From a2ea90567998b1ab93ce7fe3acc25922862e4c9c Mon Sep 17 00:00:00 2001
From: Frank Barchard <fbarchard@google.com>
Date: Tue, 8 Dec 2015 18:59:48 -0800
Subject: [PATCH] BlendPlane any width.

Benchmark
out\release\libyuv_unittest --libyuv_width=1279 --libyuv_height=719 --libyuv_repeat=999 --libyuv_flags=-1 --gtest_filter=*Blend* | sortms

Was
I420Blend_Any (2321 ms)
I420Blend_Unaligned (1684 ms)
I420Blend_Opt (1675 ms)
I420Blend_Invert (1653 ms)
BlendPlane_Invert (1556 ms)
BlendPlane_Any (1552 ms)
BlendPlane_Unaligned (1548 ms)
BlendPlane_Opt (1535 ms)
ARGBBlend_Unaligned (659 ms)
ARGBBlend_Any (596 ms)
ARGBBlend_Invert (591 ms)
ARGBBlend_Opt (508 ms)
BlendPlaneRow_Unaligned (186 ms)
BlendPlaneRow_Opt (171 ms)

Now
ARGBBlend_Any (621 ms)
ARGBBlend_Unaligned (585 ms)
ARGBBlend_Invert (564 ms)
ARGBBlend_Opt (512 ms)
I420Blend_Unaligned (347 ms)
I420Blend_Invert (345 ms)
I420Blend_Any (337 ms)
I420Blend_Opt (327 ms)
BlendPlane_Unaligned (187 ms)
BlendPlaneRow_Unaligned (187 ms)
BlendPlane_Invert (186 ms)
BlendPlane_Any (186 ms)
BlendPlaneRow_Opt (173 ms)
BlendPlane_Opt (171 ms)

which is comparable to aligned case
out\release\libyuv_unittest --libyuv_width=1280 --libyuv_height=720 --libyuv_repeat=999 --libyuv_flags=-1 --gtest_filter=*Blend* | sortms
ARGBBlend_Any (625 ms)
ARGBBlend_Unaligned (602 ms)
ARGBBlend_Invert (508 ms)
ARGBBlend_Opt (506 ms)
I420Blend_Any (353 ms)
I420Blend_Unaligned (322 ms)
I420Blend_Invert (304 ms)
I420Blend_Opt (301 ms)
BlendPlaneRow_Unaligned (188 ms)
BlendPlane_Unaligned (186 ms)
BlendPlane_Invert (185 ms)
BlendPlane_Any (184 ms)
BlendPlaneRow_Opt (173 ms)
BlendPlane_Opt (169 ms)

R=dhrosa@google.com, harryjin@google.com
BUG=libyuv:527

Review URL: https://codereview.chromium.org/1513443002 .
---
 README.chromium            |  2 +-
 include/libyuv/version.h   |  2 +-
 source/planar_functions.cc | 10 ++++------
 source/row_any.cc          |  6 ++++++
 4 files changed, 12 insertions(+), 8 deletions(-)

diff --git a/README.chromium b/README.chromium
index 811ec9b52..e76949316 100644
--- a/README.chromium
+++ b/README.chromium
@@ -1,6 +1,6 @@
 Name: libyuv
 URL: http://code.google.com/p/libyuv/
-Version: 1550
+Version: 1551
 License: BSD
 License File: LICENSE
 
diff --git a/include/libyuv/version.h b/include/libyuv/version.h
index 57e0c0a05..93cbb7868 100644
--- a/include/libyuv/version.h
+++ b/include/libyuv/version.h
@@ -11,6 +11,6 @@
 #ifndef INCLUDE_LIBYUV_VERSION_H_  // NOLINT
 #define INCLUDE_LIBYUV_VERSION_H_
 
-#define LIBYUV_VERSION 1550
+#define LIBYUV_VERSION 1551
 
 #endif  // INCLUDE_LIBYUV_VERSION_H_  NOLINT
diff --git a/source/planar_functions.cc b/source/planar_functions.cc
index c73fa8ab4..2731bd0da 100644
--- a/source/planar_functions.cc
+++ b/source/planar_functions.cc
@@ -610,8 +610,7 @@ int BlendPlane(const uint8* src_y0, int src_stride_y0,
 
 #if defined(HAS_BLENDPLANEROW_SSSE3)
   if (TestCpuFlag(kCpuHasSSSE3)) {
-// TODO(fbarchard): Implement any versions for odd width.
-//  BlendPlaneRow = BlendPlaneRow_Any_SSSE3;
+  BlendPlaneRow = BlendPlaneRow_Any_SSSE3;
     if (IS_ALIGNED(width, 8)) {
       BlendPlaneRow = BlendPlaneRow_SSSE3;
     }
@@ -619,7 +618,7 @@ int BlendPlane(const uint8* src_y0, int src_stride_y0,
 #endif
 #if defined(HAS_BLENDPLANEROW_AVX2)
   if (TestCpuFlag(kCpuHasAVX2)) {
-//  BlendPlaneRow = BlendPlaneRow_Any_AVX2;
+  BlendPlaneRow = BlendPlaneRow_Any_AVX2;
     if (IS_ALIGNED(width, 32)) {
       BlendPlaneRow = BlendPlaneRow_AVX2;
     }
@@ -678,8 +677,7 @@ int I420Blend(const uint8* src_y0, int src_stride_y0,
 
 #if defined(HAS_BLENDPLANEROW_SSSE3)
   if (TestCpuFlag(kCpuHasSSSE3)) {
-// TODO(fbarchard): Implement any versions for odd width.
-//  BlendPlaneRow = BlendPlaneRow_Any_SSSE3;
+  BlendPlaneRow = BlendPlaneRow_Any_SSSE3;
     if (IS_ALIGNED(halfwidth, 8)) {
       BlendPlaneRow = BlendPlaneRow_SSSE3;
     }
@@ -687,7 +685,7 @@ int I420Blend(const uint8* src_y0, int src_stride_y0,
 #endif
 #if defined(HAS_BLENDPLANEROW_AVX2)
   if (TestCpuFlag(kCpuHasAVX2)) {
-//  BlendPlaneRow = BlendPlaneRow_Any_AVX2;
+  BlendPlaneRow = BlendPlaneRow_Any_AVX2;
     if (IS_ALIGNED(halfwidth, 32)) {
       BlendPlaneRow = BlendPlaneRow_AVX2;
     }
diff --git a/source/row_any.cc b/source/row_any.cc
index 511ca1cc8..fef7ecd9a 100644
--- a/source/row_any.cc
+++ b/source/row_any.cc
@@ -83,6 +83,12 @@ ANY31(I422ToYUY2Row_Any_NEON, I422ToYUY2Row_NEON, 1, 1, 4, 15)
 #ifdef HAS_I422TOUYVYROW_NEON
 ANY31(I422ToUYVYRow_Any_NEON, I422ToUYVYRow_NEON, 1, 1, 4, 15)
 #endif
+#ifdef HAS_BLENDPLANEROW_AVX2
+ANY31(BlendPlaneRow_Any_AVX2, BlendPlaneRow_AVX2, 0, 0, 1, 31)
+#endif
+#ifdef HAS_BLENDPLANEROW_SSSE3
+ANY31(BlendPlaneRow_Any_SSSE3, BlendPlaneRow_SSSE3, 0, 0, 1, 7)
+#endif
 #undef ANY31
 
 // Note that odd width replication includes 444 due to implementation