nvidia gpu overheating, how to drop temperature?

[tarauser]

Few days ago i installed tara and i noticed on Thermal tab of nvidia x server settings that right after boot without opening anything else the gpu temperature is at ~72 C. My card is an ~8 years old nvidia geforce 9600GT and i am using these drivers https://imgur.com/zWNAxl7

A week or two ago before putting tara on a ssd, i used to run mint 18.3 on a hdd and after about 15-20 minutes of 0ad (relatively low spec game) gameplay the pc shut down, presumably due to gpu overheating.

So today for the first time i opened it up (it was not dusty, i have taken care of that several times before) and changed the 8 year old heat sink paste (was a sad sight) with the dot method.
Now right after boot the temperature according to nvidia x server settings is ~42 C but in a period of 3-4 minutes it steadily grows up to 65-68 C at about 1 degree per 10 seconds while only having the nvidia setting open (these are my start up programs, basically the same as a clean tara install, i even disabled the print applet https://imgur.com/26EF42B)

could i try anything else to reduce the temperature? maybe change some settings from nvidia x server settings? in the PowerMizer tab there is a Preferred Mode option set to Auto while it can also be set to Adaptive or Prefer Maximum Performance

thanks

EDIT: the heat sink paste i used https://www.aliexpress.com/item/3pcsx5g ... 38004.html
perhaps it's low quality

EDIT2: output of inxi -Fxz

Code: Select all

System:    Host: tarauser-Z87M-D3H Kernel: 4.15.0-24-generic x86_64
           bits: 64 gcc: 7.3.0
           Desktop: Cinnamon 3.8.8 (Gtk 3.22.30-1ubuntu1)
           Distro: Linux Mint 19 Tara
Machine:   Device: desktop Mobo: Gigabyte model: Z87M-D3H v: x.x serial: N/A
           BIOS: American Megatrends v: F11 date: 08/12/2014
CPU:       Quad core Intel Core i5-4460 (-MCP-) 
           arch: Haswell rev.3 cache: 6144 KB
           flags: (lm nx sse sse2 sse3 sse4_1 sse4_2 ssse3 vmx) bmips: 25539
           clock speeds: max: 3400 MHz 1: 1318 MHz 2: 2328 MHz 3: 2050 MHz
           4: 2340 MHz
Graphics:  Card: NVIDIA G94 [GeForce 9600 GT] bus-ID: 01:00.0
           Display Server: x11 (X.Org 1.19.6 )
           drivers: nvidia (unloaded: modesetting,fbdev,vesa,nouveau)
           Resolution: 1920x1080@60.00hz
           OpenGL: renderer: GeForce 9600 GT/PCIe/SSE2
           version: 3.3.0 NVIDIA 340.106 Direct Render: Yes
Audio:     Card Intel 8 Series/C220 Series High Def. Audio Controller
           driver: snd_hda_intel bus-ID: 00:1b.0
           Sound: Advanced Linux Sound Architecture v: k4.15.0-24-generic
Network:   Card: Realtek RTL8111/8168/8411 PCIE Gigabit Ethernet Controller
           driver: r8169 v: 2.3LK-NAPI port: d000 bus-ID: 03:00.0
           IF: enp3s0 state: up speed: 100 Mbps duplex: full mac: <filter>
Drives:    HDD Total Size: 250.1GB (9.5% used)
           ID-1: /dev/sda model: Samsung_SSD_860 size: 250.1GB
Partition: ID-1: / size: 220G used: 15G (7%) fs: ext4 dev: /dev/dm-1
           ID-2: /boot size: 704M used: 215M (33%) fs: ext4 dev: /dev/sda1
           ID-3: swap-1 size: 8.52GB used: 0.00GB (0%)
           fs: swap dev: /dev/dm-2
RAID:      No RAID devices: /proc/mdstat, md_mod kernel module present
Sensors:   System Temperatures: cpu: 29.8C mobo: 27.8C gpu: 0.0:68C
           Fan Speeds (in rpm): cpu: N/A
Info:      Processes: 204 Uptime: 15 min Memory: 1395.8/7914.2MB
           Init: systemd runlevel: 5 Gcc sys: 7.3.0
           Client: Shell (bash 4.4.191) inxi: 2.3.56 

[trytip]

a picture of your nvidia-settings thermal section would be more useful:
in terminal say sudo sensors-detect choose YES to all and at the end select ENTER (basically say yes to everything and be sure to read confirmations)
i have nvidia-9800gt and nvidia-340 comes with nvidia-settings-390 i took it upon myself to uninstall nvidia-settings-390 and installed the correct one nvidia-settings-331 nvidia-smi in terminal will give you more info

on a special note:
i had to run windows to UPGRADE nvidia firmware because the nvidia 9800gt did not have thermal sensors enabled by default, can't say for sure if your 9600 is the same.

