The motivation is an API change: FloatMath is going to be
deprecated and/or removed. Performance is not the goal of
this change.
That said...
Math is faster than FloatMath with AOT compilation.
While making the change, occurances of:
{Float}Math.sqrt(x * x + y * y) and
{Float}Math.sqrt({Float}Math.pow(x, 2) + {Float}Math.pow(y, 2))
have been replaced with:
{(float)} Math.hypot(x, y)
Right now there is no runtime intrinsic for hypot so is not faster
in all cases for AOT compilation:
Math.sqrt(x * x + y * y) is faster than Math.hypot(x, y) with
AOT, but all other combinations of FloatMath, use of pow() etc.
are slower than hypot().
hypot() has the advantage of being self documenting and
could be optimized in future. None of the behavior differences
around NaN and rounding appear to be important for the cases
looked at: they all assume results and arguments are in range
and usually the results are cast to float.
Different implementations measured on hammerhead / L:
AOT compiled:
[FloatMath.hypot(x, y)]
benchmark=Hypot_FloatMathHypot} 633.85 ns; σ=0.32 ns @ 3 trials
[FloatMath.sqrt(x*x + y*y)]
benchmark=Hypot_FloatMathSqrtMult} 684.17 ns; σ=4.83 ns @ 3 trials
[FloatMath.sqrt(FloatMath.pow(x, 2) + FloatMath.pow(y, 2))]
benchmark=Hypot_FloatMathSqrtPow} 1270.65 ns; σ=12.20 ns @ 6 trials
[(float) Math.hypot(x, y)]
benchmark=Hypot_MathHypot} 96.80 ns; σ=0.05 ns @ 3 trials
[(float) Math.sqrt(x*x + y*y)]
benchmark=Hypot_MathSqrtMult} 23.97 ns; σ=0.01 ns @ 3 trials
[(float) Math.sqrt(Math.pow(x, 2) + Math.pow(y, 2))]
benchmark=Hypot_MathSqrtPow} 156.19 ns; σ=0.12 ns @ 3 trials
Interpreter:
benchmark=Hypot_FloatMathHypot} 1180.54 ns; σ=5.13 ns @ 3 trials
benchmark=Hypot_FloatMathSqrtMult} 1121.05 ns; σ=3.80 ns @ 3 trials
benchmark=Hypot_FloatMathSqrtPow} 3327.14 ns; σ=7.33 ns @ 3 trials
benchmark=Hypot_MathHypot} 856.57 ns; σ=1.41 ns @ 3 trials
benchmark=Hypot_MathSqrtMult} 1028.92 ns; σ=9.11 ns @ 3 trials
benchmark=Hypot_MathSqrtPow} 2539.47 ns; σ=24.44 ns @ 3 trials
Bug: https://code.google.com/p/android/issues/detail?id=36199
Change-Id: I06c91f682095e627cb547d60d936ef87941be692
Avoid AOT compilation/initialization in ART failing due to regex.Pattern
requiring native support.
Bug: 14646037
Change-Id: I9d9c7fd9ed9fab7e21d37f9b31c00d037fc5a691
Bug: 13580771
The cached value mHasAlpha on Bitmap could get out of sync with the
real value stored in the SkBitmap. Fix this by just eliminating the
cached value. This adds an extra JNI call in copy(), createBitmap(),
and setPremultiplied(), none of which are going to be significantly
impacted by the extra JNI call.
Change-Id: I4fce069a77d9b3bbb47dde04275ff8510ce7e4f0
Remove links to non-specific function name.
Bug: 13246311
Change-Id: I5643074149d0a511bd760d638751c44e3ec4c557
(cherry picked from commit 2bc49ca96eb786f66dc2b9573e1f09bdb2d82d5e)
(These CLs are already in master.)
Bug: 13246311
This cherry-picks 7 CLs:
-----------------------------------------------------------------------
Remove calls to deprecated SkBitmap::setIsOpaque()
setIsOpaque() has been removed from ToT Skia.
