linear -> sRGB: use fast approximate sqrt()

linear -> sRGB: use fast approximate sqrt()

Since we're already approximating the sRGB gamma curve with a sqrt(), we might
as well approximate with it a faster approximate sqrt().  On Intel, this
.rsqrt().invert() version is 2-3x faster than .sqrt()  (~3x faster on older
machines, ~2x faster on newer machines).

This should provide ~11 bits of precision, suspiciously exactly enough.

Running dm --config srgb, there are diffs, but none perceptible.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2046063002

Committed: https://skia.googlesource.com/skia/+/3db2028126e12a9c1c1fb2f878d552a2de0e1452

[mtklein_C, 2016-06-07 17:45:57]

Description was changed from

==========
linear -> sRGB: use fast approximate sqrt()

Since we're already approximating the sRGB gamma curve with a sqrt(), we might
as well approximate with it a faster approximate sqrt().  On Intel, this
.rsqrt().invert() version is 2-3x faster than .sqrt()  (~3x faster on older
machines, ~2x faster on newer machines).

This should provide ~11 bits of precision, suspiciously exactly enough.

Running dm --config srgb, there are diffs, but none perceptible.

BUG=skia:
==========

to

==========
linear -> sRGB: use fast approximate sqrt()

Since we're already approximating the sRGB gamma curve with a sqrt(), we might
as well approximate with it a faster approximate sqrt().  On Intel, this
.rsqrt().invert() version is 2-3x faster than .sqrt()  (~3x faster on older
machines, ~2x faster on newer machines).

This should provide ~11 bits of precision, suspiciously exactly enough.

Running dm --config srgb, there are diffs, but none perceptible.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2046063002
==========

[mtklein_C, 2016-06-07 17:49:14]

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

[mtklein_C, 2016-06-07 17:49:15]

Are there more performance-critical .sqrt()s for me to update?

[mtklein_C, 2016-06-07 17:49:21]

The CQ bit was checked by mtklein@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-06-07 17:49:36]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/2046063002/1

[commit-bot: I haz the power, 2016-06-07 18:05:03]

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

[commit-bot: I haz the power, 2016-06-07 18:05:03]

Dry run: This issue passed the CQ dry run.

[mtklein_C, 2016-06-07 18:23:27]

mtklein@chromium.org changed reviewers:
+ fmalita@chromium.org

[mtklein_C, 2016-06-07 18:23:28]

+Florin

Refining my grep found another spot in the linear gradient code.  I think this
is it.

[mtklein_C, 2016-06-07 18:23:33]

The CQ bit was checked by mtklein@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-06-07 18:23:43]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/2046063002/20001

[herb_g, 2016-06-07 18:35:47]

lgtm

[commit-bot: I haz the power, 2016-06-07 18:36:02]

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

[commit-bot: I haz the power, 2016-06-07 18:36:03]

Dry run: This issue passed the CQ dry run.

[reed1, 2016-06-07 19:08:15]

lgtm

[mtklein, 2016-06-07 19:11:34]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2016-06-07 19:11:45]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/2046063002/20001

[commit-bot: I haz the power, 2016-06-07 19:12:42]

Description was changed from

==========
linear -> sRGB: use fast approximate sqrt()

Since we're already approximating the sRGB gamma curve with a sqrt(), we might
as well approximate with it a faster approximate sqrt().  On Intel, this
.rsqrt().invert() version is 2-3x faster than .sqrt()  (~3x faster on older
machines, ~2x faster on newer machines).

This should provide ~11 bits of precision, suspiciously exactly enough.

Running dm --config srgb, there are diffs, but none perceptible.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2046063002
==========

to

==========
linear -> sRGB: use fast approximate sqrt()

Since we're already approximating the sRGB gamma curve with a sqrt(), we might
as well approximate with it a faster approximate sqrt().  On Intel, this
.rsqrt().invert() version is 2-3x faster than .sqrt()  (~3x faster on older
machines, ~2x faster on newer machines).

This should provide ~11 bits of precision, suspiciously exactly enough.

Running dm --config srgb, there are diffs, but none perceptible.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2046063002

Committed:
https://skia.googlesource.com/skia/+/3db2028126e12a9c1c1fb2f878d552a2de0e1452
==========

[commit-bot: I haz the power, 2016-06-07 19:12:43]

Committed patchset #2 (id:20001) as
https://skia.googlesource.com/skia/+/3db2028126e12a9c1c1fb2f878d552a2de0e1452

[f(malita), 2016-06-07 19:51:48]

LGTM!