[Intel-gfx] [PATCH 1/3] drm/i915/gt: Prefer soft-rc6 over RPS DOWN_TIMEOUT
Chris Wilson
chris at chris-wilson.co.uk
Fri Apr 24 13:37:31 UTC 2020
Quoting Andi Shyti (2020-04-24 13:50:40)
> Hi Chris,
>
> On Wed, Apr 22, 2020 at 01:17:01AM +0100, Chris Wilson wrote:
> > The RPS DOWN_TIMEOUT interrupt is signaled after a period of rc6, and
> > upon receipt of that interrupt we reprogram the GPU clocks down to the
> > next idle notch [to help convserve power during rc6]. However, on
> > execlists, we benefit from soft-rc6 immediately parking the GPU and
> > setting idle frequencies upon idling [within a jiffie], and here the
> > interrupt prevents us from restarting from our last frequency.
> >
> > In the process, we can simply opt for a static pm_events mask and rely
> > on the enable/disable interrupts to flush the worker on parking.
> >
> > This will reduce the amount of oscillation observed during steady
> > workloads with microsleeps, as each time the rc6 timeout occurs we
> > immediately follow with a waitboost for a dropped frame.
> >
> > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > ---
> > drivers/gpu/drm/i915/gt/intel_rps.c | 41 +++++++++++++----------------
> > 1 file changed, 18 insertions(+), 23 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/gt/intel_rps.c b/drivers/gpu/drm/i915/gt/intel_rps.c
> > index 4dcfae16a7ce..785cd58fba76 100644
> > --- a/drivers/gpu/drm/i915/gt/intel_rps.c
> > +++ b/drivers/gpu/drm/i915/gt/intel_rps.c
> > @@ -57,7 +57,7 @@ static u32 rps_pm_mask(struct intel_rps *rps, u8 val)
> > if (val < rps->max_freq_softlimit)
> > mask |= GEN6_PM_RP_UP_EI_EXPIRED | GEN6_PM_RP_UP_THRESHOLD;
> >
> > - mask &= READ_ONCE(rps->pm_events);
> > + mask &= rps->pm_events;
>
> we are giving up the read/write ordering here because we thing
> that removing the down interval we won't have anymore
> inconsistent reads?
Correct. Since we are now only setting rps->pm_events during
initialisation, it remains constant at runtime, and we do not need to
worry about read tearing or repeated reads.
-Chris
More information about the Intel-gfx
mailing list