[OLPC-devel] Power Management?

[Jim Gettys]

Our biggest challenge is doing suspend/resume very, very fast.  The
Geode is very slow getting off the mark (150ms), so we can't go faster
than that :=(.

As soon as we get a scope into the hardware lab we're setting up, I plan
to look at how fast a iPAQ suspends and resume.  It's pretty quick, but
exactly how quick it is needs more serious investigation.  It was taking
a while to suspend: this turned out to be a sleep(1) in apmd!  Right
now, it is hard to see how long it takes to resume; turning the screen
back on is done in late in the process; in our case, it will happen
instantly, even before the CPU gets power again.  In any case, I know
Linux is well under a second (on non-brain-damaged hardware).  VESA is
braindead; this is part of why we want to go to using fbdev (along with
being able to punt VESA emulation in VSA).

If you look at the power management summit report
http://lwn.net/Articles/181687/ you'll also get a link to a scheme Mark
Foster has come up with that parallelizes resume (it is not unusual for
drivers to have delays in them due to hardware requiring time to
initialize.

So if measurements show we need to go that way, we will.  There was
great interest in Mark's idea at the PM summit; it got quite a warm
reception.
                                        - Jim

On Mon, 2006-05-15 at 15:41 -0500, Goodrich,Steven wrote:
> Ron raised a good point WRT the 30 part-time student developers: Power
> Management.  Has the technical/architectural design for OLPC's PM
> begun yet?
>  
>     -- Steve G.
> 
> Steve Goodrich (Steven.Goodrich at amd.com)
> Advanced Micro Devices, Inc.
> 1351 S. Sunset Street
> Mail Stop 83-36
> Longmont, CO  80501
> 303-774-5129    (fax: 303-774-5801)
> http://www.amd.com/embeddedprocessors 
>  
> _______________________________________________
> Devel mailing list
> Devel at laptop.org
> http://mailman.laptop.org/mailman/listinfo/devel
-- 
Jim Gettys
One Laptop Per Child