From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
To: "Tejun Heo" <tj@kernel.org>, "Michal Koutný" <mkoutny@suse.com>
Cc: Intel-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org,
cgroups@vger.kernel.org, linux-kernel@vger.kernel.org,
"Johannes Weiner" <hannes@cmpxchg.org>,
"Zefan Li" <lizefan.x@bytedance.com>,
"Dave Airlie" <airlied@redhat.com>,
"Daniel Vetter" <daniel.vetter@ffwll.ch>,
"Rob Clark" <robdclark@chromium.org>,
"Stéphane Marchesin" <marcheu@chromium.org>,
"T . J . Mercier" <tjmercier@google.com>,
Kenny.Ho@amd.com, "Christian König" <christian.koenig@amd.com>,
"Brian Welty" <brian.welty@intel.com>,
"Tvrtko Ursulin" <tvrtko.ursulin@intel.com>
Subject: Re: [RFC v3 00/12] DRM scheduling cgroup controller
Date: Thu, 26 Jan 2023 18:14:46 +0000 [thread overview]
Message-ID: <2c14b00c-c3bc-75dd-361b-54c6846fc59a@linux.intel.com> (raw)
In-Reply-To: <b8a0872c-fe86-b174-ca3b-0fc04a98e224@linux.intel.com>
On 26/01/2023 17:57, Tvrtko Ursulin wrote:
> On 26/01/2023 17:04, Tejun Heo wrote:
>> driver folks think about the current RFC tho. Is at least AMD on board
>> with
>> the approach?
>
> Yes I am keenly awaiting comments from the DRM colleagues as well.
Forgot to mention one thing on this point which may interest AMD.
Some time ago I tested the super primitive "throttling via lowering the
scheduling priority" on a GuC based i915 GPU, so only three supported
priority levels, and FWIW it can be somewhat effective.
It certainly was effective for my main use case which is "run this GPU
workload in the background while I use the GPU for something else".
The actual test was along the lines of running a GPU hog in parallel to
an interactive client which can measure dropped frames.
With equal drm.weights the interactive client was seeing ~10 (i915
pre-GuC) or ~27 (i915 GuC) dropped frames per second (60 fps target).
With the GPU hog drm.weight lowered to 1:10 that dropped to ~3 dropped
frames per second (all 3 before the over budget condition was noticed by
the controller).
Main take here is that improved user experience is possible even with
this primitive throttling method and even on GPUs which support only
three scheduling priority levels.
Although main thing still is that individual drivers are completely free
to improve their method of handling to the over budget signal. Nothing
in the controller itself should be precluding that.
Regards,
Tvrtko
WARNING: multiple messages have this Message-ID (diff)
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
To: "Tejun Heo" <tj@kernel.org>, "Michal Koutný" <mkoutny@suse.com>
Cc: "Rob Clark" <robdclark@chromium.org>,
Kenny.Ho@amd.com, "Dave Airlie" <airlied@redhat.com>,
"Stéphane Marchesin" <marcheu@chromium.org>,
"Daniel Vetter" <daniel.vetter@ffwll.ch>,
Intel-gfx@lists.freedesktop.org,
"Brian Welty" <brian.welty@intel.com>,
linux-kernel@vger.kernel.org, dri-devel@lists.freedesktop.org,
"Christian König" <christian.koenig@amd.com>,
"Tvrtko Ursulin" <tvrtko.ursulin@intel.com>,
"Zefan Li" <lizefan.x@bytedance.com>,
"Johannes Weiner" <hannes@cmpxchg.org>,
cgroups@vger.kernel.org, "T . J . Mercier" <tjmercier@google.com>
Subject: Re: [RFC v3 00/12] DRM scheduling cgroup controller
Date: Thu, 26 Jan 2023 18:14:46 +0000 [thread overview]
Message-ID: <2c14b00c-c3bc-75dd-361b-54c6846fc59a@linux.intel.com> (raw)
In-Reply-To: <b8a0872c-fe86-b174-ca3b-0fc04a98e224@linux.intel.com>
On 26/01/2023 17:57, Tvrtko Ursulin wrote:
> On 26/01/2023 17:04, Tejun Heo wrote:
>> driver folks think about the current RFC tho. Is at least AMD on board
>> with
>> the approach?
>
> Yes I am keenly awaiting comments from the DRM colleagues as well.
Forgot to mention one thing on this point which may interest AMD.
Some time ago I tested the super primitive "throttling via lowering the
scheduling priority" on a GuC based i915 GPU, so only three supported
priority levels, and FWIW it can be somewhat effective.
