radeon-profile icon indicating copy to clipboard operation
radeon-profile copied to clipboard

Hainan Chip overheating

Open GatoAmarilloBicolor opened this issue 3 years ago • 4 comments

Hello there, i hope you are fine, i have a problem on Ubuntu with a discreet gpu HAINAN CHIP, because it get reboot always i try to play DRI_PRIME=1 games in an Asus x555qg:

Graphics: Device-1: Advanced Micro Devices [AMD/ATI] Wani [Radeon R5/R6/R7 Graphics] driver: amdgpu v: 5.6.19 Device-2: AMD Sun XT [Radeon HD 8670A/8670M/8690M / R5 M330 / M430 / Radeon 520 Mobile] driver: amdgpu v: 5.6.19 Display: x11 server: X.Org 1.20.8 driver: amdgpu,radeon resolution: 1366x76860Hz, 1280x102460Hz OpenGL: renderer: AMD Radeon R7 Graphics (CARRIZO DRM 3.40.0 5.4.0-58-generic LLVM 11.0.0) v: 4.6 Mesa 20.3

i have give DC and DPM parameters to the kernel grub but can't really configure a lot of things there, i did try corectrl before radeon-profile and there i only had 2 options "Low" = 300mhz(clk) & 300mhz(mclk); "High" 1030mzh (clk) & 900mhz(mclk) oh and "Auto" but this is always the high state 1030mhz x 900mhz same as High <- this is what burn my laptop btw.

Until here i just decide to unistall corecontrol and try with Radeon Profile, and it give me more information and a new mode more than core ctrl is working here 400 mhz (clk) & 600(mclk) (that is making playable some games and i really apreciate that from radeon profile thanks) But, i always have the doubt cause if this gpu can get to 1030mhz with a 105°c of temperature, maybe there is a way to get less than High state 1030mhz x 900mhz and maybe more than Auto - Battery state with 400mhz x 600mhz , i suppose it can be stable on 600mhz x 600mhz maybe? and i will happy with that, but doing $sudo echo > /pp_* files did'nt work for me never yet :(

DRM_IOCTL_AMDGPU_INFO: Invalid argument

pp_dpm_get_mclk_od was not implemented. pp_dpm_get_sclk_od was not implemented. < this thing is worrying me

Now the other thing i was thinking are a way to put ranges on the events but the less temperature event is working over the high temperature event is not working as i think with profile "high" on less than 90°c and a way to up the cycles number to get less than 90°c.... Thanks for atention, sorry for my bad english.

Edit: This output is from amdcovc (easy to copy paste.):

Adapter 0: Wani [Radeon R5/R6/R7 Graphics] Device Topology: 0:1:0 Vendor ID: 4098 (0x1002) Device ID: 39028 (0x9874) Current CoreClock: 626 MHz Current MemoryClock: 667 MHz Core Overdrive: 0 Memory Overdrive: 0 Current Voltage: 875 mV Performance Control: high GPU Load: 3% Current BusSpeed: 0 Current BusLanes: 0 Temperature (edge): 72°C Critical temperature: 0 C FanSpeed Min (Value): 0 FanSpeed Max (Value): 0 Current FanSpeed: -nan% Controlled FanSpeed: no Core Clocks: 300MHz 480MHz 533MHz 576MHz 626MHz 685MHz 720MHz 757MHz Memory Clocks: 667MHz 933MHz Adapter 1: Sun XT [Radeon HD 8670A/8670M/8690M / R5 M330 / M430 / Radeon 520 Mobile] Device Topology: 3:0:0 Vendor ID: 4098 (0x1002) Device ID: 26208 (0x6660) Current CoreClock: 0 MHz Current MemoryClock: 0 MHz Core Overdrive: 0 Memory Overdrive: 0 Performance Control: auto Current BusSpeed: 0 Current BusLanes: 0 Temperature (edge): 72°C Critical temperature: 120 C FanSpeed Min (Value): 0 FanSpeed Max (Value): 0 Current FanSpeed: -nan% Controlled FanSpeed: no

With Radeon Profile R5 SunXT Hainan give that Frequency:

300mhz x 300mhz (This is what i get in low profile) 400mhz x 600mhz (This is auto + battery) 400mhz x 900mhz 800mhz x 900mhz (If i get this as a fixed state i will be very happy, but this just happen in the credits on games and it fast scalate to high never mantain that one ) 955mhz x 900mhz 1030mhz x 900mhz (this is the auto, medium, and high profiles on the app, and always get reboot when i go ingame)

there seems to be other combinations but i dont know how to force it cause i dont have some device files for that :3 an when i try with sudo echo, the output "permission denied" or "bad parameter" thing.

And almost i forgot to tell you THANK YOU for your great work, and thanks for pay attention.

GatoAmarilloBicolor avatar Dec 26 '20 04:12 GatoAmarilloBicolor

https://github.com/marazmista/radeon-profile/issues/164

GatoAmarilloBicolor avatar Dec 26 '20 11:12 GatoAmarilloBicolor

https://gitlab.freedesktop.org/drm/amd/-/issues/1419

GatoAmarilloBicolor avatar Dec 28 '20 13:12 GatoAmarilloBicolor

You should replace your thermal compounds in your GPU.

JoneKone avatar Mar 05 '21 23:03 JoneKone

https://gitlab.freedesktop.org/drm/amd/-/issues/1419# I suspect, the same gpu is used with 2 different states for windows, but in linux there are not distinction, i am not the only who i see with the same problem out there. I already know at less 3 differen case with the same hardware. A way is to set for this model the states just to 800x900 and blacklist the 1030x900 ....

GatoAmarilloBicolor avatar Mar 26 '21 01:03 GatoAmarilloBicolor