[tarauser]

@trytip this is the summary of sudo sensors-detect

Code: Select all

Now follows a summary of the probes I have just done.
Just press ENTER to continue: 

Driver `it87':
  * ISA bus, address 0xa30
    Chip `ITE IT8728F Super IO Sensors' (confidence: 9)

Driver `coretemp':
  * Chip `Intel digital thermal sensor' (confidence: 9)

To load everything that is needed, add this to /etc/modules:
#----cut here----
# Chip drivers
coretemp
it87
#----cut here----
If you have some drivers built into your kernel, the list above will
contain too many modules. Skip the appropriate ones!

Do you want to add these lines automatically to /etc/modules? (yes/NO)no

Unloading i2c-i801... OK
Unloading cpuid... OK

output of nvidia-smi

Code: Select all

Tue Jul 17 20:33:39 2018       
+------------------------------------------------------+                       
| NVIDIA-SMI 340.106    Driver Version: 340.106        |                       
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce 9600 GT     Off  | 0000:01:00.0     N/A |                  N/A |
| 51%   67C    P0    N/A /  N/A |    209MiB /   511MiB |     N/A      Default |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Compute processes:                                               GPU Memory |
|  GPU       PID  Process name                                     Usage      |
|=============================================================================|
|    0            Not Supported                                               |
+-----------------------------------------------------------------------------+

output of apt list --installed | grep nvidia

Code: Select all

WARNING: apt does not have a stable CLI interface. Use with caution in scripts.

nvidia-340/bionic,now 340.106-0ubuntu3 amd64 [installed]
nvidia-opencl-icd-340/bionic,now 340.106-0ubuntu3 amd64 [installed]
nvidia-prime-applet/tara,tara,now 1.0.7 all [installed]
nvidia-settings/bionic,now 390.42-0ubuntu1 amd64 [installed]

it seems i also have nvidia-settings-390 installed, should i try replacing that package?

[trytip]

i did replace nvidia-settings-390 with 331 and then used synaptic to lock the package. but i had issues with old conky not showing cpu fan speed. you can install 331 and check temperatures and if they are the same you can upgrade back to settings 390

but seems that something else is issue. nvidia-smi shows all relevant info and temperatures are as if your gpu is being used. fan speed at 51% and temperature at 67C is around what my gpu is when playing a game

do you see any high usage in top

ps: i also uninstalled nvidia-prime and removed it from the startup. you can do the same if you don't have prime

is this your card? https://www.techpowerup.com/vgabios/88317/88317

YES that heatsink paste is not much help it saying that it's for thermal pads, this is for laptop cpu/gpu and is mainly glue/adhesive not thermal paste.
it needs to be this heatsink paste artic silver or mx4 (sorry didn't see your EDIT:)

[tarauser]

yeah that's the thing, i get that fan speed and temperature while doing nothing at all

nothing unusual in top, i always have system monitor open on Resources tab no matter what i am doing, nothing unusual there either, low cpu usage, low ram usage

had no idea what prime is, after some googling i understand it let's you switch between different gpus, so yeah i disabled it since i have the motherboard graphics disabled from bios

no the card on your link is not the one i have, that one has 1gb memory, as you can see on the nvidia-smi output mine has 512mb, so it must be this one https://www.geforce.com/hardware/deskto ... ifications

thanks for the info on the paste, i will try a different one and post here soon enough

it might be a stupid question but is it worth considering removing the 9600GT altogether and live with the intel processor graphics?
screenshot of Intel HD Graphics 4600 specs: https://imgur.com/LTyttVo
source: https://ark.intel.com/products/80817/In ... o-3_40-GHz

[trytip]

KEEP your 9600GT if you can salvage, it's a GEM. i used my 8800GT for 10 years before it needed thermal paste (i bought it used in 2009) but by then it was too late and had to get a compatible 9800GT. intel graphics may have more RAM but doesn't have the same acceleration in games or internet video as your 9600. do you have a link to your card? is it msi? didn't even pay attention to RAM in your nvidia-smi was only looking at fan and temp. i should read everything more thoroughly

