-
Notifications
You must be signed in to change notification settings - Fork 20
RX580 warning, writing pp table back crashes driver #6
Comments
Hello, I would need your pp_table to test it locally (and add it to tests). There were historically some bugs in read/write so it's possible the table gets corrupted. |
Sure thing! https://tmm.cx/~hp/pp_table |
Thanks man! I have updated the repo with this one, but it seems it's read/written properly (I mean no deserialization and serialization issue). Can you try to just cat the table from the driver and then to store the same table there again (without using amdtweak)? If the driver crashes this way then it's a bug and should be reported to AMD. Otherwise |
Just writing pp_table to a file and pushing it back doesn't cause a crash just a short monitor flickering. |
For your test: This pp_table is from an Asus GL702ZC laptop which has some peculiar speed restrictions and voltages. I don't know how representative it is of a 'regular' RX580 pp_table. You may want to explicitly name it so when you get your hands on a 'normal' RX580 pp_table file you can test both. |
Fixed : Polaris hard limit record is same as Tonga but with version number '1' (Tonga seams to have v '52')
|
When I run
sudo ./amdtweak --card 0 --read-card-pp --print
the first line printed isWARNING: 'HardLimitTable': Couldn't find versioned type 'HardLimitTable$1'
When trying to make what I think of as a safe like this:
sudo ./amdtweak --card 0 --read-card-pp --set FanTable.TargetTemperature=70 --write-card-pp
The amdgpu driver either just completely stops functioning or prints a kernel panic before locking up the machine.
As for what I'm trying to achieve: I'm trying to undervolt my card to see if I can improve the thermals a little bit.
The text was updated successfully, but these errors were encountered: