This CL is part of the Call-Level Adaptation Processing design doc:
https://docs.google.com/document/d/1ZyC26yOCknrrcYa839ZWLxD6o6Gig5A3lVTh4E41074/edit?usp=sharing
The POC was used to demo the new design but was never used outside of
unit tests. The plan being to iteratively improve the
ResourceAdaptationProcessor rather than to replace it, we delete the POC
classes to avoid bloat and conflicts.
Bug: webrtc:11172
Change-Id: Ic49afdc471d2d774541f8ef3316b4c6d0a9b8a14
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/172923
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Reviewed-by: Evan Shrubsole <eshr@google.com>
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#31094}
This CL introduces the Adaptation class used by VideoStreamRestrictor.
This refactors the AdaptationTarget, AdaptationTargetOrReason,
CannotAdaptReason and AdaptationAction.
What is publicly exposed is simply a Status code. If it's kValid then
we can adapt, otherwise the status code describes why we can't adapt
(just like CannotAdaptReason prior to this CL). This means
AdaptationTargetOrReason is no longer needed. Target+reason are merged.
The other classes are renamed and moved and put in the private
namespace of Adaptation: Only the VideoStreamAdapter (now a friend
class of Adaptation) and its inner class VideoSourceRestrictor needs to
know how to execute the adaptation.
Publicly, you can now tell the effects of the adaptation without
applying it with PeekNextRestrictions() - both current and next steps
are described in terms of VideoSourceRestrictions. The rest are hidden.
This would make it possible, in the future, for a Resource to accept or
reject a proposed Adaptation by examining the resulting frame rate and
resolution described by the resulting restrictions. E.g. even if we are
not overusing bandwidth at the moment, the BW resource can prevent us
from applying a restriction that would exceed the BW limit before we
apply it.
This CL also moves input to a SetInput() method, and Increase/Decrease
methods of VideoSourceRestrictor are made private in favor of
ApplyAdaptationSteps().
Bug: webrtc:11393
Change-Id: Ie5e2181836ab3713b8021c1a152694ca745aeb0d
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/170111
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Reviewed-by: Evan Shrubsole <eshr@google.com>
Cr-Commit-Position: refs/heads/master@{#30794}
This CL moves GetAdaptUpTarget(), GetAdaptDownTarget() and
ApplyAdaptationTarget() - and related code - to the VideoStreamAdapter.
This includes pieces related to calculating how to adapt, including:
- DegradationPreference
- BalancedDegradationPreference
- AdaptationRequest and last_adaptation_request_
- CanAdaptUpResolution()
The VideoStreamAdapter's interface has changed: VideoSourceRestrictor
methods are now hidden in favor of methods exposing AdaptationTarget.
This CL also does some misc moves:
- GetEncoderBitrateLimits is moved and renamed to
VideoEncoder::EncoderInfo::GetEncoderBitrateLimitsForResolution.
- EncoderSettings moved to a separate file.
// For api/video_codecs/video_encoder.[cc/h] changes, which is the
// moving of a function.
TBR=sprang@webrtc.org
Bug: webrtc:11393
Change-Id: Ie6bd8ef644ce927d7eca6ab90a0a7bcace682f3c
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/169842
Reviewed-by: Henrik Boström <hbos@webrtc.org>
Reviewed-by: Evan Shrubsole <eshr@google.com>
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30708}
This CL simply moves the VideoSourceRestrictor from being an inner class
of OveruseFrameDetectorResourceAdaptationModule to a new class,
VideoStreamAdapter.
In follow-up CLs, the responsibility of determining what the next step
for adapting up or down should also be moved to the VideoStreamAdapter.
The end-goal is that the VideoStreamAdapter takes care of "can adapt?"
and "do adapt!" type of logic so that a multi-stream aware adaptation
module can decide which stream (adapter) to adapt, and the adapter can
take care of the nitty gritty details of doing so.
In this CL the "can?"/"do!" part is realized but not the logic for
determining what the next step up or down is, and the class interface
needs improvement.
This CL also sets up the video/adaptation/ subdirectory and moves the
AdaptationCounters class here. Other adaptation-related classes (e.g.
the module and its resources) should move into this directory as well
in the future.
