Add SSE4 version of BlurImage optimizations.

Add SSE4 version of BlurImage optimizations.

Adds an SSE4.1 version of the existing BlurImage optimizations.
Performance of blur_image_filter_* benchmarks show a 10-50%
improvement on Linux/Ubuntu Core i7.

Signed-off-by: Henrik Smiding <henrik.smiding@intel.com>;

Committed: https://skia.googlesource.com/skia/+/2830632ce93c97ed7647b13348365ea92e4ea665

Committed: https://skia.googlesource.com/skia/+/5f7f9d04dc3a2d2c3ef9d8f1703d8e13c2d15c6e

[henrik.smiding, 2014-07-01 14:54:22]

Running into a problem when adding SSE4.x intrinsics. Same problem as the
previous attempts. Mac forces -mssse3 and overrides the -msse4 i set in the
opts_sse4 lib.
I've tried not setting -mssse3 for mac when building this lib, and I've tried
both removing and overriding the flag in opts.gyp, to no avail.
My knowledge of the build system is not enough to enable SSE4 intrinsics for
Mac. At least not in a nice way.
Any ideas?

[mtklein, 2014-07-01 16:41:20]

Other than the GYP and building on Mac, this LGTM.

https://codereview.chromium.org/366593004/diff/80001/gyp/opts.gyp
File gyp/opts.gyp (right):

https://codereview.chromium.org/366593004/diff/80001/gyp/opts.gyp#newcode220
gyp/opts.gyp:220: 'OTHER_CPLUSPLUSFLAGS': [
I think you will need to nest it inside 'xcode_settings':

'xcode_settings': {
  'OTHER_CPLUSPLUSFLAGS': [ '-msse4' ],
},

GYP is really weird.

https://codereview.chromium.org/366593004/diff/80001/gyp/opts.gyp#newcode222
gyp/opts.gyp:222: ],
If that doesn't work, try also adding

'OTHER_CPLUSPLUSFLAGS!': [ '-mssse3' ],

?

The ! at the end makes it a negative merge... should remove -mssse3 from the
commandline.

[henrik.smiding, 2014-07-02 12:55:14]

https://codereview.chromium.org/366593004/diff/80001/gyp/opts.gyp
File gyp/opts.gyp (right):

https://codereview.chromium.org/366593004/diff/80001/gyp/opts.gyp#newcode220
gyp/opts.gyp:220: 'OTHER_CPLUSPLUSFLAGS': [
On 2014/07/01 16:41:20, mtklein wrote:
> I think you will need to nest it inside 'xcode_settings':
> 
> 'xcode_settings': {
>   'OTHER_CPLUSPLUSFLAGS': [ '-msse4' ],
> },
> 
> GYP is really weird.

I'll try that!

https://codereview.chromium.org/366593004/diff/80001/gyp/opts.gyp#newcode222
gyp/opts.gyp:222: ],
On 2014/07/01 16:41:20, mtklein wrote:
> If that doesn't work, try also adding
> 
> 'OTHER_CPLUSPLUSFLAGS!': [ '-mssse3' ],
> 
> ?
> 
> The ! at the end makes it a negative merge... should remove -mssse3 from the
> commandline.

I tried that in Patch set #4. Not nested in xcode_settings though.

[henrik.smiding, 2014-07-03 11:14:42]

Seems to work fine now. Do you want to do a final check before i commit it?

[henrik.smiding, 2014-07-04 11:10:44]

The CQ bit was checked by henrik.smiding@intel.com

[commit-bot: I haz the power, 2014-07-04 11:11:12]

CQ is trying da patch. Follow status at

https://skia-tree-status.appspot.com/cq/henrik.smiding@intel.com/366593004/13...

[commit-bot: I haz the power, 2014-07-04 11:23:22]

Change committed as 2830632ce93c97ed7647b13348365ea92e4ea665

[reed2, 2014-07-07 01:50:24]

A revert of this CL has been created in
https://codereview.chromium.org/375503003/ by reed@chromium.org.

