Bug 65522 – Optimizing arithmetic composite filter to ARM-neon SIMD instruction set

• Status: RESOLVED
• Resolution: FIXED
• Component: SVG

[Felician Marton]

Fasten up the arithmetic composite filter.

[Gabor Loki]

Is there any progress on this topic? or should we close this bug?

[Felician Marton]

Created attachment 109774 [details]
Arithmetic composite neon brief patch

The MAC project file update is missing from the patch.

[Zoltan Herczeg]

Created attachment 112164 [details]
patch

[WebKit Review Bot]

Comment on attachment 112164 [details]
patch

Attachment 112164 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://queues.webkit.org/results/10201753

[Gyuyoung Kim]

Comment on attachment 112164 [details]
patch

Attachment 112164 [details] did not pass efl-ews (efl):
Output: http://queues.webkit.org/results/10203712

[Early Warning System Bot]

Comment on attachment 112164 [details]
patch

Attachment 112164 [details] did not pass qt-ews (qt):
Output: http://queues.webkit.org/results/10200772

[Zoltan Herczeg]

Created attachment 112165 [details]
patch again

[Dirk Schulze]

Comment on attachment 112165 [details]
patch again

View in context: https://bugs.webkit.org/attachment.cgi?id=112165&action=review

LGTM. I can just trust you on the Neon code. Just some snippets.

> Source/WebCore/ChangeLog:12
> +        for all four RGBA channels in the same time.

at the same time?

> Source/WebCore/ChangeLog:13
> +

Just mention that it won't affect the general behavior and existing tests will cover the changes.

> Source/WebCore/platform/graphics/filters/FEComposite.cpp:123
> -    float scaledK4;
> -    float scaledK1;
> +    float scaledK1, scaledK4;

that is not allowed according the webkit style rules. Just leave it as it was.

> Source/WebCore/platform/graphics/filters/FEComposite.cpp:180
> +    ASSERT(!(pixelArrayLength & 0x3));

where is pixelArrayLength?

[Zoltan Herczeg]

Landed in trac.webkit.org/changeset/99183
Closing bug.