It certainly was effective for my main use case which is "run this GPU
workload in the background while I use the GPU for something else".
The actual test was along the lines of running a GPU hog in parallel to
an interactive client which can measure dropped frames.
With equal drm.weights the interactive client was seeing ~10 (i915
pre-GuC) or ~27 (i915 GuC) dropped frames per second (60 fps target).
With the GPU hog drm.weight lowered to 1:10 that dropped to ~3 dropped
frames per second (all 3 before the over budget condition was noticed by
the controller).
Main take here is that improved user experience is possible even with
this primitive throttling method and even on GPUs which support only
three scheduling priority levels.
Although main thing still is that individual drivers are completely free
to improve their method of handling to the over budget signal. Nothing
in the controller itself should be precluding that.
Regards,
Tvrtko
WARNING: multiple messages have this Message-ID (diff)
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
To: "Tejun Heo" <tj@kernel.org>, "Michal Koutný" <mkoutny@suse.com>
Cc: "Rob Clark" <robdclark@chromium.org>,
Kenny.Ho@amd.com, "Dave Airlie" <airlied@redhat.com>,
"Stéphane Marchesin" <marcheu@chromium.org>,
"Daniel Vetter" <daniel.vetter@ffwll.ch>,
Intel-gfx@lists.freedesktop.org, linux-kernel@vger.kernel.org,
dri-devel@lists.freedesktop.org,
"Christian König" <christian.koenig@amd.com>,
"Zefan Li" <lizefan.x@bytedance.com>,
"Johannes Weiner" <hannes@cmpxchg.org>,
cgroups@vger.kernel.org, "T . J . Mercier" <tjmercier@google.com>
Subject: Re: [Intel-gfx] [RFC v3 00/12] DRM scheduling cgroup controller
Date: Thu, 26 Jan 2023 18:14:46 +0000 [thread overview]
Message-ID: <2c14b00c-c3bc-75dd-361b-54c6846fc59a@linux.intel.com> (raw)
In-Reply-To: <b8a0872c-fe86-b174-ca3b-0fc04a98e224@linux.intel.com>
On 26/01/2023 17:57, Tvrtko Ursulin wrote:
> On 26/01/2023 17:04, Tejun Heo wrote:
>> driver folks think about the current RFC tho. Is at least AMD on board
>> with
>> the approach?
>
> Yes I am keenly awaiting comments from the DRM colleagues as well.
Forgot to mention one thing on this point which may interest AMD.
Some time ago I tested the super primitive "throttling via lowering the
scheduling priority" on a GuC based i915 GPU, so only three supported
priority levels, and FWIW it can be somewhat effective.
It certainly was effective for my main use case which is "run this GPU
workload in the background while I use the GPU for something else".
The actual test was along the lines of running a GPU hog in parallel to
an interactive client which can measure dropped frames.
With equal drm.weights the interactive client was seeing ~10 (i915
pre-GuC) or ~27 (i915 GuC) dropped frames per second (60 fps target).
With the GPU hog drm.weight lowered to 1:10 that dropped to ~3 dropped
frames per second (all 3 before the over budget condition was noticed by
the controller).
Main take here is that improved user experience is possible even with
this primitive throttling method and even on GPUs which support only
three scheduling priority levels.
Although main thing still is that individual drivers are completely free
to improve their method of handling to the over budget signal. Nothing
in the controller itself should be precluding that.
Regards,
Tvrtko
WARNING: multiple messages have this Message-ID (diff)
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
To: "Tejun Heo" <tj@kernel.org>, "Michal Koutný" <mkoutny@suse.com>
Cc: "Rob Clark" <robdclark@chromium.org>,
Kenny.Ho@amd.com, "Dave Airlie" <airlied@redhat.com>,
"Stéphane Marchesin" <marcheu@chromium.org>,
"Daniel Vetter" <daniel.vetter@ffwll.ch>,
Intel-gfx@lists.freedesktop.org, linux-kernel@vger.kernel.org,
dri-devel@lists.freedesktop.org,
"Christian König" <christian.koenig@amd.com>,
"Zefan Li" <lizefan.x@bytedance.com>,
"Johannes Weiner" <hannes@cmpxchg.org>,
cgroups@vger.kernel.org, "T . J . Mercier" <tjmercier@google.com>
Subject: Re: [RFC v3 00/12] DRM scheduling cgroup controller
Date: Thu, 26 Jan 2023 18:14:46 +0000 [thread overview]
Message-ID: <2c14b00c-c3bc-75dd-361b-54c6846fc59a@linux.intel.com> (raw)
In-Reply-To: <b8a0872c-fe86-b174-ca3b-0fc04a98e224@linux.intel.com>
On 26/01/2023 17:57, Tvrtko Ursulin wrote:
> On 26/01/2023 17:04, Tejun Heo wrote:
>> driver folks think about the current RFC tho. Is at least AMD on board
>> with
>> the approach?
