You can sign up to get a daily email of our articles, see the Mailing List page.
Strange performance drop, strange temperature readings and strange behavior from r9 380x
TT-392 Apr 24, 2022
I have been using my r9 380x on linux for quite a while now. But since a bit ago its performance seems to be getting worse (random stutters, low framerate). I was thinking about doing a repaste. But before I was gonna do that. I thought I'd run some tests, and I got some... confusing results.

I started by trying to get my gpu temperature, which, in my case meant running: sudo sensors-detect this tool seemed to suggest it had found a temperature sensors in my gpu:
AMD Family 17h thermal sensors...                           Success!
    (driver `k10temp')

next, I ran: sensors
Which gave me the following info:
nvme-pci-0900
Adapter: PCI adapter
Composite:    +42.9°C  (low  = -273.1°C, high = +109.8°C)
                       (crit = +129.8°C)
Sensor 2:     +62.9°C  (low  = -273.1°C, high = -273.1°C)

k10temp-pci-00c3
Adapter: PCI adapter
Tctl:         +37.4°C  

From what I saw from the sensors-detect output, I kinda assumed that k10temp-pci-00c3 would be my gpu temp. But after running furmark for a while, I got the following readings:
nvme-pci-0900
Adapter: PCI adapter
Composite:    +48.9°C  (low  = -273.1°C, high = +109.8°C)
                       (crit = +129.8°C)
Sensor 2:     +62.9°C  (low  = -273.1°C, high = -273.1°C)

k10temp-pci-00c3
Adapter: PCI adapter
Tctl:         +38.2°C  

And, well, 1°C after like 20 min of furmark doesn't seem right. The only temp that really went up was the one labeled: "composite". And, 48.9°C still sounds quite low. But the thing is, I looked at my gpu, and, my gpu fans aren't even spinning (they do some spinning when I boot up the computer, so I assume the fans work at least), so, it is almost like the gpu thinks it is actually not getting much hotter.

So, all of this just left me with a bunch of questions. Am I looking at the right sensors? If not, how do I get a readout for the right sensors? And if I am looking at the right sensors, why is the temperature barely going up?
It is like my gpu has some thermal cap, which makes it throttle super hard at 49 degrees without even bothering to turn on the fans. But that just seems strange. I have never even touched stuff like the fan curves and other gpu settings from linux, also, these super low framerates I have been getting seem to be a recent thing, so you'd think something must have changed.
tuubi Apr 24, 2022
I don't think k10temp-pci-00c3 is your GPU.

Sensors outputs a separate section like this for my 5700 XT:
 
amdgpu-pci-0b00
Adapter: PCI adapter
vddgfx:      725.00 mV 
fan1:           0 RPM  (min =    0 RPM, max = 3200 RPM)
edge:         +44.0°C  (crit = +100.0°C, hyst = -273.1°C)
                       (emerg = +105.0°C)
junction:     +44.0°C  (crit = +107.0°C, hyst = -273.1°C)
                       (emerg = +112.0°C)
mem:          +50.0°C  (crit = +105.0°C, hyst = -273.1°C)
                       (emerg = +110.0°C)
slowPPT:       9.00 W  (cap = 195.00 W)
tuxintuxedo Apr 24, 2022
According to the documentation k10temp is for AMD CPUs, not for VGA.
TT-392 Apr 24, 2022
Quoting: tuubiI don't think k10temp-pci-00c3 is your GPU.

Sensors outputs a separate section like this for my 5700 XT:
 
