Re: [PATCH v2 0/3] EM / PM: Inefficient OPPs

[Lukasz Luba]

Hi Viresh,

On 5/26/21 10:33 AM, Viresh Kumar wrote:
> On 26-05-21, 09:56, Lukasz Luba wrote:
> > No, these OPPs have to stay because they are used in thermal for cooling
> > states.
>
> This won't break the thermal tables. Thermal just sets the max-freq for a CPU,
> and it doesn't depend on the OPP table for that.
>
> > DT cooling devices might have them set as a scope of possible
> > states. We don't want to break existing platforms, don't we?
>
> I don't think we will end up breaking anything here.
>
> > We want to 'avoid' those OPPs when possible (no thermal pressure), but
> > we might have to use them sometimes.
>
> Why would we want to use them if they are inefficient ? Thermal or something
> else as well ?
>
> More in the other reply I am sending to Vincent.

I have responded to your email there. I don't know if you have seen it.
As I said there, these OPPs, which from energy perspective we call
'inefficient', might be used to provide enough performance under thermal
constraints.

Regards,
Lukasz