>
> Yes I am keenly awaiting comments from the DRM colleagues as well.
Forgot to mention one thing on this point which may interest AMD.
Some time ago I tested the super primitive "throttling via lowering the
scheduling priority" on a GuC based i915 GPU, so only three supported
priority levels, and FWIW it can be somewhat effective.
It certainly was effective for my main use case which is "run this GPU
workload in the background while I use the GPU for something else".
The actual test was along the lines of running a GPU hog in parallel to
an interactive client which can measure dropped frames.
With equal drm.weights the interactive client was seeing ~10 (i915
pre-GuC) or ~27 (i915 GuC) dropped frames per second (60 fps target).
With the GPU hog drm.weight lowered to 1:10 that dropped to ~3 dropped
frames per second (all 3 before the over budget condition was noticed by
the controller).
Main take here is that improved user experience is possible even with
this primitive throttling method and even on GPUs which support only
three scheduling priority levels.
Although main thing still is that individual drivers are completely free
to improve their method of handling to the over budget signal. Nothing
in the controller itself should be precluding that.
Regards,
Tvrtko
next prev parent reply other threads:[~2023-01-26 18:16 UTC|newest]
Thread overview: 126+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-01-12 16:55 [Intel-gfx] [RFC v3 00/12] DRM scheduling cgroup controller Tvrtko Ursulin
2023-01-12 16:55 ` Tvrtko Ursulin
2023-01-12 16:55 ` Tvrtko Ursulin
2023-01-12 16:55 ` Tvrtko Ursulin
2023-01-12 16:55 ` [RFC 01/12] drm: Track clients by tgid and not tid Tvrtko Ursulin
2023-01-12 16:55 ` Tvrtko Ursulin
2023-01-12 16:55 ` Tvrtko Ursulin
2023-01-12 16:55 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-12 16:55 ` [RFC 02/12] drm: Update file owner during use Tvrtko Ursulin
2023-01-12 16:55 ` Tvrtko Ursulin
2023-01-12 16:55 ` Tvrtko Ursulin
2023-01-12 16:55 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-12 16:56 ` [RFC 03/12] cgroup: Add the DRM cgroup controller Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-12 16:56 ` [RFC 04/12] drm/cgroup: Track clients per owning process Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-17 16:03 ` Stanislaw Gruszka
2023-01-17 16:03 ` Stanislaw Gruszka
2023-01-17 16:03 ` [Intel-gfx] " Stanislaw Gruszka
2023-01-17 16:03 ` Stanislaw Gruszka
2023-01-17 16:25 ` Tvrtko Ursulin
2023-01-17 16:25 ` Tvrtko Ursulin
2023-01-17 16:25 ` Tvrtko Ursulin
2023-01-17 16:25 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-12 16:56 ` [RFC 05/12] drm/cgroup: Allow safe external access to file_priv Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-12 16:56 ` [RFC 06/12] drm/cgroup: Add ability to query drm cgroup GPU time Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-12 16:56 ` [RFC 07/12] drm/cgroup: Add over budget signalling callback Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-12 16:56 ` [RFC 08/12] drm/cgroup: Only track clients which are providing drm_cgroup_ops Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-12 16:56 ` [RFC 09/12] cgroup/drm: Client exit hook Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-12 16:56 ` [RFC 10/12] cgroup/drm: Introduce weight based drm cgroup control Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-14 21:20 ` kernel test robot
2023-01-27 13:01 ` Michal Koutný
2023-01-27 13:01 ` Michal Koutný
2023-01-27 13:01 ` [Intel-gfx] " Michal Koutný
2023-01-27 13:01 ` Michal Koutný
