Bug: 16712006
Initial work towards benchmarking HWUI systems
Currently this will just create a screen full of
"cards" to simulate a high load scenario for
shadows and clipping
Change-Id: Ie9f9a9570844e136db8053e8fc62fe06cb922a5f
Graphics memory usually gets trimmed in applications when the
activity goes into the background. We use quite a lot of graphics
memory when the shade/lockscreen is open, and some of them never gets
freed unless the recents activity is closed, because we don't have
these activity-trimming-heuristics for the shade. This change
proactively trims the graphics memory when the shade gets closed or
when the lockscreen is hidden, to emulate the same heuristics as for
activities.
This change also adds trimMemory on RenderThread to systrace to
verify that no jank is introduced with this change.
This change immediately saves around 10-30 MB on an xxhdpi device
after the shade is closed.
Bug: 17581375
Change-Id: I4fb622efb51815fe08187be97ba15d012d4de5d4
This is helping spot shadow for 15%-20% increase.
With the new algorithm, we are less sensitive to the floating point error.
b/16712006
Change-Id: Ie30a6ce01e73d56054a0cf65a84549454339a7fd
Bug: 17765082
DeferredLayerUpdater had fallen behind RT updates. Re-snap to
latest expectations, ensuring to call requireGlContext() prior
to detachSurfaceTexture to avoid leaking SurfaceTextures
Change-Id: Ic65fb9831e5284f658866da8da9ad5af1d227699
+ adjusting spot and ambient shadow opacity constants to achieve desired appearance
+ reducing ambient scale ratio back to 1.0 to address over-lightening at higher elevations
+ partially revert ag/546290
Change-Id: I9d7f664f73a7b9b83df73b739103c97054bd4f6e
bug:17600162
Transparent draws are not safe to reject for all xfermodes other than
clear. Now, to be safe, only perform the rejection for SrcOver draws
since other modes are fairly uncommon.
We could specifically determine whether the xfermode could change the
output given a transparent input, but there's little to be gained from
the additional complexity.
Change-Id: Ia699ac4bdc4da3353955840b53f1922d3cb1d85d
bug:17463894
Fixes the setViewport method to immediately affect the return values
of getViewportWidth/Height methods.
Also works around tiling extension issues observed on first frame
after window resize by disabling tiling for that frame.
Change-Id: Ie172d572d20d74a1be9cc58ad389af2cffa0e4b6
Bug: 17479800
FBO cache is very expensive and no longer necessary, disable
it by just setting size to 0.
Change-Id: I664616f262c8339919e1d20baaafa5de2b628d7e
bug:17405627
Previously, the input content to the color matrix computation was left
premultiplied. Since the color matrix could reduce the alpha channel,
the alpha was re-multiplied, but this was incomplete, and incorrect.
Instead, apply the color matrix in unpremultiplied space.
Change-Id: I87b8e03d2e228e6ded81f7bbfea952605d7a095c
bug:17379260
Unclipped save layers need to isolate their clips and matrices
so that the save time readout of background content matches clip and
position with the restore/compose step.
Change-Id: I2de474cda76e960d080852f1716d6ddfa8a512d4
Fix the valid umbra detection.
This looks better b/c every vertex will have one ray shooting at it, such that
we don't miss the corner.
This performs better too, due to the polygon intersection is removed and less ray
intersection. 2x performance for rect and circle for spot shadow in test app.
b/17288227
b/15598793
b/16712006
Change-Id: I4a5ee397b9e192e93c8e35e6260b499e3e38a6f4
bug:17441218
GLConsumer uses glBindTexture() directly instead of going through
Caches::bindTexture(). This can cause libhwui to draw with the wrong
texture bound in the following case which involves 2 TextureViews:
Frame 1:
GLConsumer::updateTexImage() calls glBindTexture(GL_TEXTURE_EXTERNAL_OES, 1)
HWUI renders TextureView A:
calls Caches::bindTexture(GL_TEXTURE_EXTERNAL_OES, 1) and draws
Frame 2:
GLConsumer::updateTexImage() calls glBindTexture(GL_TEXTURE_EXTERNAL_OES, 1)
GLConsumer::updateTexImage() calls glBindTexture(GL_TEXTURE_EXTERNAL_OES, 2)
HWUI renders TextureView A:
calls Caches::bindTexture(GL_TEXTURE_EXTERNAL_OES, 1) and draws
HWUI renders TextureView B:
calls Caches::bindTexture(GL_TEXTURE_EXTERNAL_OES, 2) and draws
In this case, HWUI will incorrectly draw TextureView A using texture 2 on
frame 2, because mBoundTextures[0]=1, even though the texture currently
bound to GL_TEXTURE_EXTERNAL_OES is 2.
Since GLConsumer is always used with a target of GL_TEXTURE_EXTERNAL_OES,
work around this problem by having mBoundTextures[] store only the
textures bound to the target GL_TEXTURE_2D. This is the common case
where the extra performance is needed. Since it's legal to have
different textures bound to GL_TEXTURE_2D and GL_TEXTURE_EXTERNAL_OES
on one texture unit, Caches::bindTexture() does not need to clear
mBoundTextures[mTextureUnit] when target != GL_TEXTURE_2D.
Change-Id: I8bc54ab8adcfacad7f3ed17a31236dc7a86c967a
Signed-off-by: Fred Fettinger <fred.fettinger@motorola.com>
bug:17401066
Now correctly accounts for scale, both in hairline case (where scale
needs to be accounted for), and in standard case (where scale
shouldn't be applied, since bounds are in local space)
Change-Id: I597a20834dce42ddb741b46e4c1a4f3169a48ccc
bug:17208461
They are destroyed via finalizer-enqueued destroy method, so it's not
valid to check that they've been destroyed at gl context destruction
time.
Change-Id: I670f69825547facd5f31d44acb406418881fee00
