with the optimus control (putting the GPU into a defined state, which has been "off" in your case)

bOlaE · 2017-10-17 21:07:14

I don't think that there is option to disable the GPU in my BIOS/UEFI

mod note: converted oversized image to url. Please see the CoC regarding acceptable image sizes. -- WorMzy

Last edited by bOlaE (2017-10-18 04:48:46)

bOlaE · 2017-10-18 07:11:12

Also I have these error on booting

kernel: ACPI Error: [\_SB_.PCI0.XHC_.RHUB.HS11] Namespace lookup failure, AE_NOT_FOUND (20170531/dswload-210)
kernel: ACPI Exception: AE_NOT_FOUND, During name lookup/catalog (20170531/psobject-252)
kernel: ACPI Exception: AE_NOT_FOUND, (SSDT:xh_OEMBD) while loading table (20170531/tbxfload-228)
kernel: ACPI Error: 1 table load failures, 10 successful (20170531/tbxfload-246)

kernel: mce: [Hardware Error]: CPU 0: Machine Check: 0 Bank 6: ee2000000040110a
kernel: mce: [Hardware Error]: TSC 0 ADDR fef1ffc0 MISC 7880010086
kernel: mce: [Hardware Error]: PROCESSOR 0:906e9 TIME 1508312190 SOCKET 0 APIC 0 microcode 42
kernel: mce: [Hardware Error]: Machine check events logged
kernel: mce: [Hardware Error]: CPU 0: Machine Check: 0 Bank 7: ee2000000040110a
kernel: mce: [Hardware Error]: TSC 0 ADDR fef1ce40 MISC 47880010086
kernel: mce: [Hardware Error]: PROCESSOR 0:906e9 TIME 1508312190 SOCKET 0 APIC 0 microcode 42
kernel: mce: [Hardware Error]: CPU 0: Machine Check: 0 Bank 8: ee2000000040110a
kernel: mce: [Hardware Error]: TSC 0 ADDR fef1ff40 MISC 7880010086
kernel: mce: [Hardware Error]: PROCESSOR 0:906e9 TIME 1508312190 SOCKET 0 APIC 0 microcode 42
kernel: mce: [Hardware Error]: CPU 0: Machine Check: 0 Bank 9: ee2000000040110a
kernel: mce: [Hardware Error]: TSC 0 ADDR fef1ff00 MISC 3880010086
kernel: mce: [Hardware Error]: PROCESSOR 0:906e9 TIME 1508312190 SOCKET 0 APIC 0 microcode 42

kernel: tpm_crb MSFT0101:00: [Firmware Bug]: ACPI region does not cover the entire command/response buffer. [mem 0xfed40000-0xfed4087f flags 0x200] vs fed40080
kernel: tpm_crb MSFT0101:00: [Firmware Bug]: ACPI region does not cover the entire command/response buffer. [mem 0xfed40000-0xfed4087f flags 0x200] vs fed40080

kernel: [drm:lspcon_init [i915]] *ERROR* Failed to probe lspcon
kernel: [drm:intel_ddi_init [i915]] *ERROR* LSPCON init failed on port B

Last edited by bOlaE (2017-10-18 07:20:54)

Arch Linux

#1 2017-10-14 05:24:38

temperature nvidia with Bumblebee in normal use is arround 40C

#2 2017-10-15 00:39:21

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#3 2017-10-15 09:19:37

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#4 2017-10-15 09:44:37

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#5 2017-10-15 10:03:02

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#6 2017-10-15 14:49:37

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#7 2017-10-15 17:53:47

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#8 2017-10-15 18:23:30

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#9 2017-10-15 18:42:38

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#10 2017-10-15 19:03:35

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#11 2017-10-15 19:34:26

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#12 2017-10-15 20:01:45

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#13 2017-10-16 14:42:08

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#14 2017-10-17 10:16:26

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#15 2017-10-17 11:06:38

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#16 2017-10-17 13:13:35

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#17 2017-10-17 17:07:55

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#18 2017-10-17 18:43:13

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#19 2017-10-17 19:03:07

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#20 2017-10-17 20:08:31

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#21 2017-10-17 20:13:24

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#22 2017-10-17 20:23:49

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#23 2017-10-17 20:34:12

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#24 2017-10-17 21:07:14

Re: temperature nvidia with Bumblebee in normal use is arround 40C

#25 2017-10-18 07:11:12

Re: temperature nvidia with Bumblebee in normal use is arround 40C

Board footer