The reason for reverting is: breaks linker on chrome

[04:36:09.966000] [503/5965] LIB obj\chrome\installer_util.lib
[04:36:10.466000] FAILED:
C:\Users\chrome-bot\buildbot\third_party\depot_tools\python276_bin\python.exe
gyp-win-tool link-with-manifests environment.x86 True skia.dll
"C:\Users\chrome-bot\buildbot\third_party\depot_tools\python276_bin\python.exe
gyp-win-tool link-wrapper environment.x86 False link.exe /nologo
/IMPLIB:skia.dll.lib /DLL /OUT:skia.dll @skia.dll.rsp" 2 mt.exe rc.exe
"obj\skia\skia.skia.dll.intermediate.manifest"
obj\skia\skia.skia.dll.generated.manifest
[04:36:10.466000] skia.opts_check_x86.obj : error LNK2019: unresolved external
symbol "bool __cdecl SkBoxBlurGetPlatformProcs_SSE4(void (__cdecl**)(unsigned
int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned
int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned
int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned
int const *,int,unsigned int *,int,int,int,int,int))"
(?SkBoxBlurGetPlatformProcs_SSE4@@YA_NPAP6AXPBIHPAIHHHHH@Z222@Z) referenced in
function "bool __cdecl SkBoxBlurGetPlatformProcs(void (__cdecl**)(unsigned int
const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int
const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int
const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int
const *,int,unsigned int *,int,int,int,int,int))"
(?SkBoxBlurGetPlatformProcs@@YA_NPAP6AXPBIHPAIHHHHH@Z222@Z)
[04:36:10.466000] 
[04:36:10.466000] skia.dll : fatal error LNK1120: 1 unresolved externals
.

[henrik.smiding, 2014-07-07 11:09:37]

On 2014/07/07 01:50:24, reed2 wrote:
> A revert of this CL has been created in
> https://codereview.chromium.org/375503003/ by mailto:reed@chromium.org.
> 
> The reason for reverting is: breaks linker on chrome
> 
> [04:36:09.966000] [503/5965] LIB obj\chrome\installer_util.lib
> [04:36:10.466000] FAILED:
> C:\Users\chrome-bot\buildbot\third_party\depot_tools\python276_bin\python.exe
> gyp-win-tool link-with-manifests environment.x86 True skia.dll
> "C:\Users\chrome-bot\buildbot\third_party\depot_tools\python276_bin\python.exe
> gyp-win-tool link-wrapper environment.x86 False link.exe /nologo
> /IMPLIB:skia.dll.lib /DLL /OUT:skia.dll @skia.dll.rsp" 2 mt.exe rc.exe
> "obj\skia\skia.skia.dll.intermediate.manifest"
> obj\skia\skia.skia.dll.generated.manifest
> [04:36:10.466000] skia.opts_check_x86.obj : error LNK2019: unresolved external
> symbol "bool __cdecl SkBoxBlurGetPlatformProcs_SSE4(void (__cdecl**)(unsigned
> int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned
> int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned
> int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned
> int const *,int,unsigned int *,int,int,int,int,int))"
> (?SkBoxBlurGetPlatformProcs_SSE4@@YA_NPAP6AXPBIHPAIHHHHH@Z222@Z) referenced in
> function "bool __cdecl SkBoxBlurGetPlatformProcs(void (__cdecl**)(unsigned int
> const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int
> const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int
> const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int
> const *,int,unsigned int *,int,int,int,int,int))"
> (?SkBoxBlurGetPlatformProcs@@YA_NPAP6AXPBIHPAIHHHHH@Z222@Z)
> [04:36:10.466000] 
> [04:36:10.466000] skia.dll : fatal error LNK1120: 1 unresolved externals
> .

It's not just that the new files are missing in Chrome's build scripts?

[mtklein, 2014-07-07 14:38:27]

On 2014/07/07 11:09:37, henrik.smiding wrote:
> On 2014/07/07 01:50:24, reed2 wrote:
> > A revert of this CL has been created in
> > https://codereview.chromium.org/375503003/ by mailto:reed@chromium.org.
> > 
> > The reason for reverting is: breaks linker on chrome
> > 
> > [04:36:09.966000] [503/5965] LIB obj\chrome\installer_util.lib
> > [04:36:10.466000] FAILED:
> >
C:\Users\chrome-bot\buildbot\third_party\depot_tools\python276_bin\python.exe
> > gyp-win-tool link-with-manifests environment.x86 True skia.dll
> >
"C:\Users\chrome-bot\buildbot\third_party\depot_tools\python276_bin\python.exe
> > gyp-win-tool link-wrapper environment.x86 False link.exe /nologo
> > /IMPLIB:skia.dll.lib /DLL /OUT:skia.dll @skia.dll.rsp" 2 mt.exe rc.exe
> > "obj\skia\skia.skia.dll.intermediate.manifest"
> > obj\skia\skia.skia.dll.generated.manifest
> > [04:36:10.466000] skia.opts_check_x86.obj : error LNK2019: unresolved
external
> > symbol "bool __cdecl SkBoxBlurGetPlatformProcs_SSE4(void
(__cdecl**)(unsigned
> > int const *,int,unsigned int *,int,int,int,int,int),void
(__cdecl**)(unsigned
> > int const *,int,unsigned int *,int,int,int,int,int),void
(__cdecl**)(unsigned
> > int const *,int,unsigned int *,int,int,int,int,int),void
(__cdecl**)(unsigned
> > int const *,int,unsigned int *,int,int,int,int,int))"
> > (?SkBoxBlurGetPlatformProcs_SSE4@@YA_NPAP6AXPBIHPAIHHHHH@Z222@Z) referenced
in
> > function "bool __cdecl SkBoxBlurGetPlatformProcs(void (__cdecl**)(unsigned
int
> > const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned
int
> > const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned
int
> > const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned
int
> > const *,int,unsigned int *,int,int,int,int,int))"
> > (?SkBoxBlurGetPlatformProcs@@YA_NPAP6AXPBIHPAIHHHHH@Z222@Z)
> > [04:36:10.466000] 
> > [04:36:10.466000] skia.dll : fatal error LNK1120: 1 unresolved externals
> > .
> 
> It's not just that the new files are missing in Chrome's build scripts?