Update setters for mIsPremultiplied and hasAlpha to take the
other into consideration.
cherry-pick from: I1b36b0b0ce7126031eb7b769b563c17dcd4b306a
-----------------------------------------------------------------------
Merge AssetStream with AssetStreamAdaptor.
Add enums to the constructor for AssetStreamAdaptor to choose the
different behaviors used by the (former) two different classes.
The old clients of AssetStream now get the following features of
AssetStreamAdaptor
- Debugging statements on error.
- The stream is an SkStreamRewindable.
- getLength() returns the correct value, and the old way of getting
the length (read(NULL, 0)) is no longer implemented, since it is
no longer used.
- isAtEnd() returns the correct value. ToT Skia makes it pure virtual,
so some implementation is necessary.
cherry-pick from: I2a5395914e4f53830aaefee396556459083a1c56
-----------------------------------------------------------------------
Deprecate Android-specific SkPaint functions.
The following functions were problematic:
const SkGlyph& getUnicharMetrics(SkUnichar, const SkMatrix*);
const SkGlyph& getGlyphMetrics(uint16_t, const SkMatrix*);
const void* findImage(const SkGlyph&, const SkMatrix*);
Replacing them with calls through SkGlyphCache solved a nasty crash
bug, so they have all been deprecated.
Bug: 11968757
cherry-pick from: Id746315d41aec5b211b78b172a883c2061130f08
-----------------------------------------------------------------------
pass SkGlyphCache into updateGlyphCache()
Doing so prevents us from double-locking the glyph cache, thereby
effectively locking ourselves out of reusing work that we'd just done.
Bug: 11968757
cherry-pick from: I5c552f2d0bbe30af2ce9054ba684e7da756a0d89
-----------------------------------------------------------------------
Updates to the Skia API needed to merge the WebView m33 version of Skia.
cherry-pick from: I0f63b53f2aae58871413b132742fc84138f069a3
Bugfix for screenshots (recent apps) due to incorrect rowBytes computation
bug: 12915192
cherry-pick from: I4d5fe2a2f75baf66099e0970fb646686a1992714
-----------------------------------------------------------------------
Fix bug in AndroidPixelRef where we did not store the correct imageInfo for a recycled bitmap.
cherry-pick from: I882483b78886e2f19fa4e43a86e69f5a82b3b7e5
-----------------------------------------------------------------------
Change-Id: Ie2b731a9f0795802418cfecddb4b684c92c64d33
This changes BaseObj to support 64-bit IDs. There are a few caveats:
1. Since it is deprecated, RSG will not support 64-bit.
2. Currently, methods that pass arrays of IDs to the driver are not supported in 64-bit. This will be fixed in a later CL.
bug 11332320
Change-Id: If0dbecc8b285e260f767e441e05088b6a1b749a2
For storing pointers, long is used in
android/graphics/Camera class, as native
pointers can be 64-bit.
In addition, some minor changes have been done
to conform with standard JNI practice (e.g. use of
jint instead of int in JNI function prototypes)
Change-Id: Ic05ebf2051a225a1638a43f476bab6176c0b5c38
Signed-off-by: Ashok Bhat <ashok.bhat@arm.com>
Signed-off-by: Marcus Oakland <marcus.oakland@arm.com>
Changes in this patch include
[x] Long is used to store native pointers as they can
be 64-bit.
[x] Some minor changes have been done to conform with
standard JNI practice (e.g. use of jint instead of int
in JNI function prototypes)
[x] AssetAtlasManager is not completely 64-bit compatible
yet. Specifically mAtlasMap member has to be converted
to hold native pointer using long. Added a TODO to
AssetAtlasManager.java to indicate the change required.
Change-Id: I940433f601c6db998c1a8ffff338f5361200d5ed
Signed-off-by: Ashok Bhat <ashok.bhat@arm.com>
Signed-off-by: Craig Barber <craig.barber@arm.com>
Signed-off-by: Kévin PETIT <kevin.petit@arm.com>
Signed-off-by: Marcus Oakland <marcus.oakland@arm.com>