[Webkit-unassigned] [Bug 54456] Optimizing lightning filter to ARM-neon SIMD instruction set

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=54456





--- Comment #18 from Gabor Loki <loki at webkit.org>  2011-03-01 01:39:39 PST ---
(From update of attachment 83296)
View in context: https://bugs.webkit.org/attachment.cgi?id=83296&action=review

> Source/JavaScriptCore/wtf/Platform.h:352
> +#if defined(__ARM_NEON__)
> +#define ENABLE_ARM_NEON 1
> +#endif

Hmm, maybe it should be WTF_CPU_ARM_NEON instead of ENABLE_ARM_NEON. ENABLE_* macros are mainly those kind of features which can be turn on/off irrespectively of build or host system.

> Source/WebCore/ChangeLog:9
> +        allows to speed-up the lighting filter by 4 times on arm

s/arm/ARM/

> Source/WebCore/platform/graphics/filters/FELighting.cpp:36
> +#if CPU(ARM) && ENABLE(ARM_NEON) && COMPILER(GCC)

If CPU(ARM_NEON) is used the condition will be sorter as well.

> Source/WebCore/platform/graphics/filters/FELighting.h:84
> +#if CPU(ARM) && ENABLE(ARM_NEON)
> +    void drawInteriorPixels(LightingData&, LightSource::PaintingData&);

You missed the COMPILER(GCC) part.


At last I had some time to walk through the assembly code, and I could say it is reasonable and optimal.
So, I would like to propose it to be r+ -ed with those little changes. ;)

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.