[tarauser]

no it is Albatron, the fan is quite small, i tried googling a bit but couldn't find the exact card i have, i mean if you want a picture i could pm you one later when i close the pc

[trytip]

show me a picture of nvidia-settings PowerMizer it should say how much power it has. my 9800 is this

even the lowest auto power settings is higher than your intel graphics, it should tell you how powerful it is

[tarauser]

[trytip]

you're good there, the only difference between mine and yours is a bit of Mhz in fact i see yours is more powerful in certain conditions than my 9800.
if you're not seeing any weird characters or artifacts as the bios boots you should be good to go for another good 5 to 10 years depending on how much you use it at 100% playing games. i used my 8800 for hours and hours at a time, i know the hardware all depends on manufacturer hopefully yours is built to last.
replace the thermal paste and don't overdo it and your video should last you a long time.

[tarauser]

UPDATE: i removed that chinese paste and applied the Arctic MX-2 i got today, right after boot now it starts at 38 C (was 42 C) and in about the same period of 3-4 mins it us up to 65 C (was 68 C) with the fan at 45 % (was ~52 %).

So there is a slight improvement but it still is, i think, quite a high temperature for idle state, i might open it up and re apply the MX-2 sometime soon because the result after applying wasn't exactly a dot method (the paste is quite thick and the result was messy).

[trytip]

that's an improvement but still not normal. i read reviews on amazon they say mx4 is thin you say mx2 is thick? i'm guessing it matters not thickness as long as the paste won't run over into circuits. hard to tell how much to apply, i watched a lot of youtube and kind of tried to lather it on, but found out not the way to do it. use a drop in the middle and don't spread it, the pressure will do that for you https://www.youtube.com/watch?v=-hNgFNH7zhQ

if the thermal paste is done like this guy here in the video (cpu or gpu makes no diff applying paste should be the same) then you have fan issues do you have power connecting to the card? is it a 6 pin? next time you take it apart take some pictures, don't have to be perfect

i had to do mine cpu a couple of times because i used the spread method at first.

ps: don't overdo it when tightening fan screws i broke a few thinking just a liiitle bit ... snap

[tarauser]

so here are some pictures https://www.dropbox.com/sh/i2vm503ad409 ... NZCERMeoAa

i took the card apart again, at first you can see how i applied the MX-2 when i posted here few days ago, having tightened the screws all the way i can see now that it was too much,
in the other pictures you can see how i cleaned it using only some cotton swabs (don't have isopropyl alcohol) and in the last one how i applied the MX-2 using the dot method (well, 2 dots because it wouldn't stop coming out of the syringe)
this time i didn't tighten the screws all the way and i can't stop thinking whether i left them too loose :/

temperature / fan speed has not changed, gets stabilized at ~65 C ~45% after few minutes of waiting on desktop after booting

i forgot to check the power pins, it's either 4 or 6, i will edit this later

[trytip]

this is heartbreaking since you can't really tell how the paste spreads by how much you applied and you probably wasted almost all of it. i can tell you from experience that the more you open it the more chance of breaking or over tighten will kill your card . on my last attempt i thought it's not tight enough, and one more turn BROKE the screw leaving the part of it stuck inside the nut

from what i see i believe that should be your normal temperature since the fan is not performance fan if you should open it again use LESS thermal paste place it in the middle, do twist the screws until you can not twist but DO NOT over tighten the screws( use tips of fingers on screwdriver so you have less power). by now 3 times removing and replacing screws will break.

i can't find a 8600gt albatron that has the fan by the dvi base. most of them i see the fan is completely on the opposite side at the end of the card

here are pictures of my broken nvidia 8800GT the fan is HUGE. in the second snap you can see the broken screw bottom right even with this big fan my normal temperatures were 48C in the winter and 57C in summer

[Faust]

UPDATE: i removed that chinese paste and applied the Arctic MX-2 i got today, right after boot now it starts at 38 C (was 42 C) and in about the same period of 3-4 mins it us up to 65 C (was 68 C) with the fan at 45 % (was ~52 %).

So there is a slight improvement but it still is, i think, quite a high temperature for idle state, i might open it up and re apply the MX-2 sometime soon because the result after applying wasn't exactly a dot method (the paste is quite thick and the result was messy).

I read this a couple of days ago , so I did some digging around for a thermal paste guide that I wrote years back ( nothing has really changed since then ) .
I've posted it in the " Hardware" section :-
viewtopic.php?f=49&t=274067

It might be some help .... ?