From: Jesse Barnes Subject: Re: [PATCH v3] drm/i915/bdw: BDW Software Turbo Date: Thu, 14 Aug 2014 12:37:53 -0700 Message-ID: <20140814123753.59fc9e63@jbarnes-desktop> References: <1407780518-26610-1-git-send-email-daisy.sun@intel.com> <20140811213357.GH10500@phenom.ffwll.local> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Received: from mail-pa0-f41.google.com (mail-pa0-f41.google.com [209.85.220.41]) by gabe.freedesktop.org (Postfix) with ESMTP id A42B76E73D for ; Thu, 14 Aug 2014 12:37:24 -0700 (PDT) Received: by mail-pa0-f41.google.com with SMTP id rd3so2192275pab.28 for ; Thu, 14 Aug 2014 12:37:24 -0700 (PDT) In-Reply-To: <20140811213357.GH10500@phenom.ffwll.local> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" To: Daniel Vetter Cc: intel-gfx@lists.freedesktop.org List-Id: intel-gfx@lists.freedesktop.org On Mon, 11 Aug 2014 23:33:57 +0200 Daniel Vetter wrote: > On Mon, Aug 11, 2014 at 11:08:38AM -0700, Daisy Sun wrote: > > BDW supports GT C0 residency reporting in constant time unit. Driver > > calculates GT utilization based on C0 residency and adjusts RP > > frequency up/down accordingly. For offscreen workload specificly, > > set frequency to RP0. > > > > Offscreen task is not restricted by frame rate, it can be > > executed as soon as possible. Transcoding and serilized workload > > between CPU and GPU both need high GT performance, RP0 is a good > > option in this case. RC6 will kick in to compensate power > > consumption when GT is not active. > > > > v2: Rebase on recent drm-intel-nightly > > v3: Add flip timerout monitor, when no flip is deteced within > > 100ms, set frequency to RP0. > > Ok, let's make this really clear: > > If you wire this into the flip handling in any way, I will not merge your > patch. The timer should be fully independant and tie into the gpu > busy/idle handling we already have. Sounds like Daisy won't be able to spend any more time on this either. So we're left with this patch, which does improve things for most cases, or no patch, which leaves things universally bad. Unless someone wants to pick up the additional work and testing of using a timer scheme, making sure we don't have needless wakeups, and generally improve power/perf across even more cases than this patch. -- Jesse Barnes, Intel Open Source Technology Center