Commit 95730719 authored by fbarchard@google.com's avatar fbarchard@google.com

Neon RGB24 to I420

BUG=none
TEST=convert_test
Review URL: https://webrtc-codereview.appspot.com/965018

git-svn-id: http://libyuv.googlecode.com/svn/trunk@481 16f28f9a-4ce2-e073-06de-1de4eb20be90
parent 522d757c
Name: libyuv
URL: http://code.google.com/p/libyuv/
Version: 480
Version: 481
License: BSD
License File: LICENSE
......
......@@ -362,6 +362,16 @@ void ARGBToUV411Row_NEON(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
int pix);
void ARGBToUVRow_NEON(const uint8* src_argb, int src_stride_argb,
uint8* dst_u, uint8* dst_v, int pix);
void BGRAToUVRow_NEON(const uint8* src_bgra, int src_stride_bgra,
uint8* dst_u, uint8* dst_v, int pix);
void ABGRToUVRow_NEON(const uint8* src_abgr, int src_stride_abgr,
uint8* dst_u, uint8* dst_v, int pix);
void RGBAToUVRow_NEON(const uint8* src_rgba, int src_stride_rgba,
uint8* dst_u, uint8* dst_v, int pix);
void RGB24ToUVRow_NEON(const uint8* src_rgb24, int src_stride_rgb24,
uint8* dst_u, uint8* dst_v, int pix);
void RAWToUVRow_NEON(const uint8* src_raw, int src_stride_raw,
uint8* dst_u, uint8* dst_v, int pix);
void RGB565ToUVRow_NEON(const uint8* src_rgb565, int src_stride_rgb565,
uint8* dst_u, uint8* dst_v, int pix);
void ARGB1555ToUVRow_NEON(const uint8* src_argb1555, int src_stride_argb1555,
......@@ -433,6 +443,16 @@ void ARGBToUV411Row_Any_NEON(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
int pix);
void ARGBToUVRow_Any_NEON(const uint8* src_argb, int src_stride_argb,
uint8* dst_u, uint8* dst_v, int pix);
void BGRAToUVRow_Any_NEON(const uint8* src_bgra, int src_stride_bgra,
uint8* dst_u, uint8* dst_v, int pix);
void ABGRToUVRow_Any_NEON(const uint8* src_abgr, int src_stride_abgr,
uint8* dst_u, uint8* dst_v, int pix);
void RGBAToUVRow_Any_NEON(const uint8* src_rgba, int src_stride_rgba,
uint8* dst_u, uint8* dst_v, int pix);
void RGB24ToUVRow_Any_NEON(const uint8* src_rgb24, int src_stride_rgb24,
uint8* dst_u, uint8* dst_v, int pix);
void RAWToUVRow_Any_NEON(const uint8* src_raw, int src_stride_raw,
uint8* dst_u, uint8* dst_v, int pix);
void RGB565ToUVRow_Any_NEON(const uint8* src_rgb565, int src_stride_rgb565,
uint8* dst_u, uint8* dst_v, int pix);
void ARGB1555ToUVRow_Any_NEON(const uint8* src_argb1555,
......@@ -449,6 +469,10 @@ void ABGRToUVRow_C(const uint8* src_abgr, int src_stride_abgr,
uint8* dst_u, uint8* dst_v, int width);
void RGBAToUVRow_C(const uint8* src_rgba, int src_stride_rgba,
uint8* dst_u, uint8* dst_v, int width);
void RGB24ToUVRow_C(const uint8* src_rgb24, int src_stride_rgb24,
uint8* dst_u, uint8* dst_v, int width);
void RAWToUVRow_C(const uint8* src_raw, int src_stride_raw,
uint8* dst_u, uint8* dst_v, int width);
void RGB565ToUVRow_C(const uint8* src_rgb565, int src_stride_rgb565,
uint8* dst_u, uint8* dst_v, int width);
void ARGB1555ToUVRow_C(const uint8* src_argb1555, int src_stride_argb1555,
......
......@@ -11,6 +11,6 @@
#ifndef INCLUDE_LIBYUV_VERSION_H_ // NOLINT
#define INCLUDE_LIBYUV_VERSION_H_
#define LIBYUV_VERSION 480
#define LIBYUV_VERSION 481
#endif // INCLUDE_LIBYUV_VERSION_H_ NOLINT
......@@ -1012,6 +1012,12 @@ int BGRAToI420(const uint8* src_bgra, int src_stride_bgra,
if (IS_ALIGNED(width, 8)) {
BGRAToYRow = BGRAToYRow_NEON;
}
if (width >= 16) {
BGRAToUVRow = BGRAToUVRow_Any_NEON;
if (IS_ALIGNED(width, 16)) {
BGRAToUVRow = BGRAToUVRow_NEON;
}
}
}
#endif
......@@ -1074,6 +1080,12 @@ int ABGRToI420(const uint8* src_abgr, int src_stride_abgr,
if (IS_ALIGNED(width, 8)) {
ABGRToYRow = ABGRToYRow_NEON;
}
if (width >= 16) {
ABGRToUVRow = ABGRToUVRow_Any_NEON;
if (IS_ALIGNED(width, 16)) {
ABGRToUVRow = ABGRToUVRow_NEON;
}
}
}
#endif
......@@ -1136,6 +1148,12 @@ int RGBAToI420(const uint8* src_rgba, int src_stride_rgba,
if (IS_ALIGNED(width, 8)) {
RGBAToYRow = RGBAToYRow_NEON;
}
if (width >= 16) {
RGBAToUVRow = RGBAToUVRow_Any_NEON;
if (IS_ALIGNED(width, 16)) {
RGBAToUVRow = RGBAToUVRow_NEON;
}
}
}
#endif
......@@ -1173,6 +1191,25 @@ int RGB24ToI420(const uint8* src_rgb24, int src_stride_rgb24,
src_rgb24 = src_rgb24 + (height - 1) * src_stride_rgb24;
src_stride_rgb24 = -src_stride_rgb24;
}
#if defined(HAS_RGB24TOYROW_NEON)
void (*RGB24ToUVRow)(const uint8* src_rgb24, int src_stride_rgb24,
uint8* dst_u, uint8* dst_v, int width) = RGB24ToUVRow_C;
void (*RGB24ToYRow)(const uint8* src_rgb24, uint8* dst_y, int pix) =
RGB24ToYRow_C;
if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
RGB24ToYRow = RGB24ToYRow_Any_NEON;
if (IS_ALIGNED(width, 8)) {
RGB24ToYRow = RGB24ToYRow_NEON;
}
if (width >= 16) {
RGB24ToUVRow = RGB24ToUVRow_Any_NEON;
if (IS_ALIGNED(width, 16)) {
RGB24ToUVRow = RGB24ToUVRow_NEON;
}
}
}
#else // HAS_RGB24TOYROW_NEON
SIMD_ALIGNED(uint8 row[kMaxStride * 2]);
void (*RGB24ToARGBRow)(const uint8* src_rgb, uint8* dst_argb, int pix) =
RGB24ToARGBRow_C;
......@@ -1183,15 +1220,7 @@ int RGB24ToI420(const uint8* src_rgb24, int src_stride_rgb24,
RGB24ToARGBRow = RGB24ToARGBRow_SSSE3;
}
}
#elif defined(HAS_RGB24TOARGBROW_NEON)
if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
RGB24ToARGBRow = RGB24ToARGBRow_Any_NEON;
if (IS_ALIGNED(width, 8)) {
RGB24ToARGBRow = RGB24ToARGBRow_NEON;
}
}
#endif
void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
#if defined(HAS_ARGBTOUVROW_SSSE3)
......@@ -1202,23 +1231,6 @@ int RGB24ToI420(const uint8* src_rgb24, int src_stride_rgb24,
}
}
#endif
#if defined(HAS_RGB24TOYROW_NEON)
void (*RGB24ToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
RGB24ToYRow_C;
if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
RGB24ToYRow = RGB24ToYRow_Any_NEON;
if (IS_ALIGNED(width, 8)) {
RGB24ToYRow = RGB24ToYRow_NEON;
}
if (width >= 16) {
ARGBToUVRow = ARGBToUVRow_Any_NEON;
if (IS_ALIGNED(width, 16)) {
ARGBToUVRow = ARGBToUVRow_NEON;
}
