Color dodge and burn with SkPMFloat.

Color dodge and burn with SkPMFloat.

Both 25-35% faster with SSE.
With NEON, Burn measures as a ~10% regression, Dodge a huge 2.9x improvement.

The Burn regression is somewhat artificial: we're drawing random colored rects onto an opaque white dst, so we're heavily biased toward the (d==da) fast path in the serial code.  In the vector code there's no short-circuiting and we always pay a fixed cost for ColorBurn regardless of src or dst content.

Dodge's fast paths, in contrast, only trigger when (s==sa) or (d==0), neither of which happens any more than randomly in our benchmark.  I don't think (d==0) should happen at all.  Similarly, the (s==0) Burn fast path is really only going to happen as often as SkRandom allows.

In practice, the existing Burn benchmark is hitting its fast path 100% of the time.  So I actually feel really great that this only dings the benchmark by 10%.

Chrome's still guarded by SK_SUPPORT_LEGACY_XFERMODES, which I'll lift after finishing the last xfermode, SoftLight.

BUG=skia:

Committed: https://skia.googlesource.com/skia/+/2aab22a58a366df4752c1cf0f004092c6e7be335

[mtklein, 2015-06-26 00:43:40]

mtklein@google.com changed reviewers:
+ mtklein@google.com

[mtklein, 2015-06-26 00:43:41]

https://codereview.chromium.org/1214443002/diff/40001/src/core/SkPMFloat.h
File src/core/SkPMFloat.h (right):

https://codereview.chromium.org/1214443002/diff/40001/src/core/SkPMFloat.h#ne...
src/core/SkPMFloat.h:30: Sk4f alphas() const;  // argb -> aaaa
, generally faster than equivalent Sk4f(this->a())

[mtklein, 2015-06-26 15:19:48]

The CQ bit was checked by mtklein@google.com to run a CQ dry run

[commit-bot: I haz the power, 2015-06-26 15:20:03]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1214443002/60001

[commit-bot: I haz the power, 2015-06-26 15:48:37]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-06-26 15:48:38]

Dry run: This issue passed the CQ dry run.

[mtklein_C, 2015-06-26 16:01:42]

mtklein@chromium.org changed reviewers:
+ reed@google.com
- mtklein@google.com

[reed1, 2015-06-26 17:17:59]

lgtm w/ nits

https://codereview.chromium.org/1214443002/diff/60001/src/core/Sk4pxXfermode.h
File src/core/Sk4pxXfermode.h (right):

https://codereview.chromium.org/1214443002/diff/60001/src/core/Sk4pxXfermode....
src/core/Sk4pxXfermode.h:113: auto sa = s.alphas(),
in general, I just love the expressiveness of SkPMFloat, etc.

https://codereview.chromium.org/1214443002/diff/60001/src/core/Sk4pxXfermode....
src/core/Sk4pxXfermode.h:121: auto colors = (d == Sk4f(0)).thenElse(dstover,
After staring for a while, it looks like this is a nested if else?

if ... then
   ...
else
    if ... then
        ...
    else
        ...

If so, I wonder if we can write this (and ColorBurn) in some different way to
make this tri-state-ness clearer.

[mtklein, 2015-06-26 17:39:55]

mtklein@google.com changed reviewers:
+ mtklein@google.com

[mtklein, 2015-06-26 17:39:56]

https://codereview.chromium.org/1214443002/diff/60001/src/core/Sk4pxXfermode.h
File src/core/Sk4pxXfermode.h (right):

https://codereview.chromium.org/1214443002/diff/60001/src/core/Sk4pxXfermode....
src/core/Sk4pxXfermode.h:121: auto colors = (d == Sk4f(0)).thenElse(dstover,
On 2015/06/26 17:17:58, reed1 wrote:
> After staring for a while, it looks like this is a nested if else?
> 
> if ... then
>    ...
> else
>     if ... then
>         ...
>     else
>         ...
> 
> If so, I wonder if we can write this (and ColorBurn) in some different way to
> make this tri-state-ness clearer.

Acknowledged.

[mtklein, 2015-06-26 17:40:03]

The CQ bit was checked by mtklein@google.com

[mtklein, 2015-06-26 17:40:03]

The patchset sent to the CQ was uploaded after l-g-t-m from reed@google.com
Link to the patchset: https://codereview.chromium.org/1214443002/#ps80001
(title: "reed")

[commit-bot: I haz the power, 2015-06-26 17:40:21]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1214443002/80001

[commit-bot: I haz the power, 2015-06-26 17:46:34]

Committed patchset #5 (id:80001) as
https://skia.googlesource.com/skia/+/2aab22a58a366df4752c1cf0f004092c6e7be335