We have to cache the starting point for the latest "move", then apply it after
we close the path.
b/18214929
Change-Id: I8e8e5c810d720a1c194b8f59d74867a0efbb7662
Transition PackageManager internals away from heavier HashMap/HashSet
to use drop-in ArrayMap/ArraySet replacements. Saves ~38% RAM and
thousands of objects on a typical device.
Bug: 18115729
Change-Id: Ie107d2fee4b7baa4e3c3923231b4be877d1a5d2f
Also adds IntArray, which is like LongArray for integers, and prevents
the AM/PM label text in the time picker header from wrapping.
BUG: 17468036
Change-Id: I7120089885709f23e20368927e4b3ed9db2e5393
On modern versions of Android running in AOT mode
FloatMath is slower than Math. Calls to Math.sqrt(),
etc. are replaced by intrinsics which can be as small
as a single CPU opcode.
When running in interpreted mode the new
implementation is unfortunately slower, but I'm
judging this acceptable and likely to be improved
over time. This change saves a small amount of native
code.
Example timings:
Mako AOSP AOT:
Method: Original / New / Direct call to Math
ceil: 596ns / 146.ns / 111ns
sqrt: 694ns / 56ns / 25ns
Mako AOSP interpreted:
Method: Original / New / Direct call to Math
ceil: 1900ns / 2307ns / 1485ns
sqrt: 1998ns / 2603ns / 1788ns
Other calls Mako AOT:
Method: Original / New
cos: 635ns / 270ns
exp: 566ns / 324ns
floor: 604ns / 150ns
hypot: 631ns / 232ns
pow: 936ns / 643ns
sin: 641ns / 299ns
The advice to use Math directly, in preference to
FloatMath, is still good. FloatMath will be deprecated
separately.
Bug: https://code.google.com/p/android/issues/detail?id=36199
Change-Id: If07fcbd78543d13bc6d75f9743f999860e8d58d7
Somehow the change at
https://android-review.googlesource.com/#/c/109481/1/core/java/android/util/Spline.java
isn't in master, which seems to be causing a boot failure.
07-28 07:44:51.338 771 771 E art : No implementation found for float android.util.FloatMath.hypot(float, float) (tried Java_android_util_FloatMath_hypot and Java_android_util_FloatMath_hypot__FF)
07-28 07:44:51.338 771 771 W SystemServer: ***********************************************
07-28 07:44:51.339 771 771 F SystemServer: BOOT FAILURE making Power Manager Service ready
07-28 07:44:51.339 771 771 F SystemServer: java.lang.UnsatisfiedLinkError: No implementation found for float android.util.FloatMath.hypot(float, float) (tried Java_android_util_FloatMath_hypot and Java_android_util_FloatMath_hypot__FF)
07-28 07:44:51.339 771 771 F SystemServer: at android.util.FloatMath.hypot(Native Method)
07-28 07:44:51.339 771 771 F SystemServer: at android.util.Spline$MonotoneCubicSpline.<init>(Spline.java:168)
07-28 07:44:51.339 771 771 F SystemServer: at android.util.Spline.createMonotoneCubicSpline(Spline.java:72)
07-28 07:44:51.339 771 771 F SystemServer: at android.util.Spline.createSpline(Spline.java:47)
07-28 07:44:51.339 771 771 F SystemServer: at com.android.server.display.DisplayPowerController.createAutoBrightnessSpline(DisplayPowerController.java:1100)
Change-Id: Ic6a85d4f9125b921fa4f307e9d5723c09104ca4d
On modern versions of Android running in AOT mode
FloatMath is slower than Math. Calls to Math.sqrt(),
etc. are replaced by intrinsics which can be as small
as a single CPU opcode.
When running in interpreted mode the new
implementation is unfortunately slower, but I'm
judging this acceptable and likely to be improved
over time. This change saves a small amount of native
code.
Example timings:
Mako AOSP AOT:
Method: Original / New / Direct call to Math
ceil: 596ns / 146.ns / 111ns
sqrt: 694ns / 56ns / 25ns
Mako AOSP interpreted:
Method: Original / New / Direct call to Math
ceil: 1900ns / 2307ns / 1485ns
sqrt: 1998ns / 2603ns / 1788ns
Other calls Mako AOT:
Method: Original / New
cos: 635ns / 270ns
exp: 566ns / 324ns
floor: 604ns / 150ns
hypot: 631ns / 232ns
pow: 936ns / 643ns
sin: 641ns / 299ns
The advice to use Math directly, in preference to
FloatMath, is still good. FloatMath will be deprecated
separately.
Bug: https://code.google.com/p/android/issues/detail?id=36199
Change-Id: I8d1947d88b3c576643138b1df589fb9da7c1ab88
Now "1e-5" will not be separated as "1e" and "-5".
Add one test for this use case.
Make sure we print out the pathData when path parsing has error.
b/17919923
Change-Id: I10a00ce21166cfb5a009c49c1a93f40eeb956d83
Like "0.0.0" will be separated to "0.0 .0" now, just to make sure we are more
complied with svg path data.
b/17892882
Change-Id: Id7b64e9882f5174aa794a0256e2a29d66c724876
Fix Slog.wtf to not acquire the activity manager lock in its code
path, so that it can never deadlock. This was the original intention
of it, but part was missed.
Now we can put back in the code to detect when strict mode data is
getting large (a little more targeted now to the actual problem),
and use Slog.wtf to report it. And as a bonus, when this happens
we will now clear all of the collected violations, to avoid getting
in to the bad case where IPCs start failing. So this should be
good enough for L to fix the problem, with wtf reports for us to
see if the underlying issue is still happening.
Finally, switch a butch of stuff in the system process from Log.wtf
to Slog.wtf, since many of those are deadlocks waiting to happen.
Oh and fix a crash in the settings provider I noticed in APR.
Change-Id: I307d51b7a4db238fd1e5fe2f3f9bf1b9c6f1c041
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
Some devices may contain things like transflective displays where
they actually want the backlight brightness to decrease in the face
of high ambient light scenarios. In the interest of time, just use a
linear interpolation based on the brightness control points for these
cases but in the future we should consider adding a non-monotonic
cubic spline.
Bug: 15611140
Change-Id: I8fcee061b18c8ae9cc9a1b38d8b850aaec4478ce
Getting the indexOf is useful for doing compound operations
like:
int i = set.indexOf(key);
if (i >= 0) {
Object o = set.valueAt(i);
o.blah();
set.removeAt(i);
}
Change-Id: I3d4b77d1461ba969fc6b4d332d52d4d084b5b53c
...can lead to launching of un-exported activities
We now validate the array map after unparcelling to make sure there
are no duplicate keys.
And to make up for the performance overhead this introduces, I switched
the parcelling/unparcelling code to write keys as explicit string
objects rather than generic values. There was no reason to use generic
values since the write method itself only accepts an array map with
String keys.
Change-Id: I57bda9eb79ceaaa9c1b94ad49d9e462b52102149
Like new settings actions for some of the new settings panels.
And fix voice interaction services so they require a recognizer.
And tweak array map doc to be correct.
Blah blah blah.
Change-Id: Ib5e66b574b10e7b3fa39723b21046a74e6ead204
