Voltage/power domain framework?

Voltage/power domain framework?

[Ikhwan Lee]

I've seen quite a few discussions mentioning the necessity of voltage/power
domain framework. Is there any project going on? I heard that there were
some activities by Nokia, but I could not find any on the web.

Ikhwan

Re: Voltage/power domain framework?

[Mark Gross]

On Fri, Aug 25, 2006 at 04:50:42PM +0900, Ikhwan Lee wrote:
> I've seen quite a few discussions mentioning the necessity of voltage/power
> domain framework. Is there any project going on? I heard that there were
> some activities by Nokia, but I could not find any on the web.
> 

There isn't anything formal going on.  However; there is a loose
collection of folks working on the "operating point" patches with an eye
to defining a lower level interface to enable voltage/power/clock domain
framework's.  

This activity is fall out from last April's PM-summit
http://lwn.net/Articles/181888/, one can also infer the interested
parties from the mail activity on this list.

Do you have some design ideas, patches,  or requirements in this area that could be
taken into account?

--mgross

Re: Voltage/power domain framework?

[Paul Mundt]

On Fri, Aug 25, 2006 at 04:50:42PM +0900, Ikhwan Lee wrote:
> I've seen quite a few discussions mentioning the necessity of voltage/power
> domain framework. Is there any project going on? I heard that there were
> some activities by Nokia, but I could not find any on the web.
> 
Yes, Nokia did some work on OMAP for this that could probably be
generalized. It doesn't seem like it's been posted anywhere yet though,
Sampsa?

Re: Voltage/power domain framework?

[Ikhwan Lee]

Thanks for your note,

> > I've seen quite a few discussions mentioning the necessity of
voltage/power
> > domain framework. Is there any project going on? I heard that there were
> > some activities by Nokia, but I could not find any on the web.
>
> There isn't anything formal going on.  However; there is a loose
> collection of folks working on the "operating point" patches with an eye
> to defining a lower level interface to enable voltage/power/clock domain
> framework's.

I have been following the threads on both PowerOp and OpPoint; however, I
never thought they would do the job of voltage/power/clock domain
frameworks. The way I interpreted is that they _need_ the domain frameworks
so that they could handle lower-level physics better.

> Do you have some design ideas, patches,  or requirements in this area that
could be
> taken into account?

I have requirements and some very rough ideas. And I was searching for an
existing implementation or a discussion so that I can find a place to start
with. Maybe we can start by writing down the requirements. As far as I know,
the CELF PM Requirement document does not have a section for voltage/power
domain frameworks.

Regards,
Ikhwan

Re: Voltage/power domain framework?

[Eugeny S. Mints]

2006/8/27, Ikhwan Lee <ikhwan.lee@samsung.com>:
> Thanks for your note,
>
> > > I've seen quite a few discussions mentioning the necessity of
> voltage/power
> > > domain framework. Is there any project going on? I heard that there were
> > > some activities by Nokia, but I could not find any on the web.
> >
> > There isn't anything formal going on.  However; there is a loose
> > collection of folks working on the "operating point" patches with an eye
> > to defining a lower level interface to enable voltage/power/clock domain
> > framework's.
>
> I have been following the threads on both PowerOp and OpPoint; however, I
> never thought they would do the job of voltage/power/clock domain
> frameworks. The way I interpreted is that they _need_ the domain frameworks
> so that they could handle lower-level physics better.
OpPoint does not have any integration with voltage/clock framework
idea what is completely unaccepted considering the announced goals for
OpPoint.

Yes, clock/voltage framework is separate entity and PowerOP does not
do the job of clock/voltage framewrok because PowerOP is completely
different layer. But PowerOP is only a part of PM framework Matt and I
are proposing and PowerOP leverages clock/voltage framework as the
lowest layer in the PM framework - PowerOP sits on top of PM Core
while PM Core interacts directly with clock/voltage framework (please
see PM framework architecture picture overview in my patch set
postings).

Clock framework is implemented for omap1/2 arches and is utilized by
PM Core for omap in my previous posts of PowerOP.

Thanks,
Eugeny
> > Do you have some design ideas, patches,  or requirements in this area that
> could be
> > taken into account?
>
> I have requirements and some very rough ideas. And I was searching for an
> existing implementation or a discussion so that I can find a place to start
> with. Maybe we can start by writing down the requirements. As far as I know,
> the CELF PM Requirement document does not have a section for voltage/power
> domain frameworks.
>
> Regards,
> Ikhwan
>
>
> _______________________________________________
> linux-pm mailing list
> linux-pm@lists.osdl.org
> https://lists.osdl.org/mailman/listinfo/linux-pm
>

Re: Voltage/power domain framework?

[Mark Gross]

On Sun, Aug 27, 2006 at 08:48:01PM +0900, Ikhwan Lee wrote:
> Thanks for your note,
> 
> > > I've seen quite a few discussions mentioning the necessity of
> voltage/power
> > > domain framework. Is there any project going on? I heard that there were
> > > some activities by Nokia, but I could not find any on the web.
> >
> > There isn't anything formal going on.  However; there is a loose
> > collection of folks working on the "operating point" patches with an eye
> > to defining a lower level interface to enable voltage/power/clock domain
> > framework's.
> 
> I have been following the threads on both PowerOp and OpPoint; however, I
> never thought they would do the job of voltage/power/clock domain
> frameworks. The way I interpreted is that they _need_ the domain frameworks
> so that they could handle lower-level physics better.
> 
> > Do you have some design ideas, patches,  or requirements in this area that
> could be
> > taken into account?
> 
> I have requirements and some very rough ideas. And I was searching for an
> existing implementation or a discussion so that I can find a place to start
> with. Maybe we can start by writing down the requirements. As far as I know,
> the CELF PM Requirement document does not have a section for voltage/power
> domain frameworks.

If you provide me with content then I would be happy to add such a
section or welcome you to edit the public wiki yourself ;)

--mgross