Perlin noise adjustments

Perlin noise adjustments

1 ) Added GPU simplex noise as a comparison to GPU Perlin noise. It can be enabled simply by using a #define (SK_USE_SIMPLEX_NOISE)
2 ) Enabled GPU Perlin noise tests
3 ) Added a small benchmark for noise
4 ) Disabled GPU Perlin noise on Android based on benchmarks

Committed: https://code.google.com/p/skia/source/detail?r=8719

[sugoi, 2013-04-10 19:44:50]

I still need to run test on trybots to see if any of them are failing, but this
should essentially be it for noise before I integrate to blink.

[Stephen White, 2013-04-10 19:55:09]

https://codereview.chromium.org/14087002/diff/1/src/effects/SkPerlinNoiseShad...
File src/effects/SkPerlinNoiseShader.cpp (right):

https://codereview.chromium.org/14087002/diff/1/src/effects/SkPerlinNoiseShad...
src/effects/SkPerlinNoiseShader.cpp:563: virtual void setData(const
GrGLUniformManager&, const GrDrawEffect&);
I think you should have SK_OVERRIDE here too.

https://codereview.chromium.org/14087002/diff/1/src/effects/SkPerlinNoiseShad...
src/effects/SkPerlinNoiseShader.cpp:580: int numOctaves() const { return
fNumOctaves & 0xFF; /*[0,255] octaves allowed*/ }
Shouldn't this limit be applied on set, rather than get?

https://codereview.chromium.org/14087002/diff/1/src/effects/SkPerlinNoiseShad...
src/effects/SkPerlinNoiseShader.cpp:596: fAlpha == s.fAlpha;
Need to compare fNumOctaves here too?

[sugoi, 2013-04-10 20:02:42]

https://codereview.chromium.org/14087002/diff/1/src/effects/SkPerlinNoiseShad...
File src/effects/SkPerlinNoiseShader.cpp (right):

https://codereview.chromium.org/14087002/diff/1/src/effects/SkPerlinNoiseShad...
src/effects/SkPerlinNoiseShader.cpp:563: virtual void setData(const
GrGLUniformManager&, const GrDrawEffect&);
Fixed in all 3 places

https://codereview.chromium.org/14087002/diff/1/src/effects/SkPerlinNoiseShad...
src/effects/SkPerlinNoiseShader.cpp:580: int numOctaves() const { return
fNumOctaves & 0xFF; /*[0,255] octaves allowed*/ }
It is already set in SkPerlinNoiseShader's constructor, I'll remove this one.

https://codereview.chromium.org/14087002/diff/1/src/effects/SkPerlinNoiseShad...
src/effects/SkPerlinNoiseShader.cpp:596: fAlpha == s.fAlpha;
On 2013/04/10 19:55:09, Stephen White wrote:
> Need to compare fNumOctaves here too?

Done.

[bsalomon, 2013-04-10 20:21:10]

lgtm (two nits and a question)

https://codereview.chromium.org/14087002/diff/1/bench/PerlinNoiseBench.cpp
File bench/PerlinNoiseBench.cpp (right):

https://codereview.chromium.org/14087002/diff/1/bench/PerlinNoiseBench.cpp#ne...
bench/PerlinNoiseBench.cpp:25: test(canvas, 0, 0,
SkPerlinNoiseShader::kFractalNoise_Type,
this->

https://codereview.chromium.org/14087002/diff/1/bench/PerlinNoiseBench.cpp#ne...
bench/PerlinNoiseBench.cpp:25: test(canvas, 0, 0,
SkPerlinNoiseShader::kFractalNoise_Type,
One run is enough?

https://codereview.chromium.org/14087002/diff/1/bench/PerlinNoiseBench.cpp#ne...
bench/PerlinNoiseBench.cpp:51: drawClippedRect(canvas, x, y, paint);
this->

[sugoi, 2013-04-10 20:28:01]

https://codereview.chromium.org/14087002/diff/1/bench/PerlinNoiseBench.cpp
File bench/PerlinNoiseBench.cpp (right):

https://codereview.chromium.org/14087002/diff/1/bench/PerlinNoiseBench.cpp#ne...
bench/PerlinNoiseBench.cpp:25: test(canvas, 0, 0,
SkPerlinNoiseShader::kFractalNoise_Type,
On 2013/04/10 20:21:10, bsalomon wrote:
> this->

Done.

https://codereview.chromium.org/14087002/diff/1/bench/PerlinNoiseBench.cpp#ne...
bench/PerlinNoiseBench.cpp:25: test(canvas, 0, 0,
SkPerlinNoiseShader::kFractalNoise_Type,
On 2013/04/10 20:21:10, bsalomon wrote:
> One run is enough?

Well, it is very simplistic, but from what I saw, it was scaling fairly linearly
in the [1-10] range for the number of octaves (the parameter that I set to 3
here), and other parameters don't affect performance significantly, so I didn't
bother adding more since it felt somewhat redundant.

https://codereview.chromium.org/14087002/diff/1/bench/PerlinNoiseBench.cpp#ne...
bench/PerlinNoiseBench.cpp:51: drawClippedRect(canvas, x, y, paint);
On 2013/04/10 20:21:10, bsalomon wrote:
> this->

Done.

[Stephen White, 2013-04-10 20:34:12]

LGTM

[sugoi1, 2013-04-17 13:49:03]

Committed patchset #2 manually as r8719 (presubmit successful).