amdgpu-pci-0b00
Adapter: PCI adapter
vddgfx:      725.00 mV 
fan1:           0 RPM  (min =    0 RPM, max = 3200 RPM)
edge:         +44.0°C  (crit = +100.0°C, hyst = -273.1°C)
                       (emerg = +105.0°C)
junction:     +44.0°C  (crit = +107.0°C, hyst = -273.1°C)
                       (emerg = +112.0°C)
mem:          +50.0°C  (crit = +105.0°C, hyst = -273.1°C)
                       (emerg = +110.0°C)
slowPPT:       9.00 W  (cap = 195.00 W)
Quoting: tuxintuxedoAccording to the documentation k10temp is for AMD CPUs, not for VGA.
I guess it is good to know that that is not the gpu temperature. But the sensors output I posted was the whole sensors output. There isn't a gpu section. So I have no idea how I would actually get that information at this point. Also, my fans did kick in at some point, but only at very low power. And stopped spinning again after a few seconds. So that kinda suggests that the gpu isnt't really getting hot. I tried some other gpu related tools. Like, corectrl, which just shows gpu 0 MHz memory 0 MHz, power 0 W, activity 0%, used memory: 940 MB, and gives me no options to do anything with that gpu. I tried amdgpu-fan, which tells me: "no compatible cards found, exiting", and I also don't have a /sys/class/drm/card0/hwmon/hwmon0/pwm1 file to manually adjust fan speed.

So.. idk if I just have a niche card, but a lot of the information on the web doesn't seem to work for me. Only thing that got me some amount of information is:
sudo lshw -c display
Which rerturns:
  *-display                 
       description: VGA compatible controller
       product: Tonga XT / Amethyst XT [Radeon R9 380X / R9 M295X]
       vendor: Advanced Micro Devices, Inc. [AMD/ATI]
       physical id: 0
       bus info: pci@0000:08:00.0
       version: f1
       width: 64 bits
       clock: 33MHz
       capabilities: pm pciexpress msi vga_controller bus_master cap
       configuration: driver=amdgpu latency=0
       resources: irq:69 memory:e0000000-efffffff memory:f0000000-f0

which doesn't really tell me more, except that something is running at 33MHz?? I doubt that is the actual gpu clockspeed, cause it doesn't go up with load at all, and... with a speed that low... would it even be able to run my desktop smoothly?
At this point I'd almost think it is running on cpu graphics. But I am running an R7 1700, which doesn't have an onboard gpu, so that isn't possible.
tuxintuxedo Apr 24, 2022
What do
glxinfo|grep OpenGL
and
sudo lshw -c video|grep driver
say?
33MHz is something else, so don't worry. (33MHz is probably the PCIe bus clock speed during idle)

Last edited by tuxintuxedo on 24 April 2022 at 4:35 pm UTC
TT-392 Apr 24, 2022
Quoting: tuxintuxedoWhat do
glxinfo|grep OpenGL
and
sudo lshw -c video|grep driver
say?
33MHz is something else, so don't worry. (33MHz is probably the PCIe bus clock speed during idle)
glxinfo|grep OpenGL

OpenGL vendor string: AMD
OpenGL renderer string: AMD Radeon R9 380 Series (tonga, LLVM 13.0.1, DRM 3.42, 5.15.35-1-lts)
OpenGL core profile version string: 4.6 (Core Profile) Mesa 22.0.1
OpenGL core profile shading language version string: 4.60
OpenGL core profile context flags: (none)
OpenGL core profile profile mask: core profile
OpenGL core profile extensions:
OpenGL version string: 4.6 (Compatibility Profile) Mesa 22.0.1
OpenGL shading language version string: 4.60
OpenGL context flags: (none)
OpenGL profile mask: compatibility profile
OpenGL extensions:
OpenGL ES profile version string: OpenGL ES 3.2 Mesa 22.0.1
OpenGL ES profile shading language version string: OpenGL ES GLSL ES 3.20
OpenGL ES profile extensions:



sudo lshw -c video|grep driver

       configuration: driver=amdgpu latency=0
tuxintuxedo Apr 24, 2022
Strange. Everything looks good here, so I don't understand why sensors-detect doesn't find any "amdgpu" related sensors.
While you're here, please consider supporting GamingOnLinux on:

Reward Tiers: Patreon. Plain Donations: PayPal.

This ensures all of our main content remains totally free for everyone! Patreon supporters can also remove all adverts and sponsors! Supporting us helps bring good, fresh content. Without your continued support, we simply could not continue!

You can find even more ways to support us on this dedicated page any time. If you already are, thank you!
Login / Register


Or login with...
Sign in with Steam Sign in with Google
Social logins require cookies to stay logged in.