All of lore.kernel.org
 help / color / mirror / Atom feed
From: Chris Wilson <chris@chris-wilson.co.uk>
To: intel-gfx@lists.freedesktop.org
Subject: Re: [PATCH 29/33] drm/i915: Implement an "idle" barrier
Date: Fri, 25 Jan 2019 08:43:38 +0000	[thread overview]
Message-ID: <154840581816.2593.16496649224827780481@skylake-alporthouse-com> (raw)
In-Reply-To: <20190125023005.1007-29-chris@chris-wilson.co.uk>

Quoting Chris Wilson (2019-01-25 02:30:01)
> We have a number of tasks that we like to run when idle and parking the
> GPU into a powersaving mode. A few of those tasks are using the global
> idle point as a convenient moment when all previous execution has been
> required (and so we know that the GPU is not still touching random
> user memory). However, on a busy system we are unlikely to see global
> idle points, and would prefer a much more incremental system of being
> able to retire after all current execution has completed.
> 
> Enter the idle barrier and idle tasks.
> 
> To determine a point in the future when all current tasks are complete,
> we schedule a new low priority request that will be executed after all
> current work is complete, and by imposing a barrier before all future
> work. We therefore know we retire that barrier, the GPU is no longer
> touching any memory released before the barrier was submitting allowing
> us to run a set of idle tasks clear of any dangling GPU references.

The biggest downside is that this does impose a global barrier onto the
system, a point which PI cannot pass.

After sleeping on it, I think I prefer not to take this approach, as
interesting as it is, just yet (for HW semaphores, we can just keep track
of the read references).
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

  reply	other threads:[~2019-01-25  8:43 UTC|newest]

Thread overview: 50+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-01-25  2:29 [PATCH 01/33] drm/i915/execlists: Move RPCS setup to context pin Chris Wilson
2019-01-25  2:29 ` [PATCH 02/33] drm/i915: Measure the required reserved size for request emission Chris Wilson
2019-01-25  8:34   ` Mika Kuoppala
2019-01-25  9:52     ` Chris Wilson
2019-01-25  2:29 ` [PATCH 03/33] drm/i915: Remove manual breadcumb counting Chris Wilson
2019-01-25  9:21   ` Mika Kuoppala
2019-01-25  2:29 ` [PATCH 04/33] drm/i915: Compute the HWS offsets explicitly Chris Wilson
2019-01-25  9:26   ` Mika Kuoppala
2019-01-25  2:29 ` [PATCH 05/33] drm/i915/execlists: Suppress preempting self Chris Wilson
2019-01-25  2:29 ` [PATCH 06/33] drm/i915/execlists: Suppress redundant preemption Chris Wilson
2019-01-25  2:29 ` [PATCH 07/33] drm/i915/selftests: Apply a subtest filter Chris Wilson
2019-01-25 11:44   ` Mika Kuoppala
2019-01-25 11:48     ` Chris Wilson
2019-01-29 10:37   ` Joonas Lahtinen
2019-01-25  2:29 ` [PATCH 08/33] drm/i915: Make all GPU resets atomic Chris Wilson
2019-01-25  2:29 ` [PATCH 09/33] drm/i915/guc: Disable global reset Chris Wilson
2019-01-25  2:29 ` [PATCH 10/33] drm/i915: Remove GPU reset dependence on struct_mutex Chris Wilson
2019-01-25 12:50   ` Mika Kuoppala
2019-01-25  2:29 ` [PATCH 11/33] drm/i915/selftests: Trim struct_mutex duration for set-wedged selftest Chris Wilson
2019-01-25  2:29 ` [PATCH 12/33] drm/i915: Issue engine resets onto idle engines Chris Wilson
2019-01-25  2:29 ` [PATCH 13/33] drm/i915: Stop tracking MRU activity on VMA Chris Wilson
2019-01-25  2:29 ` [PATCH 14/33] drm/i915: Pull VM lists under the VM mutex Chris Wilson
2019-01-25  2:29 ` [PATCH 15/33] drm/i915: Move vma lookup to its own lock Chris Wilson
2019-01-25  2:29 ` [PATCH 16/33] drm/i915: Always allocate an object/vma for the HWSP Chris Wilson
2019-01-25  2:29 ` [PATCH 17/33] drm/i915: Add timeline barrier support Chris Wilson
2019-01-25  2:29 ` [PATCH 18/33] drm/i915: Move list of timelines under its own lock Chris Wilson
2019-01-25  2:29 ` [PATCH 19/33] drm/i915: Introduce concept of per-timeline (context) HWSP Chris Wilson
2019-01-25  2:29 ` [PATCH 20/33] drm/i915: Enlarge vma->pin_count Chris Wilson
2019-01-25  2:29 ` [PATCH 21/33] drm/i915: Allocate a status page for each timeline Chris Wilson
2019-01-25  2:29 ` [PATCH 22/33] drm/i915: Share per-timeline HWSP using a slab suballocator Chris Wilson
2019-01-25  2:29 ` [PATCH 23/33] drm/i915: Track the context's seqno in its own timeline HWSP Chris Wilson
2019-01-25  2:29 ` [PATCH 24/33] drm/i915: Track active timelines Chris Wilson
2019-01-25  2:29 ` [PATCH 25/33] drm/i915: Identify active requests Chris Wilson
2019-01-25  2:29 ` [PATCH 26/33] drm/i915: Remove the intel_engine_notify tracepoint Chris Wilson
2019-01-25 14:10   ` Tvrtko Ursulin
2019-01-25  2:29 ` [PATCH 27/33] drm/i915: Replace global breadcrumbs with per-context interrupt tracking Chris Wilson
2019-01-25 13:54   ` Tvrtko Ursulin
2019-01-25 14:26     ` Chris Wilson
2019-01-25 14:39       ` Chris Wilson
2019-01-25  2:30 ` [PATCH 28/33] drm/i915: Drop fake breadcrumb irq Chris Wilson
2019-01-25 11:07   ` Tvrtko Ursulin
2019-01-25  2:30 ` [PATCH 29/33] drm/i915: Implement an "idle" barrier Chris Wilson
2019-01-25  8:43   ` Chris Wilson [this message]
2019-01-25  2:30 ` [PATCH 30/33] drm/i915: Keep timeline HWSP allocated until the system is idle Chris Wilson
2019-01-25  2:30 ` [PATCH 31/33] drm/i915/execlists: Refactor out can_merge_rq() Chris Wilson
2019-01-25  2:30 ` [PATCH 32/33] drm/i915: Use HW semaphores for inter-engine synchronisation on gen8+ Chris Wilson
2019-01-25  2:30 ` [PATCH 33/33] drm/i915: Prioritise non-busywait semaphore workloads Chris Wilson
2019-01-25  3:13 ` ✗ Fi.CI.CHECKPATCH: warning for series starting with [01/33] drm/i915/execlists: Move RPCS setup to context pin Patchwork
2019-01-25  3:27 ` ✗ Fi.CI.SPARSE: " Patchwork
2019-01-25  3:34 ` ✗ Fi.CI.BAT: failure " Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=154840581816.2593.16496649224827780481@skylake-alporthouse-com \
    --to=chris@chris-wilson.co.uk \
    --cc=intel-gfx@lists.freedesktop.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.