renderengine: Skip cross-frame blur cache for dynamic blur input#1
Open
zylhdrXP wants to merge 20 commits into
Open
renderengine: Skip cross-frame blur cache for dynamic blur input#1zylhdrXP wants to merge 20 commits into
zylhdrXP wants to merge 20 commits into
Conversation
This reverts commit 7c93509. Change-Id: I1aa4f4d64d49022b4f9d97e6ebe15d81216d65c5 Signed-off-by: Frost <159105703+Karan-Frost@users.noreply.github.com> Signed-off-by: Dmitrii <bankersenator@gmail.com> Signed-off-by: iHSAN <ihsanulrahman@proton.me> Signed-off-by: Ghosuto <clash.raja10@gmail.com>
This reverts commit 6d223cb. Change-Id: Iaa6f872313cb6c6433c73d715d5f16d853d4a3f9 Signed-off-by: Frost <159105703+Karan-Frost@users.noreply.github.com> Signed-off-by: Dmitrii <bankersenator@gmail.com> Signed-off-by: iHSAN <ihsanulrahman@proton.me> Signed-off-by: Ghosuto <clash.raja10@gmail.com>
This reverts commit 1e7c09b. Change-Id: I9b81b02c293ee0018a79a6028ae97bf628968fc3 Signed-off-by: Frost <159105703+Karan-Frost@users.noreply.github.com> Signed-off-by: Dmitrii <bankersenator@gmail.com> Signed-off-by: iHSAN <ihsanulrahman@proton.me> Signed-off-by: Ghosuto <clash.raja10@gmail.com>
Layer dataspace is initialized as unknown by default, unless overriden by the buffer source. We don't require expensive rendering for color conversion when the dataspace is unknown. Change-Id: I079c520f63a65c77ba3162664656e607eafff991 Signed-off-by: Ghosuto <clash.raja10@gmail.com>
If a invisible parent layer set a fixedTransformHint, the sub layers of this parent layer should be updated in time, especially in the occation that when a landscape Activity exit and return to vertical Activity. It can optimize performance by avoiding unnecessary buffer reallocation while dequeuBuffer for sub layers. Change-Id: I5327591af6c2a4ba290d2ea2059a407171e47ecf Signed-off-by: jiayongqiang <jiayongqiang@xiaomi.com> Signed-off-by: Ghosuto <clash.raja10@gmail.com>
No layers definitely doesn't mean new changes happened; so there's no need to enforce the maximum refresh rate. Maintain current refresh rate instead. [gotenksIN: adapt to 14] Change-Id: I31c9edc1a8ff3ff8340afd2f731df66755ffc9e0 Co-authored-by: Omkar Chandorkar <gotenksIN@aospa.co> Signed-off-by: Omkar Chandorkar <gotenksIN@aospa.co> Signed-off-by: Ghosuto <clash.raja10@gmail.com>
presentTime is missing when the user isn't touching the screen. To make the estimation more accurate regardless of the user's input, remove this check. Change-Id: I27c55989b7f3d5ef5efe6b0d709135122bb6ba87 Signed-off-by: Juhyung Park <qkrwngud825@gmail.com> Signed-off-by: Ghosuto <clash.raja10@gmail.com>
With the previous commit entering idle aggressively, it is important that touch boost works well. Since there are valid cases where we would want touch boost to work when there are no layers detected (e.g., notification panel pull down if it was not accounted for during the initial vote type set), change touch boost to work regardless of layer's status. Change-Id: I0a125cf9027440de205fa4ca611657b70b8a088f Signed-off-by: Juhyung Park <qkrwngud825@gmail.com> Signed-off-by: Adithya R <gh0strider.2k18.reborn@gmail.com> Signed-off-by: Ghosuto <clash.raja10@gmail.com>
30 frames for 300ms is quite enough for approximated calculation. As an added bonus, the device reacts quicker when a higher refresh rate should be used. Change-Id: Id8930ab7343cc4e974b78929615e6efdd1c2b27e Signed-off-by: Juhyung Park <qkrwngud825@gmail.com> Signed-off-by: Ghosuto <clash.raja10@gmail.com>
tuned heap utilization, usap pool, and madvise limits for high-ram devices Change-Id: I20f6009aa21b9286b11fa34347becdb9b61f0ee1 Signed-off-by: Ghosuto <clash.raja10@gmail.com>
6gb: utilization 0.5->0.6 with usap/madvise/pinner, 4gb: bumping growthlimit and maxfree with usap/madvise, 2gb/1gb: raising minfree from 512k to reduce micro-gc churn Change-Id: I4495ee8708a100db9e13c00120bb94735f50a744 Signed-off-by: Ghosuto <clash.raja10@gmail.com>
client comp cache hit was clearing the expensive rendering hint mid-blur, causing the gpu boost to flap on/off frame to frame during steady state blur like qs held open. keep it asserted while a blur layer is still present. Change-Id: I3632b4cca7ebb06d21ea1098764bef8de5e78f1c Signed-off-by: Ghosuto <clash.raja10@gmail.com>