Yes, I think so.  I will go coordinate landing this with our tree sheriff.

[mtklein, 2014-07-07 15:04:27]

On 2014/07/07 14:38:27, mtklein wrote:
> On 2014/07/07 11:09:37, henrik.smiding wrote:
> > On 2014/07/07 01:50:24, reed2 wrote:
> > > A revert of this CL has been created in
> > > https://codereview.chromium.org/375503003/ by mailto:reed@chromium.org.
> > > 
> > > The reason for reverting is: breaks linker on chrome
> > > 
> > > [04:36:09.966000] [503/5965] LIB obj\chrome\installer_util.lib
> > > [04:36:10.466000] FAILED:
> > >
> C:\Users\chrome-bot\buildbot\third_party\depot_tools\python276_bin\python.exe
> > > gyp-win-tool link-with-manifests environment.x86 True skia.dll
> > >
> "C:\Users\chrome-bot\buildbot\third_party\depot_tools\python276_bin\python.exe
> > > gyp-win-tool link-wrapper environment.x86 False link.exe /nologo
> > > /IMPLIB:skia.dll.lib /DLL /OUT:skia.dll @skia.dll.rsp" 2 mt.exe rc.exe
> > > "obj\skia\skia.skia.dll.intermediate.manifest"
> > > obj\skia\skia.skia.dll.generated.manifest
> > > [04:36:10.466000] skia.opts_check_x86.obj : error LNK2019: unresolved
> external
> > > symbol "bool __cdecl SkBoxBlurGetPlatformProcs_SSE4(void
> (__cdecl**)(unsigned
> > > int const *,int,unsigned int *,int,int,int,int,int),void
> (__cdecl**)(unsigned
> > > int const *,int,unsigned int *,int,int,int,int,int),void
> (__cdecl**)(unsigned
> > > int const *,int,unsigned int *,int,int,int,int,int),void
> (__cdecl**)(unsigned
> > > int const *,int,unsigned int *,int,int,int,int,int))"
> > > (?SkBoxBlurGetPlatformProcs_SSE4@@YA_NPAP6AXPBIHPAIHHHHH@Z222@Z)
referenced
> in
> > > function "bool __cdecl SkBoxBlurGetPlatformProcs(void (__cdecl**)(unsigned
> int
> > > const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned
> int
> > > const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned
> int
> > > const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned
> int
> > > const *,int,unsigned int *,int,int,int,int,int))"
> > > (?SkBoxBlurGetPlatformProcs@@YA_NPAP6AXPBIHPAIHHHHH@Z222@Z)
> > > [04:36:10.466000] 
> > > [04:36:10.466000] skia.dll : fatal error LNK1120: 1 unresolved externals
> > > .
> > 
> > It's not just that the new files are missing in Chrome's build scripts?
> 
> Yes, I think so.  I will go coordinate landing this with our tree sheriff.

Relanding this.  I have Chrome building now with
https://codereview.chromium.org/378523003/ patched in.

[mtklein, 2014-07-07 15:04:32]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-07-07 15:05:34]

CQ is trying da patch. Follow status at

https://skia-tree-status.appspot.com/cq/henrik.smiding@intel.com/366593004/13...

[commit-bot: I haz the power, 2014-07-07 15:05:44]

Change committed as 5f7f9d04dc3a2d2c3ef9d8f1703d8e13c2d15c6e

[henrik.smiding, 2014-07-08 08:13:42]

On 2014/07/07 15:05:44, I haz the power (commit-bot) wrote:
> Change committed as 5f7f9d04dc3a2d2c3ef9d8f1703d8e13c2d15c6e

Thanks!