Add MSA (MIPS SIMD Arch) optimized WebGL image conversion functions

Add MSA (MIPS SIMD Arch) optimized WebGL image conversion functions

We add following MSA optimized WebGL image conversion functions:
- unpackOneRowOfBGRA8LittleToRGBA8MSA
- unpackOneRowOfRGBA4444ToRGBA8MSA
- packOneRowOfRGBA8LittleToRGBA8MSA
- packOneRowOfRGBA8ToUnsignedShort5551MSA
- packOneRowOfRGBA8ToUnsignedShort565MSA

BUG=642653
Committed: https://crrev.com/0b87fed4b772f1203c6178c5f0645b7790c61740
Cr-Commit-Position: refs/heads/master@{#418183}

[Prashant.Patil, 2016-09-02 15:34:14]

Description was changed from

==========
Add MSA (MIPS SIMD Arch) optimized WebGL image conversion functions

We add following MSA optimized WebGL image conversion functions:
- unpackOneRowOfBGRA8LittleToRGBA8MSA
- unpackOneRowOfRGBA4444ToRGBA8MSA
- packOneRowOfRGBA8LittleToRGBA8MSA
- packOneRowOfRGBA8ToUnsignedShort5551MSA
- packOneRowOfRGBA8ToUnsignedShort565MSA

R=dpranke@chromium.org
BUG=642653
==========

to

==========
Add MSA (MIPS SIMD Arch) optimized WebGL image conversion functions

We add following MSA optimized WebGL image conversion functions:
- unpackOneRowOfBGRA8LittleToRGBA8MSA
- unpackOneRowOfRGBA4444ToRGBA8MSA
- packOneRowOfRGBA8LittleToRGBA8MSA
- packOneRowOfRGBA8ToUnsignedShort5551MSA
- packOneRowOfRGBA8ToUnsignedShort565MSA

R=dpranke@chromium.org
BUG=642653
==========

[Prashant.Patil, 2016-09-02 15:34:14]

prashant.patil@imgtec.com changed reviewers:
+ kbr@chromium.org, schenney@chromium.org, zmo@chromium.org

[Dirk Pranke, 2016-09-02 16:51:42]

Description was changed from

==========
Add MSA (MIPS SIMD Arch) optimized WebGL image conversion functions

We add following MSA optimized WebGL image conversion functions:
- unpackOneRowOfBGRA8LittleToRGBA8MSA
- unpackOneRowOfRGBA4444ToRGBA8MSA
- packOneRowOfRGBA8LittleToRGBA8MSA
- packOneRowOfRGBA8ToUnsignedShort5551MSA
- packOneRowOfRGBA8ToUnsignedShort565MSA

R=dpranke@chromium.org
BUG=642653
==========

to

==========
Add MSA (MIPS SIMD Arch) optimized WebGL image conversion functions

We add following MSA optimized WebGL image conversion functions:
- unpackOneRowOfBGRA8LittleToRGBA8MSA
- unpackOneRowOfRGBA4444ToRGBA8MSA
- packOneRowOfRGBA8LittleToRGBA8MSA
- packOneRowOfRGBA8ToUnsignedShort5551MSA
- packOneRowOfRGBA8ToUnsignedShort565MSA


BUG=642653
==========

[Dirk Pranke, 2016-09-02 16:51:42]

dpranke@chromium.org changed reviewers:
- dpranke@chromium.org

[Dirk Pranke, 2016-09-02 16:51:57]

dpranke@chromium.org changed reviewers:
+ dpranke@chromium.org

[Dirk Pranke, 2016-09-02 16:51:58]

I'm not really a good reviewer for this ...

[Ken Russell (switch to Gerrit), 2016-09-07 03:21:51]

Thanks for this contribution. It's a nice piece of work.

I wonder about the maintainability of this code. This is an awful lot of
handwritten assembly. Are you sure you want to maintain it in this form? Can you
think of any way to refactor it to generate more of the code from smaller
functions?

Are you sure that the WebGL conformance tests are sufficient to cover all of
these new code paths?

[Prashant.Patil, 2016-09-12 06:13:39]

On 2016/09/07 03:21:51, Ken Russell wrote:
> Thanks for this contribution. It's a nice piece of work.
> 
> I wonder about the maintainability of this code. This is an awful lot of
> handwritten assembly. Are you sure you want to maintain it in this form? Can
you
> think of any way to refactor it to generate more of the code from smaller
> functions?
> 
> Are you sure that the WebGL conformance tests are sufficient to cover all of
> these new code paths?

