Re: [Intel-gfx] [PATCH 4/4] drm/i915: re-enable rc6 for ironlake
On Wed, Apr 27, 2011 at 03:49:40PM -0700, Jesse Barnes wrote: > On Wed, 27 Apr 2011 23:03:14 +0100 > Chris Wilson wrote: > > > We should probably just enable fbc on the pipe connected to the > > > internal panel (if any) and keep it disabled otherwise. > > > > On Arrandale, fbc is potentially a bigger win than rc6 - but is also a > > potential loss - right? > > No, I think rc6 is still a much bigger win even on Arrandale (iirc >>1W > for rc6, ~0.5W for fbc best case). And yeah, there is the potential for > increased power consumption when fbc is enabled if a framebuffer that > doesn't compress very well continues to get re-compressed over a long > period of time. > > Jesse The last tests that I ran it seemed we needed to be really careful about when to turn on FBC on for Arrandale as noted in the bug that Chris assigned to me below :). I did some rc6 tests prior to the FBC tests and can't find my notes, but I actually recall it being closer to 2W. https://bugs.freedesktop.org/show_bug.cgi?id=31742 Ben ___ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH 4/4] drm/i915: re-enable rc6 for ironlake
On Wed, 27 Apr 2011 23:03:14 +0100 Chris Wilson wrote: > > We should probably just enable fbc on the pipe connected to the > > internal panel (if any) and keep it disabled otherwise. > > On Arrandale, fbc is potentially a bigger win than rc6 - but is also a > potential loss - right? No, I think rc6 is still a much bigger win even on Arrandale (iirc >>1W for rc6, ~0.5W for fbc best case). And yeah, there is the potential for increased power consumption when fbc is enabled if a framebuffer that doesn't compress very well continues to get re-compressed over a long period of time. Jesse ___ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH 4/4] drm/i915: re-enable rc6 for ironlake
On Wed, 27 Apr 2011 14:20:09 -0700, Jesse Barnes wrote: > On Wed, 27 Apr 2011 08:49:28 +0100 > Chris Wilson wrote: > > > On Fri, 18 Mar 2011 16:12:48 -0700, Ben Widawsky wrote: > > > The previous patches should fix enough of the known issues to try > > > re-enabling rc6 for general consumption > > > > Yay, and they bring back the old bugs! :) > > > > So upon enabling fbc on a 2560x1440 external panel, it goes blank again. > > The same symptoms as prompted disabling fbc last year. Can anyone confirm > > this bad behaviour? > > Re-enabling rc6 causes problems when you enable fbc? Yes, I'm pretty sure that's the causal link. Disabling the LVDS makes the external DP go haywire, with fbc enabled for the external, but worked fine until we tried to enable rc6 again. If I remember my bugs correctly, pure fbc failures resulted in a stripy display, whereas this is a solid corruption. And I'm pretty certain this is the same bug that convinced me to disable rc6 the last time. > We should probably just enable fbc on the pipe connected to the > internal panel (if any) and keep it disabled otherwise. On Arrandale, fbc is potentially a bigger win than rc6 - but is also a potential loss - right? -Chris -- Chris Wilson, Intel Open Source Technology Centre ___ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH 4/4] drm/i915: re-enable rc6 for ironlake
On Wed, 27 Apr 2011 08:49:28 +0100 Chris Wilson wrote: > On Fri, 18 Mar 2011 16:12:48 -0700, Ben Widawsky wrote: > > The previous patches should fix enough of the known issues to try > > re-enabling rc6 for general consumption > > Yay, and they bring back the old bugs! :) > > So upon enabling fbc on a 2560x1440 external panel, it goes blank again. > The same symptoms as prompted disabling fbc last year. Can anyone confirm > this bad behaviour? Re-enabling rc6 causes problems when you enable fbc? We should probably just enable fbc on the pipe connected to the internal panel (if any) and keep it disabled otherwise. Jesse ___ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH 4/4] drm/i915: re-enable rc6 for ironlake
On Fri, 18 Mar 2011 16:12:48 -0700, Ben Widawsky wrote: > The previous patches should fix enough of the known issues to try > re-enabling rc6 for general consumption Yay, and they bring back the old bugs! :) So upon enabling fbc on a 2560x1440 external panel, it goes blank again. The same symptoms as prompted disabling fbc last year. Can anyone confirm this bad behaviour? -Chris -- Chris Wilson, Intel Open Source Technology Centre ___ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/intel-gfx
[Intel-gfx] [PATCH 4/4] drm/i915: re-enable rc6 for ironlake
The previous patches should fix enough of the known issues to try re-enabling rc6 for general consumption Signed-off-by: Ben Widawsky --- drivers/gpu/drm/i915/i915_drv.c |2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c index 22ec066..e3c808d 100644 --- a/drivers/gpu/drm/i915/i915_drv.c +++ b/drivers/gpu/drm/i915/i915_drv.c @@ -49,7 +49,7 @@ module_param_named(powersave, i915_powersave, int, 0600); unsigned int i915_semaphores = 0; module_param_named(semaphores, i915_semaphores, int, 0600); -unsigned int i915_enable_rc6 = 0; +unsigned int i915_enable_rc6 = 1; module_param_named(i915_enable_rc6, i915_enable_rc6, int, 0600); unsigned int i915_lvds_downclock = 0; -- 1.7.3.4 ___ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/intel-gfx