3

My GPU gets ungodly hot very quickly and I would like to see what is pushing it, how hard it is being pushed and then make an informed choice about upgrade, adding cooling or reconfiguring.

How do I profile my GPU's use and the applications making demands upon it?

I'm using: 12.04 (precise) [64-bit]

It's a GeForce GT that I want to profile.

2

1 Answer 1

0

This answer is more generic because you don't give more info like if you run a mobile device and what drivers you are use.

First of all, check your module:

$ modinfo nvidia|grep -e Mobile
parm:           NVreg_Mobile:int

On notebooks you should have set NVreg_Mobile=1.

$ cat /etc/modprobe.d/nvidia.conf
options nvidia NVreg_Mobile=1 NVreg_EnableMSI=1

and rebuild you initrd with update-initramfs -c -t -k all.

The MSI setting is generally good for memory copy with a fast interface.

$ grep nvidia /proc/interrupts
45:  3969293       0   PCI-MSI-edge      nvidia

You should also use the lm-sensors package, which shows you the system temperatures, voltages and fan speeds:

$ sensors
coretemp-isa-0000
Adapter: ISA adapter
Physical id 0:  +43.0 C  (high = +80.0 C, crit = +85.0 C)
Core 0:         +40.0 C  (high = +80.0 C, crit = +85.0 C)
Core 1:         +42.0 C  (high = +80.0 C, crit = +85.0 C)

pkg-temp-0-virtual-0
Adapter: Virtual device
temp1:        +44.0 C  

nct6775-isa-0290
Adapter: ISA adapter
Vcore:          +0.98 V  (min =  +0.00 V, max =  +1.74 V)
in1:            +1.10 V  (min =  +0.00 V, max =  +0.00 V)  ALARM
AVCC:           +3.33 V  (min =  +2.98 V, max =  +3.63 V)
+3.3V:          +3.33 V  (min =  +2.98 V, max =  +3.63 V)
in4:            +1.02 V  (min =  +0.00 V, max =  +0.00 V)  ALARM
in5:            +1.04 V  (min =  +0.00 V, max =  +0.00 V)  ALARM
in6:            +1.07 V  (min =  +0.00 V, max =  +0.00 V)  ALARM
3VSB:           +3.38 V  (min =  +2.98 V, max =  +3.63 V)
Vbat:           +3.31 V  (min =  +2.70 V, max =  +3.63 V)
fan1:          1110 RPM  (min =    0 RPM, div = 32)  ALARM
fan2:          1028 RPM  (min =    0 RPM, div = 32)  ALARM
fan3:           562 RPM  (min =    0 RPM, div = 32)  ALARM
fan4:             0 RPM  (div = 128)
SYSTIN:         +34.0 C  (high = +85.0 C, hyst = +75.0 C)  sensor = CPU diode
CPUTIN:         +35.0 C  (high = +80.0 C, hyst = +75.0 C)  sensor = CPU diode
PECI Agent 0:   +43.5 C  (high = +80.0 C, hyst = +75.0 C)  sensor = Intel PECI
PCH_CHIP_TEMP:  +49.0 C  
PECI Agent 1:    +0.0 C  (high = +80.0 C, hyst = +75.0 C)
cpu0_vid:      +0.000 V
intrusion0:    OK

If you think your GPU-fan is to slow, and the particular fan is listed, then you can use the fancontrol tool to adjust it.

To read the temperature of you GPU use:

$ DISPLAY=:0 nvidia-settings -q GPUCoreTemp -n -t
51 
1
  • So how do I profile the applications making demands upon the GPU? (I know how to play with the fan and take a temperature reading). I really would like to know at what sort of level of floating point operations etc it was cooking to 105°C with. Mar 25, 2014 at 0:48

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .