From mboxrd@z Thu Jan  1 00:00:00 1970
From: Chris Wilson <chris@chris-wilson.co.uk>
Subject: Re: [PATCH] drm/i915: Don't wait for vblank for sprite
 plane flips
Date: Fri, 28 Jun 2013 17:05:18 +0100
Message-ID: <20130628160518.GD32183@cantiga.alporthouse.com>
References: <1372428931-24144-1-git-send-email-vijay.a.purushothaman@intel.com>
	<20130628142450.GH5004@intel.com>
Mime-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Return-path: <intel-gfx-bounces+gcfxdi-intel-gfx=m.gmane.org@lists.freedesktop.org>
Received: from fireflyinternet.com (s16502780.onlinehome-server.info
	[87.106.93.118])
	by gabe.freedesktop.org (Postfix) with ESMTP id 98018E66CC
	for <intel-gfx@lists.freedesktop.org>;
	Fri, 28 Jun 2013 09:05:22 -0700 (PDT)
Content-Disposition: inline
In-Reply-To: <20130628142450.GH5004@intel.com>
List-Unsubscribe: <http://lists.freedesktop.org/mailman/options/intel-gfx>,
	<mailto:intel-gfx-request@lists.freedesktop.org?subject=unsubscribe>
List-Archive: <http://lists.freedesktop.org/archives/intel-gfx>
List-Post: <mailto:intel-gfx@lists.freedesktop.org>
List-Help: <mailto:intel-gfx-request@lists.freedesktop.org?subject=help>
List-Subscribe: <http://lists.freedesktop.org/mailman/listinfo/intel-gfx>,
	<mailto:intel-gfx-request@lists.freedesktop.org?subject=subscribe>
Sender: intel-gfx-bounces+gcfxdi-intel-gfx=m.gmane.org@lists.freedesktop.org
Errors-To: intel-gfx-bounces+gcfxdi-intel-gfx=m.gmane.org@lists.freedesktop.org
To: Ville =?iso-8859-1?Q?Syrj=E4l=E4?= <ville.syrjala@linux.intel.com>
Cc: Intel Graphics <intel-gfx@lists.freedesktop.org>
List-Id: intel-gfx@lists.freedesktop.org

On Fri, Jun 28, 2013 at 05:24:50PM +0300, Ville Syrj=E4l=E4 wrote:
> On Fri, Jun 28, 2013 at 07:45:31PM +0530, Vijay Purushothaman wrote:
> > Since the sprite planes are using synchronized MMIO based flip, no need
> > to wait for vblank. Removing this wait allows us to get a nice
> > performance boost to both 3D & media workloads based on sprite (~60 fps
> > from ~20 fps)
> =

> Nak. We can't unpin the buffer until the hardware has finished reading
> from it.
> =

> The proper fix is to do the unpin asynchronously after the flip has
> completed. That's one part of the bigger atomic pageflip story.

The interested reader is invited to review the patches to do async
unpinning here and in set-base sent many, many moons ago.
-Chris

-- =

Chris Wilson, Intel Open Source Technology Centre