AMD Radeon Monitoring - Odd Behavior

ryouiki

Member
Has anyone else seen odd behavior for monitoring Memory Clock frequency on Radeon 6000 series cards? It is possible that this is some type of driver issue/ADL/ALDX problem perhaps, but under HWiNFO 7.36 and 7.40 I've noticed that "GPU Memory Clock" values appear to stop working.

The issue first presents itself as the current/min/max/avg values being changed to a grey color, and then eventually the fields themselves become completely blank. At that point if you try to disable/hide this item, then it completely disappears from the "Configure Sensors" and cannot be re-enabled. The first time I encountered this, closing HWiNFO and reopening it did not restore the sensor, only after a complete system reboot would it become visible again.

Unfortunately I don't know of a way to specifically trigger this scenario, I've only noticed it after the fact when I am occasionally checking water temperature values.
 
Please attach a HWiNFO Debug File of this issue. You don't need to run in Debug Mode all time until it happens, just when it happens run HWiNFO again in Debug Mode to capture the data.
 
I believe I have captured this situation,

After the sensor disappeared, shutdown HWiNFO, changed HWiNFO64.INI to add DebugMode=1, and re-ran with sensor missing then closed the application again. Sensor was not available again until system was rebooted.
 
This looks like a bug in the GPU driver/its SMU firmware. HWiNFO uses multiple methods to read the memory clock but they all fail and return 0.
There were problems with reporting of memory clock long ago and AMD also acknowledged this in driver release notes. Try to upgrade the drivers, maybe they have/will fix this...
 
Thank you :)

I'll submit a bug report to AMD, I'm currently on the latest drivers (23.2.2), maybe they have broken something in the recent releases.
 
This bug still occurs on latest AMD drivers (23.3.1). Already submitted bug report to AMD but letting others know.
 
Back
Top