This is intrinsic based code(and not handwritten assembly) which would run on
all MIPS-MSA current and future targets. 
The code appears bigger because of unrolling and thereby using of all 32 vector
registers optimally at the same time also covering all the corner cases. 
Additionally we are going to add WebGLImageConversionTest in
blink_platform_unittests  to cover these kernels.

We are happy to maintain MSA code.

[Ken Russell (switch to Gerrit), 2016-09-12 19:43:55]

On 2016/09/12 06:13:39, Prashant.Patil wrote:
> On 2016/09/07 03:21:51, Ken Russell wrote:
> > Thanks for this contribution. It's a nice piece of work.
> > 
> > I wonder about the maintainability of this code. This is an awful lot of
> > handwritten assembly. Are you sure you want to maintain it in this form? Can
> you
> > think of any way to refactor it to generate more of the code from smaller
> > functions?
> > 
> > Are you sure that the WebGL conformance tests are sufficient to cover all of
> > these new code paths?
> 
> This is intrinsic based code(and not handwritten assembly) which would run on
> all MIPS-MSA current and future targets. 
> The code appears bigger because of unrolling and thereby using of all 32
vector
> registers optimally at the same time also covering all the corner cases. 
> Additionally we are going to add WebGLImageConversionTest in
> blink_platform_unittests  to cover these kernels.
> 
> We are happy to maintain MSA code.

OK. Thanks for your feedback. LGTM

[Prashant.Patil, 2016-09-13 04:46:21]

The CQ bit was checked by prashant.patil@imgtec.com

[commit-bot: I haz the power, 2016-09-13 04:46:29]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-09-13 06:14:41]

Description was changed from

==========
Add MSA (MIPS SIMD Arch) optimized WebGL image conversion functions

We add following MSA optimized WebGL image conversion functions:
- unpackOneRowOfBGRA8LittleToRGBA8MSA
- unpackOneRowOfRGBA4444ToRGBA8MSA
- packOneRowOfRGBA8LittleToRGBA8MSA
- packOneRowOfRGBA8ToUnsignedShort5551MSA
- packOneRowOfRGBA8ToUnsignedShort565MSA


BUG=642653
==========

to

==========
Add MSA (MIPS SIMD Arch) optimized WebGL image conversion functions

We add following MSA optimized WebGL image conversion functions:
- unpackOneRowOfBGRA8LittleToRGBA8MSA
- unpackOneRowOfRGBA4444ToRGBA8MSA
- packOneRowOfRGBA8LittleToRGBA8MSA
- packOneRowOfRGBA8ToUnsignedShort5551MSA
- packOneRowOfRGBA8ToUnsignedShort565MSA


BUG=642653
==========

[commit-bot: I haz the power, 2016-09-13 06:14:43]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-09-13 06:16:40]

Description was changed from

==========
Add MSA (MIPS SIMD Arch) optimized WebGL image conversion functions

We add following MSA optimized WebGL image conversion functions:
- unpackOneRowOfBGRA8LittleToRGBA8MSA
- unpackOneRowOfRGBA4444ToRGBA8MSA
- packOneRowOfRGBA8LittleToRGBA8MSA
- packOneRowOfRGBA8ToUnsignedShort5551MSA
- packOneRowOfRGBA8ToUnsignedShort565MSA


BUG=642653
==========

to

==========
Add MSA (MIPS SIMD Arch) optimized WebGL image conversion functions

We add following MSA optimized WebGL image conversion functions:
- unpackOneRowOfBGRA8LittleToRGBA8MSA
- unpackOneRowOfRGBA4444ToRGBA8MSA
- packOneRowOfRGBA8LittleToRGBA8MSA
- packOneRowOfRGBA8ToUnsignedShort5551MSA
- packOneRowOfRGBA8ToUnsignedShort565MSA

BUG=642653

Committed: https://crrev.com/0b87fed4b772f1203c6178c5f0645b7790c61740
Cr-Commit-Position: refs/heads/master@{#418183}
==========

[commit-bot: I haz the power, 2016-09-13 06:16:41]

Patchset 1 (id:??) landed as
https://crrev.com/0b87fed4b772f1203c6178c5f0645b7790c61740
Cr-Commit-Position: refs/heads/master@{#418183}