}
}
#else
void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
ARGBToYRow_C;
#if defined(HAS_ARGBTOUVROW_SSSE3)
......@@ -1235,13 +1247,14 @@ int RGB24ToI420(const uint8* src_rgb24, int src_stride_rgb24,
#endif // HAS_RGB24TOYROW_NEON
for (int y = 0; y < height - 1; y += 2) {
RGB24ToARGBRow(src_rgb24, row, width);
RGB24ToARGBRow(src_rgb24 + src_stride_rgb24, row + kMaxStride, width);
ARGBToUVRow(row, kMaxStride, dst_u, dst_v, width);
#if defined(HAS_RGB24TOYROW_NEON)
RGB24ToUVRow(src_rgb24, src_stride_rgb24, dst_u, dst_v, width);
RGB24ToYRow(src_rgb24, dst_y, width);
RGB24ToYRow(src_rgb24 + src_stride_rgb24, dst_y + dst_stride_y, width);
#else
RGB24ToARGBRow(src_rgb24, row, width);
RGB24ToARGBRow(src_rgb24 + src_stride_rgb24, row + kMaxStride, width);
ARGBToUVRow(row, kMaxStride, dst_u, dst_v, width);
ARGBToYRow(row, dst_y, width);
ARGBToYRow(row + kMaxStride, dst_y + dst_stride_y, width);
#endif
......@@ -1251,11 +1264,12 @@ int RGB24ToI420(const uint8* src_rgb24, int src_stride_rgb24,
dst_v += dst_stride_v;
}
if (height & 1) {
RGB24ToARGBRow_C(src_rgb24, row, width);
ARGBToUVRow(row, 0, dst_u, dst_v, width);
#if defined(HAS_RGB24TOYROW_NEON)
RGB24ToUVRow(src_rgb24, 0, dst_u, dst_v, width);
RGB24ToYRow(src_rgb24, dst_y, width);
#else
RGB24ToARGBRow(src_rgb24, row, width);
ARGBToUVRow(row, 0, dst_u, dst_v, width);
ARGBToYRow(row, dst_y, width);
#endif
}
......@@ -1263,7 +1277,6 @@ int RGB24ToI420(const uint8* src_rgb24, int src_stride_rgb24,
}
// Convert RAW to I420.
// Same as RGB24 but RGB vs BGR
LIBYUV_API
int RAWToI420(const uint8* src_raw, int src_stride_raw,
uint8* dst_y, int dst_stride_y,
......@@ -1281,6 +1294,25 @@ int RAWToI420(const uint8* src_raw, int src_stride_raw,
src_raw = src_raw + (height - 1) * src_stride_raw;
src_stride_raw = -src_stride_raw;
}
#if defined(HAS_RAWTOYROW_NEON)
void (*RAWToUVRow)(const uint8* src_raw, int src_stride_raw,
uint8* dst_u, uint8* dst_v, int width) = RAWToUVRow_C;
void (*RAWToYRow)(const uint8* src_raw, uint8* dst_y, int pix) =
RAWToYRow_C;
if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
RAWToYRow = RAWToYRow_Any_NEON;
if (IS_ALIGNED(width, 8)) {
RAWToYRow = RAWToYRow_NEON;
}
if (width >= 16) {
RAWToUVRow = RAWToUVRow_Any_NEON;
if (IS_ALIGNED(width, 16)) {
RAWToUVRow = RAWToUVRow_NEON;
}
}
}
#else // HAS_RAWTOYROW_NEON
SIMD_ALIGNED(uint8 row[kMaxStride * 2]);
void (*RAWToARGBRow)(const uint8* src_rgb, uint8* dst_argb, int pix) =
RAWToARGBRow_C;
......@@ -1291,15 +1323,7 @@ int RAWToI420(const uint8* src_raw, int src_stride_raw,
RAWToARGBRow = RAWToARGBRow_SSSE3;
}
}
#elif defined(HAS_RAWTOARGBROW_NEON)
if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
RAWToARGBRow = RAWToARGBRow_Any_NEON;
if (IS_ALIGNED(width, 8)) {
RAWToARGBRow = RAWToARGBRow_NEON;
}
}
#endif
void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
