Family 17h CPUs use a slightly different frequency formula than previous
generation models.
PR: 221621
Details
Details
Diff Detail
Diff Detail
- Lint
Lint Passed - Unit
No Test Coverage - Build Status
Buildable 11167 Build 11549: arc lint + arc unit
Event Timeline
Comment Actions
So I will note that on my system hwpstate ends up opting to get these from acpi_perf:
kernel: hwpstate0: going to fetch info from acpi_perf ... mjoras@icarium ~> sysctl dev.hwpstate.0.freq_settings dev.hwpstate.0.freq_settings: 3400/3825 2800/2765 2200/1952
Forcing it to come from an MSR and using your patch I get:
dev.hwpstate.0.freq_settings: 3400/-1 2800/-1 2200/-1
So that looks good. The power dissipation is unknown, but apparently available when using ACPI. This is the case for older families too. powerd(8) only uses the power information in verbose mode to track total energy usage, apparently.
This patch doesn't fix the PR, as already noted there. It seems like this would in fact fix powerd(8) for the case where getting the information from ACPI fails, so I might mention that this fixes it for the MSR case specifically in the title.