2023-01-27 13:31 ` Tvrtko Ursulin
2023-01-27 13:31 ` Tvrtko Ursulin
2023-01-27 13:31 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-27 13:31 ` Tvrtko Ursulin
2023-01-27 14:11 ` Michal Koutný
2023-01-27 14:11 ` [Intel-gfx] " Michal Koutný
2023-01-27 14:11 ` Michal Koutný
2023-01-27 15:21 ` Tvrtko Ursulin
2023-01-27 15:21 ` Tvrtko Ursulin
2023-01-27 15:21 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-27 15:21 ` Tvrtko Ursulin
2023-01-28 1:11 ` Tejun Heo
2023-01-28 1:11 ` Tejun Heo
2023-01-28 1:11 ` [Intel-gfx] " Tejun Heo
2023-01-28 1:11 ` Tejun Heo
2023-02-02 14:26 ` Tvrtko Ursulin
2023-02-02 14:26 ` Tvrtko Ursulin
2023-02-02 14:26 ` [Intel-gfx] " Tvrtko Ursulin
2023-02-02 14:26 ` Tvrtko Ursulin
2023-02-02 20:00 ` Tejun Heo
2023-02-02 20:00 ` Tejun Heo
2023-02-02 20:00 ` [Intel-gfx] " Tejun Heo
2023-02-02 20:00 ` Tejun Heo
2023-01-12 16:56 ` [RFC 11/12] drm/i915: Wire up with drm controller GPU time query Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-12 16:56 ` [RFC 12/12] drm/i915: Implement cgroup controller over budget throttling Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` Tvrtko Ursulin
2023-01-12 16:56 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-12 17:40 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for DRM scheduling cgroup controller (rev3) Patchwork
2023-01-12 18:09 ` [Intel-gfx] ✓ Fi.CI.BAT: success " Patchwork
2023-01-13 6:52 ` [Intel-gfx] ✓ Fi.CI.IGT: " Patchwork
2023-01-23 15:42 ` [RFC v3 00/12] DRM scheduling cgroup controller Michal Koutný
2023-01-23 15:42 ` Michal Koutný
2023-01-23 15:42 ` [Intel-gfx] " Michal Koutný
2023-01-23 15:42 ` Michal Koutný
2023-01-25 18:11 ` Tvrtko Ursulin
2023-01-25 18:11 ` Tvrtko Ursulin
2023-01-25 18:11 ` Tvrtko Ursulin
2023-01-25 18:11 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-26 13:00 ` Michal Koutný
2023-01-26 13:00 ` Michal Koutný
2023-01-26 13:00 ` [Intel-gfx] " Michal Koutný
2023-01-26 13:00 ` Michal Koutný
2023-01-26 17:04 ` Tejun Heo
2023-01-26 17:04 ` Tejun Heo
2023-01-26 17:04 ` [Intel-gfx] " Tejun Heo
2023-01-26 17:04 ` Tejun Heo
2023-01-26 17:57 ` Tvrtko Ursulin
2023-01-26 17:57 ` Tvrtko Ursulin
2023-01-26 17:57 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-26 17:57 ` Tvrtko Ursulin
2023-01-26 18:14 ` Tvrtko Ursulin [this message]
2023-01-26 18:14 ` Tvrtko Ursulin
2023-01-26 18:14 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-26 18:14 ` Tvrtko Ursulin
2023-01-27 10:04 ` Michal Koutný
2023-01-27 10:04 ` Michal Koutný
2023-01-27 10:04 ` [Intel-gfx] " Michal Koutný
2023-01-27 10:04 ` Michal Koutný
2023-01-27 11:40 ` Tvrtko Ursulin
2023-01-27 11:40 ` Tvrtko Ursulin
2023-01-27 11:40 ` Tvrtko Ursulin
2023-01-27 11:40 ` [Intel-gfx] " Tvrtko Ursulin
2023-01-27 13:00 ` Michal Koutný
2023-01-27 13:00 ` [Intel-gfx] " Michal Koutný
2023-01-27 13:00 ` Michal Koutný
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=2c14b00c-c3bc-75dd-361b-54c6846fc59a@linux.intel.com \
--to=tvrtko.ursulin@linux.intel.com \
--cc=Intel-gfx@lists.freedesktop.org \
--cc=Kenny.Ho@amd.com \
--cc=airlied@redhat.com \
--cc=brian.welty@intel.com \
--cc=cgroups@vger.kernel.org \
--cc=christian.koenig@amd.com \
--cc=daniel.vetter@ffwll.ch \
--cc=dri-devel@lists.freedesktop.org \
--cc=hannes@cmpxchg.org \
--cc=linux-kernel@vger.kernel.org \
--cc=lizefan.x@bytedance.com \
--cc=marcheu@chromium.org \
--cc=mkoutny@suse.com \
--cc=robdclark@chromium.org \
--cc=tj@kernel.org \
--cc=tjmercier@google.com \
--cc=tvrtko.ursulin@intel.com \
/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.