[Intel-gfx] [PATCH v2 1/3] drm/i915: add rc6 residency times to debugfs

Ben Widawsky ben at bwidawsk.net
Mon Mar 26 18:49:42 CEST 2012


On Mon, 26 Mar 2012 09:33:51 -0300
Eugeni Dodonov <eugeni at dodonov.net> wrote:

> On Mon, Mar 26, 2012 at 06:34, Daniel Vetter <daniel at ffwll.ch> wrote:
> 
> > On Sun, Mar 25, 2012 at 05:33:27PM -0700, Ben Widawsky wrote:
> > > RC6 residency should be in intervals of 1.28us, and the counter
> > > wraps. Here is an example using awk to get the various RC6 and
> > > RC6+ residency times in seconds, since boot.
> > >
> > > cat /sys/kernel/debug/dri/0/i915_drpc_info  | grep residency |
> > > awk -F':'
> > -F' '  '{print $5 * 1.28 / 1000000}'
> > >
> > > This is primarily for debug, and QA/application developers using
> > > the sysfs interface looking for more insight.
> > >
> > > Untested on IVB.
> >
> > Tested on my ivb, seems to yield sane numbers for the default
> > configuration (i.e. enable rc+rc6p, disable rc6pp).
> >
> 
> Yeah, it should work on IVB except for some strange behavior
> sometimes (I noticed that on IVB we sometimes get into RC6 even if we
> pass the i915_enable_rc6=0 parameter). I don't know if it is Bios
> playing some tricks with us or something changed in the mailbox
> communication, but no harm came from this so far. I'll see what is
> going on there.
> 
> But for the patch itself:
> Reviewed-by: Eugeni Dodonov <eugeni.dodonov at gmail.com>
> 
> Nothing like new and shiny powertop power-saving options :).
> 


Well as I said... at least on SNB it seems the RC6 value is never 0.
But I suspect this isn't because we go into RC6, but rather just a
uninitialized register value. Probably would be nice to follow up with
the designers on this one.



More information about the Intel-gfx mailing list