Bug: webrtc:11393
Change-Id: I2c12c1281eca854c62791abb65f0aca47a119726
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/169542
Reviewed-by: Evan Shrubsole <eshr@google.com>
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Reviewed-by: Henrik Andreassson <henrika@webrtc.org>
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30705}
* This replaces the video stream methods for forcing adaptation
with a mock resource that triggers overuse.
* Resources can now be injected to the Module using the AddResource
function.
* Resources now have tests for adding and removing callbacks.
* Quality/EncoderUse% resources are tracked in the Resource list of
the adaptation module.
* The adaptation module ties all resources to a reason to keep stats
working as expected.
BUG=webrtc:11377
Change-Id: I1f5902f7416dc41b4915c0072e6f0da2bb3bb2b7
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/168948
Commit-Queue: Evan Shrubsole <eshr@google.com>
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Reviewed-by: Henrik Boström <hbos@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30610}
Adapting up or down is currently a "Maybe Adapt" method. In the future
we will want to be able to decide which stream to adapt, and as such we
need to be able to tell if a stream is able to do so.
This takes us one step in that direction, by refactoring
OveruseFrameDetectorResourceAdaptationModule::VideoSourceRestrictor's
methods to follow a simple pattern:
- What is the next step?
GetHigherFrameRateThan, GetLowerFrameRateThan,
GetHigherResolutionThan, GetLowerResolutionThan
- Can we adapt?
CanIncreaseFrameRate, CanDecreaseFrameRate,
CanIncreaseResolution, CanDecreaseResolution
- Do adapt!
IncreaseFrameRateTo, DecreaseFrameRateTo,
IncreaseResolutionTo, DecreaseResolutionTo
Hopefully this makes the code easier to follow.
This CL changes the "Request Higher/Lower" methods to take the target
as input instead of implicitly calculating the target from the current
input resolution or frame rate.
Bug: webrtc:11222
Change-Id: If625834e921a24a872145105f5d553fb8f9f1795
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/168966
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Reviewed-by: Evan Shrubsole <eshr@google.com>
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30600}
This reverts commit 93d9ae8a17f2e7b90641cbac28e740afc67d383a.
Reason for revert: Perf regression.
Original change's description:
> Remove ResourceAdaptationModule::OnMaybeEncodeFrame
>
> We can react just as well at OnEncodeVideoFrame, which is the same
> behaviour except after checking if the Encoder is paused and the frame
> dropper.
>
> For the initial frame drop, the frame dropper is irrelevant as the frame
> can not be dropped until we are accepting frames. If we didn't drop the
> frame, the encoder can't be paused as the data rate
> is over 0.
>
> For the quality rampup experiment, similar for encoder paused - we can't
> rampup if we are paused anyways since the data rate needs to be non-zero.
> If we are dropping frames we likely don't want to do quality rampup
> anyways.
>
> Bug: webrtc:11222
> Change-Id: Ie3e09d9d8d509dc17ba7a1443cf4747f61c04f6a
> Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/168601
> Reviewed-by: Henrik Boström <hbos@webrtc.org>
> Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
> Commit-Queue: Evan Shrubsole <eshr@google.com>
> Cr-Commit-Position: refs/heads/master@{#30539}
TBR=ilnik@webrtc.org,hbos@webrtc.org,eshr@google.com
# Not skipping CQ checks because original CL landed > 1 day ago.
No-Try: True
Bug: webrtc:11222
Change-Id: Ifb2fc74eb7572568fb0ee1b53a09e4180f87b30c
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/168880
Commit-Queue: Mirko Bonadei <mbonadei@webrtc.org>
Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30568}
We can react just as well at OnEncodeVideoFrame, which is the same
behaviour except after checking if the Encoder is paused and the frame
dropper.
For the initial frame drop, the frame dropper is irrelevant as the frame
can not be dropped until we are accepting frames. If we didn't drop the
frame, the encoder can't be paused as the data rate
is over 0.
For the quality rampup experiment, similar for encoder paused - we can't
rampup if we are paused anyways since the data rate needs to be non-zero.
If we are dropping frames we likely don't want to do quality rampup
anyways.
