You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
when 0pwm fan settings for IT8659 controller are set the fans still don't fully stop, but also when they barely spin they don't really contribute to cooling perf
How reproducible
Reproducible with stock AMI firmware, permanent. (I assume it also applies to coreboot as the fan settings in the source are almost identical to AMI options)
Fan Mode | VarStore: IT8659_SMF | VarOffset: 0x0 | Size: 0x1
Manual Mode: 0x0 <-- set to 0x0
Automatic Mode: 0x1
Cpu Fan1 Setting
Manual PWM Setting | VarStore: IT8659_SMF | VarOffset: 0x6 | Size: 0x1
Min: 0x0 | Max: 0xFF | Step: 0x1 <-- set to 0x0
...
Cpu Fan2 Setting
Fan Mode | VarStore: IT8659_SMF | VarOffset: 0x8 | Size: 0x1
Manual Mode: 0x0 <-- set to 0x0
Automatic Mode: 0x1
Cpu Fan2 Setting
Manual PWM Setting | VarStore: IT8659_SMF | VarOffset: 0xE | Size: 0x1
Min: 0x0 | Max: 0xFF | Step: 0x1 <-- set to 0x0
And with these settings the fans occasionally spin up to like 100 rpm and then turn off leaving no cooling performance benefit on the table except consuming extra energy and producing noise
Expected behavior
In manual pwm mode when setting is set to 0 then fans should just switch off. Or at least in the fan settings there should be a separate boolean toggle that instructs IT8659 to switch off fan ctl completely.
At first you might think but why not leave these 'sane' defaults on. Well my argument is if the user is allowed to set fan speeds that are so low that barely move a cubic millimeter of air per s contributing nothing to the overall heat dissipation performance especially on the i3 vp66 unit might as well allow users to turn the whole thing off the in first place saving on noise levels + giving user the freedom disconnecting extra moving parts from the system if they intend to.
Actual behavior
Fans spin for like 1 second at rouhgly 100-200 rpm and then switch off and cycle like this every 5 seconds or so
Screenshots
No response
Additional context
My findings mainly come from the AMI Bios variant of the firmware however I suspect this applies to coreboot components in the same vein as well. There are clear references to it8659 in AMI BIOS firmware & corebios. I believe should probably smart enough where it can receive a simple instruction of turning off the fan control completely.
Solutions you've tried
Patching AMI BIOS (and flashing directly on chip just to be extra sure) with tinkered Platform Thermal Configuration values to see if this was linked to anything intel thermals related, or any extra hidden values that at least had some relation to fan/pwm controls but to no avail.
I do suspect the main pwm logic lives on that super io chip and programming that chip esp with no datasheet avail would be a great undertaking.
The only other option left on the table is using a fan simulator that would trick the controller into thinking it's running a real fan, but given the supposed openness of the platform and the fact that the fan headers are glued to the unit for some reason I would consider such solution as not preferable.
The text was updated successfully, but these errors were encountered:
tmp_off - is the temperature threshold at which the fans will go off
tmp_start - is the temperature threshold at which the fans will start spinning
tmp_full - is the temperature threshold at which the fans will spin at full speed
pwm_start - is the initial duty cycle (%) at which the fan starts spinning when the temperature reaches tmp_start
slope - controls how fast the PWM duty cycle should ramp up in the tmp_start to tmp_full temperature range, from pwm_start duty cycle up to 100% duty cycle
Then simply recompile and flash the Dasharo ROM. If you wish to have setup options for the fan control, please state that this is a feature request from your side and we may pass it to Protectli. But it depends on Protectli whether it will be implemented or not.
As for the AMI firmware, I can't help much. You are on your own, unfortunately.
Component
EC firmware
Device
Protectli VP6630
Dasharo version
No response
Dasharo Tools Suite version
No response
Test case ID
No response
Brief summary
when 0pwm fan settings for IT8659 controller are set the fans still don't fully stop, but also when they barely spin they don't really contribute to cooling perf
How reproducible
Reproducible with stock AMI firmware, permanent. (I assume it also applies to coreboot as the fan settings in the source are almost identical to AMI options)
How to reproduce
Set these fan settings in AMI or corebios (link to source):
And with these settings the fans occasionally spin up to like 100 rpm and then turn off leaving no cooling performance benefit on the table except consuming extra energy and producing noise
Expected behavior
In manual pwm mode when setting is set to 0 then fans should just switch off. Or at least in the fan settings there should be a separate boolean toggle that instructs IT8659 to switch off fan ctl completely.
At first you might think but why not leave these 'sane' defaults on. Well my argument is if the user is allowed to set fan speeds that are so low that barely move a cubic millimeter of air per s contributing nothing to the overall heat dissipation performance especially on the i3 vp66 unit might as well allow users to turn the whole thing off the in first place saving on noise levels + giving user the freedom disconnecting extra moving parts from the system if they intend to.
Actual behavior
Fans spin for like 1 second at rouhgly 100-200 rpm and then switch off and cycle like this every 5 seconds or so
Screenshots
No response
Additional context
My findings mainly come from the AMI Bios variant of the firmware however I suspect this applies to coreboot components in the same vein as well. There are clear references to it8659 in AMI BIOS firmware & corebios. I believe should probably smart enough where it can receive a simple instruction of turning off the fan control completely.
Solutions you've tried
Patching AMI BIOS (and flashing directly on chip just to be extra sure) with tinkered Platform Thermal Configuration values to see if this was linked to anything intel thermals related, or any extra hidden values that at least had some relation to fan/pwm controls but to no avail.
I do suspect the main pwm logic lives on that super io chip and programming that chip esp with no datasheet avail would be a great undertaking.
The only other option left on the table is using a fan simulator that would trick the controller into thinking it's running a real fan, but given the supposed openness of the platform and the fact that the fan headers are glued to the unit for some reason I would consider such solution as not preferable.
The text was updated successfully, but these errors were encountered: