Hi Nils, Nils Goroll wrote: > Hi Anup, > > thanks for your explanations and for pointing out that cpu-threshold values > below > the cpu usage scan rate do not make sense. > > On a wider perspective, I would like to ask if it is possible to find a more > intuitive > integration with gnome power manager, once you re-visit the code anyway due > to the change of the cpu usage scan rate. > > If I have understood this correctly, the cpu-threshold is defined as the > time (a/the) > cpu needs to be "idle" (have not looked up how exactly idleness is defined > for PM, > can't be "full idle") in the current power level until the next lower power > level is > entered. > > With the current implementation, the user configures "Computer speed policy > based on processor load", but with the default setting yielding cpu-threshold > 1275, a system would need to be "idle" for 21 minutes before starting to save > power. > You are right. > I doubt whether this is what users expect, > so I would like to suggest to > > - either change the default of at least cpufreq/performance_battery to 1 > - or to transparently convey the basis of the internal implementation to the > user by introducing a GUI-configurable parameter like "start to save power > after this time idle" > We are discussing with the GPM team about this issue and get the cpu-threshold changed only when user explicitly changes it. Also will fix the performance to cpu-threshold calculation part. > Also, If I understand the goals of project Tesla correctly, the overall goal > is > to define objectives which are then to be achieved by power management, > so all of this can / will(?) become obsolete again, right? > Yes, that is one of the goals of Tesla. The current cpufreq GPM support is a interim support till we finish our current restructuring of CPU power management. We are working as part of tesla project a complete revamp of CPU power management. Will keep the alias posted. Also keep looking at the Tesla project page about new features and progress.
Really appreciate your inputs. Thanks Anup > What do you think? > > Nils > -- > This message posted from opensolaris.org > _______________________________________________ > pm-discuss mailing list > pm-discuss at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/pm-discuss > -- Anup Pemmaiah Sun Microsystems
