Port-amd64 archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]


On Tue, Feb 22, 2011 at 08:53:54AM +0200, Jukka Ruohonen wrote:
> Hi.
> I would like to remove the options(4) and the code that implements
> EST_FREQ_USERWRITE and INTEL_ONDEMAND_CLOCKMOD. But is anyone actually using
> these options?
> Both are commented out by default in GENERICs and both are of questionable
> value. Especially the latter, "T-states" in ACPI jargon (see acpicpu(4) for
> a short description), should not be something that users and system
> administrators should "tune" themselves. I also have doubts about the
> robustness of INTEL_ONDEMAND_CLOCKMOD on new CPUs.
> The rationale is that I am moving things to the newly added "cpufeaturebus",
> so that options(4) are not used directly. This provides modularity and will
> allow safe fallback to ENHANCED_SPEEDSTEP and POWERNOW_K8 if problems are
> detected with acpicpu(4). But I wouldn't want to unnecessarily add entries to
> cpu_info because of something with dubious value -- and no users?

Just a data point, some laptop CPUs (Celeron M 350 for instance) have no
power management other than clock modulation.  (Not that that laptop is
still working, but whatever.)

        Jonathan Kollasch

Home | Main Index | Thread Index | Old Index