clean clear-glass look on kawase v2, with adjustments: kMaxSurfaces 4->3, filterDepth 2.5->3.0, kInputScale 0.25->0.20 across all filter variants, plus a 4-entry cross-frame blur cache keyed on below-blur layer fingerprints so held-steady frames skip generate(). perfetto BackgroundBlur cost on felix outer 120Hz, QS expand+hold cycles: metric pre-cache with cache delta count 324 279 - avg 92.5us 58.3us -37% min 32.0us 7.0us -78% p50 72.2us 37.4us -48% p90 169.0us 155.8us -8% p95 187.5us 177.9us -5% p99 223.1us 240.4us +8% max 336.9us 292.3us -13% Change-Id: I89cb74e6fd2ab01d5c24b4f12162b8f24bad5e9f Signed-off-by: rmp22 <195054967+rmp22@users.noreply.github.com> Signed-off-by: Ghosuto <clash.raja10@gmail.com>
Change-Id: I82f555cd8731f56948473c42058652fa3931525e Signed-off-by: Pranav Vashi <neobuddy89@gmail.com> Signed-off-by: Ghosuto <clash.raja10@gmail.com>
MediaTek devices dither the framework hbm dimlayer from renderengine in order to make the background match the system brightness, effectively only making the udfps icon bright. Change-Id: I66cb2d7d2ca3d08f768c6a66521a60eaa66015aa Signed-off-by: Ghosuto <clash.raja10@gmail.com>
Change-Id: I1939c578adc5b6e5aa221634dda9a143d0ea3aa9 Signed-off-by: Ghosuto <clash.raja10@gmail.com>
Change-Id: I2a1b99545fd369cd9f56a4279010e8f3766ff27c Signed-off-by: rmp22 <195054967+rmp22@users.noreply.github.com> Signed-off-by: Ghosuto <clash.raja10@gmail.com>
Change-Id: Iedb7dfd993259fa91a8dbadbe83b32aa7c62862b Signed-off-by: Ghosuto <clash.raja10@gmail.com>
Signed-off-by: Ghosuto <clash.raja10@gmail.com>
The cross-frame blur cache added in f5ccc185cb keys cached blur images on a fingerprint of the layers below the blur, mixing in each layer's GraphicBuffer::getId(), alpha, transform and boundaries. getId() returns mId, which is assigned once at allocation and is stable for the lifetime of the buffer object. Video producers (e.g. TikTok) recycle a small buffer pool and write a new frame into the same buffer in place, so the id stays constant while the pixels change. When the QS shade is expanded over a playing video, the shade's background-blur layer covers the screen and its blur input is the video below it. The video pool cycles through only a few buffer ids, and with a held panel the blur radius and blurRect are constant, so the fingerprint collapses to a handful of values that all fit in the 4-entry cache. After the first cycle every lookup hits and generate() is skipped, leaving the on-screen blurred backdrop frozen even though the video keeps decoding. Neither the buffer id, the acquire fence nor the geometry can detect an in-place content update at composition time, so any buffer-backed layer below the blur must be treated as dynamic. Track this while computing the fingerprint and bypass the cross-frame cache (lookup and store) whenever dynamic content is present. Fully static scenes (solid-color or effect-only layers) remain eligible for the cache, preserving the QS-hold power win the cache was added for. The per-frame cachedBlurs map and the Output ClientCompositionRequestCache are unaffected. (cherry picked from commit AlphaDroid-Project/frameworks_native@598e2e9)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
8 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The cross-frame blur cache added in f5ccc185cb keys cached blur images on a fingerprint of the layers below the blur, mixing in each layer's GraphicBuffer::getId(), alpha, transform and boundaries. getId() returns mId, which is assigned once at allocation and is stable for the lifetime of the buffer object. Video producers (e.g. TikTok) recycle a small buffer pool and write a new frame into the same buffer in place, so the id stays constant while the pixels change.
When the QS shade is expanded over a playing video, the shade's background-blur layer covers the screen and its blur input is the video below it. The video pool cycles through only a few buffer ids, and with a held panel the blur radius and blurRect are constant, so the fingerprint collapses to a handful of values that all fit in the 4-entry cache. After the first cycle every lookup hits and generate() is skipped, leaving the on-screen blurred backdrop frozen even though the video keeps decoding.
Neither the buffer id, the acquire fence nor the geometry can detect an in-place content update at composition time, so any buffer-backed layer below the blur must be treated as dynamic. Track this while computing the fingerprint and bypass the cross-frame cache (lookup and store) whenever dynamic content is present. Fully static scenes (solid-color or effect-only layers) remain eligible for the cache, preserving the QS-hold power win the cache was added for. The per-frame cachedBlurs map and the Output ClientCompositionRequestCache are unaffected.
(cherry picked from commit AlphaDroid-Project/frameworks_native@598e2e9)