Bug: webrtc:11222
Change-Id: Ie3e09d9d8d509dc17ba7a1443cf4747f61c04f6a
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/168601
Reviewed-by: Henrik Boström <hbos@webrtc.org>
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Commit-Queue: Evan Shrubsole <eshr@google.com>
Cr-Commit-Position: refs/heads/master@{#30539}
The Resource interface (previously a skeleton not used outside of
testing) is updated to inform listeners of changes to resource
usage. Debugging methods are removed (Name, UsageUnitsOfMeasurements,
CurrentUsage). The interface is implemented by
OveruseFrameDetectorResourceAdaptationModule's inner classes
EncodeUsageResource and QualityScalerResource.
The new ResourceUsageListener interface is implemented by
OveruseFrameDetectorResourceAdaptationModule. In order to avoid adding
AdaptationObserverInterface::AdaptReason to the ResourceUsageListener
interface, the module figures out if the reason is "kCpu" or "kQuality"
by looking which Resource object triggered
OnResourceUsageStateMeasured(). These resources no longer need an
explicit reference to OveruseFrameDetectorResourceAdaptationModule and
could potentially be used by a different module.
In this CL, AdaptationObserverInterface::AdaptDown()'s return value is
still needed by QualityScaler. This is mirrored in the return value of
ResourceUsageListener::OnResourceUsageStateMeasured(). A TODO is added
to remove it and a comment explains how the current implementation
seems to break the contract of the method (as was the case prior to
this CL).
Follow-up work include:
- Move EncodeUsageResource and QualityScalerResource to separate files.
- Make resources injectable, allowing fake resources in testing and
removing OnResourceOveruseForTesting() methods.
(Investigate adding the necessary input signals to the Resource
interface or relevant sub-interfaces so that the module does not need
to know which Resource implementation is used.)
- And more! See whiteboard :)
Bug: webrtc:11222
Change-Id: I0a46ace4a2e617874e3ee97e67e3a199fef420a2
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/168180
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Reviewed-by: Erik Språng <sprang@webrtc.org>
Reviewed-by: Evan Shrubsole <eshr@google.com>
Cr-Commit-Position: refs/heads/master@{#30469}
It would be nice for this to stay in video stream encoder,
but this feature is mostly related to quality scaling. Perhaps
something easier to understand is possible in the future.
Bug: webrtc:11222
Change-Id: I71705f33ff94bbcf2fb9b5c94226c8e76dcba94c
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/168051
Commit-Queue: Evan Shrubsole <eshr@google.com>
Reviewed-by: Erik Språng <sprang@webrtc.org>
Reviewed-by: Henrik Boström <hbos@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30446}
This defines the following methods:
- OnFrame(), replaces SetLastFramePixelCount().
- OnFrameDroppedDueToSize(), a rename of FrameDroppedDueToSize() to
match the other methods.
- OnEncodeStarted(), a rename of the incorrectly named FrameCaptured().
- OnEncodeCompleted(), a rename of the poorly named FrameSent().
In order to get rid of SetLastFramePixelCount(), the "we don't know the
frame size" use case - which was previously implicitly avoided by
invoking SetLastFramePixelCount() with a made-up value for
last_frame_info_ - is now avoided using ".value_or()" in
LastInputFrameSizeOrDefault(). This does mean that a constant 144p
resolution value is referenced in two places, but the fact that this is
a magic value is at least made explicit. This may help future
improvements.
Bug: webrtc:11222
Change-Id: I3b28daa8c5ecf57c6537957d4759f15e24bb2234
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/166961
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Reviewed-by: Evan Shrubsole <eshr@google.com>
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30352}
The poorly named SetEncoderStartBitrate() is renamed
SetEncoderTargetBitrate() and added to the abstract resource adaptation
module interface.
The so-called "start bitrate" was updated to match the target bitrate,
so this was only ever a "start bitrate" until we had any estimates. The
variable is renamed in VideoStreamEncoder as well, and usage of optional
types are introduced to avoid magical values in a few places in the
existing code.
