Thank you for looking in to this. After the update, the 500MHz is consistently reproducible for me so far. I did have it happen on 3.05 prior but it was less common.
It seems that disconnecting the “problematic” charger (or dock), connecting the 180w charger and then reconnecting the problematic charger will reset this issue and the laptop will start performing properly, even with the other charger.
Here is a screen cap from amdgpu_top while it is in this state:
Now here is another undesirable issue, and I believe it is related to the above issue. When connected to my dock, if I disconnect the 180w charger, the dock connection gets interrupted every time. This means I lose all my external monitor connections along with any other devices connected to the dock, including mounted drives which are forcibly disconnected with everything else. I can plug the 180w charger back in and then disconnect it again to repeat this issue.
@Quin_Chou Since this BIOS update I’ve had some odd thing happening with my battery health.
After update to BIOS 3.06, some hours later I had a crash which I first thought was the FTH issue but later realised it wasn’t after looking at kernel logs.
My battery health went down from 99% ( 85Wh~ ) to 62% ( 52Wh~ ) and wouldn’t charge beyond 53Wh~ and battery was quite warm, so I thought maybe my battery is faulty. Turned off, removed the battery, wait for it to cool down, plugged it back in and then reset BIOS.
I disabled battery extender ( before BIOS reset, charge limit was 80% and extender was turned off ) but was still only charging up to 53Wh~.
This morning the battery health went up to 93% ( 79Wh~ ) now it’s close to 100% again ( 85Wh~ )
I was thinking of buying a new battery because I suspect it might be a faulty battery but then this happened.
Is BIOS 3.06 doing a gradual battery recalibration?
With 3.06 beta I am still experiencing horrible frame pacing due to inconsistent GFX_CLK using a 140w charger, no matter the power profile, I had to restart to fix the issue. But my 90w dock seemed to benefit from the upgrade, the power profiles actually seem to do different stuff from each other.
Linux, setting highest clocks or selecting specific clocks and/or changing max wattage does not make a difference. GPU hits max of ~35w (50w with a 90w charger) and still experiences clock jumps.
It’s fantastic news that the Framework drivers will be synced with the AMD public drivers in the future. The driver delay was my main issue with my Framework 16, so I’m thrilled that there’s a plan to address it. Thanks for your work on the bios and driver updates.
Okay, reporting back after re-trying 3.06. First I tried “Load optimal defaults” from BIOS, this did not fix the issue.
Then I did the full mainboard reset procedure (the one with pressing the intrusion switch 10 times), and since then the clocks have stayed okay. Would be great if the BIOS could reset itself so I don’t have to unscrew the midplate after an update, but this is a beta BIOS afterall, so as long as the release version behaves better, that’d be great.
But for now I’m happy on 3.06 after the full mainboard reset.
That’s really cool, thank you! My first post in 2021 was about that, I think
But I wonder, could you follow other manufacturers like Dell and Lenovo in allowing both upper and lower charge bounds to be specified explicitly? I.e. if I am fine with a 25% float range, I could set the upper charge bound to 80% and the lower charge bound to 55% to let the laptop go slowly up and down?
This was actually available in the previous bios version, 3.05, inside the OS. Two values could be adjusted:
charge_control_end_threshold
charge_control_start_threshold
At least in Linux I was able to set both values how I wanted. However, so far I do not have either option available anymore on this update. I either have something set wrong in the bios that causes it to not open these values up to the OS, or some other incompatibility with the Linux kernel was introduced.
Yes, I saw those values in EC source when I looked into this, but one of them (min) seemed to do nothing in the EC source and only the max was actually used.
Edit, but good point, to check EC source once again, this could be implemented there now properly.
Trying the new BIOS, charging seems to go past the set charge limit. Originally reset the EC by resetting the board and that seemed to resolve it. But after the system went to sleep and I got it stuck in the 544MHz state, the battery charge limit was ignored.
This is what happened with me to begin with after updating to BIOS 3.06. My limit was set to 80% but went past that ( 89-93% ), I rebooted to check that the limit was set to 80% and it was
later I had a crash and noticed my battery health drastically dropped, won’t repeat anymore as I’ve already stated the rest of this earlier
I’m wondering if this new BIOS is doing some battery recalibration ( which is good if it is )
Battery health is gradually going back n forth now between 99-100% which is normal but this is with 100% charge limit and will set it back to 80% later if I’m happy with the battery
Maybe the battery was charged to 80%, but the BIOS error caused the system to believe that the battery was charged to 89%-93%, so battery health dropped