Bug: webrtc:11222
Change-Id: Idde92f68f34616aa3c34ab77a791fdbe7ea7af26
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/166880
Reviewed-by: Evan Shrubsole <eshr@google.com>
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30347}
This squashes together several input signals that were spread out
through several calls into a single method and calling place:
SetEncoderSettings(), invoked from ReconfigureEncoder(). This is added
to the abstract interface.
This makes the following methods obsolete which are removed:
- SetEncoder(): The VideoEncoder was only used for GetEncoderInfo();
the VideoEncoder::EncoderInfo is now part of the EncoderSettings.
- SetEncoderConfig(): The VideoEncoderConfig is part of
EncoderSettings. The config is used for its codec_type and
content_type enums.
- SetCodecMaxFrameRate(): The max frame rate was the same as
VideoCodec::maxFramerate. VideoCodec is now part of EncoderSettings.
There may be some overlap in information between EncoderConfig and
VideoCodec, but that is outside the scope of this CL, which only makes
sure to bundle encoder settings-like information into one input signal.
Bug: webrtc:11222
Change-Id: I67c49c49c0a859cb7d5051939a461593c695a789
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/166602
Reviewed-by: Erik Språng <sprang@webrtc.org>
Reviewed-by: Evan Shrubsole <eshr@google.com>
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30332}
This CL does two things for the sake of getting us closer to adaptation
modules being injectable and usable without knowing implementation
details.
Firstly, RefreshTargetFramerate() is removed. The target frame rate is
dependent on two things: 1) the codec max frame rate, and 2) the video
source restrictions. If either of these two changes, the target frame
rate is updated - there is no need to trigger this externally; the
module already knows if either of these factors change.
The private method MaybeUpdateTargetFrameRate() is added to ensure
overuse_detector->OnTargetFramerateUpdated() happens when necessary.
In doing this, the frame rates are updated to use
absl::optional<double>. This documents its optionality and avoids
magical values (previously -1 was not a bug but meaning "missing"). It
also matches VideoSourceRestrictions::max_frame_rate()'s type.
Secondly, ResetAdaptationCounters() is renamed
ResetVideoSourceRestrictions(). This more accurately describes what it
is doing; it is resetting the restrictions (the adaptation counters
getting reset is merely an implementation specific side-effect of
this). This method is added to the generic interface.
The usefulness of being able to ResetVideoSourceRestrictions() is
questioned in a TODO - current usage of this is when "quality rampup"
finishes. Nevertheless, any module could implement this functionality
so it belongs to the interface for now.
Bug: webrtc:11222
Change-Id: I079785df55fc9894e85087ec98be3e4ebd0713c5
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/166522
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Reviewed-by: Evan Shrubsole <eshr@google.com>
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30320}
The "has_input_video_ logic" is simplified to abort AdaptUp() and
AdaptDown() directly rather than in each calling place of the
VideoSourceRestrictor. The intent is no change in behavior.
The degradation_preference_ is removed from the VideoSourceRestrictor
as its only usage was DCHECKing (not worth it).
ResourceAdaptationModuleInterface gets SetHasInputVideo() and
SetDegradationPreference(), making these things controllable without
knowing implementation details.
StartCheckForOveruse() and StopCheckForOveruse() are renamed to
StartResourceAdaptation() and StopResourceAdaptation().
Bug: webrtc:11222
Change-Id: Id2d7f34d427dfb3ecd4831b1a245d07becae6520
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/166173
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Reviewed-by: Evan Shrubsole <eshr@google.com>
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30296}
This is part of the work for making VideoStreamEncoder responsible for
configuring its source/sink and limiting the responsibility of
OveruseFrameDetectorResourceAdaptationModule to only output relevant
VideoSourceRestrictions.
BEFORE THIS CL
Prior to this CL, OveruseFrameDetector was responsible for performing
AddOrUpdateSink() on the source, which it did using its nested class
VideoSourceProxy.
AddOrUpdateSink() could happen for both adaptation and non-adaptation
related reasons. For example:
- Adaptation related: AdaptUp() or AdaptDown() happens, causing updated
VideoSourceRestrictions.
- Non-adaptation related: VideoStreamEncoder asks the module to
reconfigure the source/sink for it, such as with
SetMaxFramerateAndAlignment() or SetWantsRotationApplied().
AFTER THIS CL
AddOrUpdateSink() is performed by VideoSourceController, which is owned
by VideoStreamEncoder. Any reconfiguration has to go through the
VideoStreamEncoder. This means that:
- Non-adaptation related settings happen between VideoStreamEncoder and
VideoSourceController directly (without going through the adaptation
module).
- Adaptation related changes can be expressed in terms of
VideoSourceRestrictions. OveruseFrameDetectorResourceAdaptationModule
only has to output the restrictions and not know or care about other
source/sink settings.
For now, VideoSourceController has to know about DegradationPreference.
In a future CL, the DegradationPreference logic should move back to
the adaptation module. The VideoSourceRestrictions are fully capable of
expressing all possible source/sink values without the "modifier" that
is the degradation preference.
Bug: webrtc:11222
Change-Id: I0f058c4700ca108e2d9f212e38b61f6f728aa419
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/162802
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Reviewed-by: Evan Shrubsole <eshr@google.com>
Cr-Commit-Position: refs/heads/master@{#30228}
The VideoSourceRestrictions describe the maximum pixels per frame and
max frame rate of a video source.
This CL makes the ResourceAdaptationModuleInterface responsible for the
reconfiguration of video sources. The VideoSourceRestrictions is the
output of an adaptation module, and the ResourceAdaptationModuleListener
handles the callback for when the source restrictions change.
The OveruseFrameDetectorResourceAdaptationModule is updated to output
its changes using these interfaces, and VideoStreamEncoder - now a
listener - is made responsible for triggering the reconfiguring the
video source.
Performing the reconfiguration still requires interacting with the
VideoSourceProxy - it is still partially responsible for keeping track
of rtc::VideoSinkWants settings and performing AddOrUpdateSink(). For
now this may look a bit weird: the VideoSourceProxy tells the
VideoStreamEncoder about the new restrictions, and then the
VideoStreamEncoder tells the VideoSourceProxy to apply these
restrictions on the source/sink. This exercises the listener though, and
unblocks the next CL.
The next CL should move all "configuring the source" logic to the
VideoStreamEncoder instead, so that the only information that is tracked
by OveruseFrameDetectorResourceAdaptationModule is what it actually
outputs to the listener. See the next CL
(https://webrtc-review.googlesource.com/c/src/+/162802) where a
VideoSourceController is introduced, to be owned by the
VideoStreamEncoder rather than the adaptation module.
Bug: webrtc:11222
Change-Id: I450ce74f51d96c4b98009a06134db671893d8fdc
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/162522
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Reviewed-by: Evan Shrubsole <eshr@google.com>
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30227}
This interface will be improved upon iteratively to aid reviewability.
The initial version only handles starting and stopping the module; input
and output of the module is still implementation-specific.
TBR=sprang@webrtc.org
Bug: webrtc:11222
Change-Id: Ie307cfe3d3211c84346c035f2c0e9a632f58221b
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/162580
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Reviewed-by: Evan Shrubsole <eshr@google.com>
Cr-Commit-Position: refs/heads/master@{#30167}
This CL adds Resource, ResourceConsumer, ResourceConsumerConfiguration
and ResourceAdaptationProcessor and implements the algorithm outlined
in
https://docs.google.com/presentation/d/13jyqCWNpIa873iKT6yDuB5Q5ma-c0CvxBpX--0tCclY/edit?usp=sharing.
Simply put, if any resource (such as "CPU") is overusing, the most
expensive consumer (e.g. encoded stream) is adapted one step down.
If all resources are underusing, the least expensive consumer is
adapted one step up.
The current resources, consumers and configurations are all fakes;
this CL has no effect on the current adaptation algorithms used in
practise, but it lays down the foundation for future work in this
area.
Bug: webrtc:11167, webrtc:11168, webrtc:11169
Change-Id: I4054ec7728a52a49e137eee6fa67fa27debd9254
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/161237
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Reviewed-by: Evan Shrubsole <eshr@google.com>
Reviewed-by: Stefan Holmer <stefan@webrtc.org>
Reviewed-by: Erik Språng <sprang@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30053}