#if defined(HAS_ARGBTOUVROW_SSSE3)
......@@ -1310,23 +1334,6 @@ int RAWToI420(const uint8* src_raw, int src_stride_raw,
}
}
#endif
#if defined(HAS_RAWTOYROW_NEON)
void (*RAWToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
RAWToYRow_C;
if (TestCpuFlag(kCpuHasNEON) && width >= 8) {
RAWToYRow = RAWToYRow_Any_NEON;
if (IS_ALIGNED(width, 8)) {
RAWToYRow = RAWToYRow_NEON;
}
if (width >= 16) {
ARGBToUVRow = ARGBToUVRow_Any_NEON;
if (IS_ALIGNED(width, 16)) {
ARGBToUVRow = ARGBToUVRow_NEON;
}
}
}
#else
void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
ARGBToYRow_C;
#if defined(HAS_ARGBTOUVROW_SSSE3)
......@@ -1343,13 +1350,14 @@ int RAWToI420(const uint8* src_raw, int src_stride_raw,
#endif // HAS_RAWTOYROW_NEON
for (int y = 0; y < height - 1; y += 2) {
RAWToARGBRow(src_raw, row, width);
RAWToARGBRow(src_raw + src_stride_raw, row + kMaxStride, width);
ARGBToUVRow(row, kMaxStride, dst_u, dst_v, width);
#if defined(HAS_RAWTOYROW_NEON)
RAWToUVRow(src_raw, src_stride_raw, dst_u, dst_v, width);
RAWToYRow(src_raw, dst_y, width);
RAWToYRow(src_raw + src_stride_raw, dst_y + dst_stride_y, width);
#else
RAWToARGBRow(src_raw, row, width);
RAWToARGBRow(src_raw + src_stride_raw, row + kMaxStride, width);
ARGBToUVRow(row, kMaxStride, dst_u, dst_v, width);
ARGBToYRow(row, dst_y, width);
ARGBToYRow(row + kMaxStride, dst_y + dst_stride_y, width);
#endif
......@@ -1359,11 +1367,12 @@ int RAWToI420(const uint8* src_raw, int src_stride_raw,
dst_v += dst_stride_v;
}
if (height & 1) {
RAWToARGBRow_C(src_raw, row, width);
ARGBToUVRow(row, 0, dst_u, dst_v, width);
#if defined(HAS_RAWTOYROW_NEON)
RAWToUVRow(src_raw, 0, dst_u, dst_v, width);
RAWToYRow(src_raw, dst_y, width);
#else
RAWToARGBRow(src_raw, row, width);
ARGBToUVRow(row, 0, dst_u, dst_v, width);
ARGBToYRow(row, dst_y, width);
#endif
}
......@@ -1550,10 +1559,12 @@ int ARGB1555ToI420(const uint8* src_argb1555, int src_stride_argb1555,
#if defined(HAS_ARGB1555TOYROW_NEON)
ARGB1555ToUVRow(src_argb1555, src_stride_argb1555, dst_u, dst_v, width);
ARGB1555ToYRow(src_argb1555, dst_y, width);
ARGB1555ToYRow(src_argb1555 + src_stride_argb1555, dst_y + dst_stride_y, width);
ARGB1555ToYRow(src_argb1555 + src_stride_argb1555, dst_y + dst_stride_y,
width);
#else
ARGB1555ToARGBRow(src_argb1555, row, width);
ARGB1555ToARGBRow(src_argb1555 + src_stride_argb1555, row + kMaxStride, width);
ARGB1555ToARGBRow(src_argb1555 + src_stride_argb1555, row + kMaxStride,
width);
ARGBToUVRow(row, kMaxStride, dst_u, dst_v, width);
ARGBToYRow(row, dst_y, width);
ARGBToYRow(row + kMaxStride, dst_y + dst_stride_y, width);
......
......@@ -543,8 +543,8 @@ int ARGB1555ToARGB(const uint8* src_argb1555, int src_stride_argb1555,
src_argb1555 = src_argb1555 + (height - 1) * src_stride_argb1555;
src_stride_argb1555 = -src_stride_argb1555;
}
void (*ARGB1555ToARGBRow)(const uint8* src_argb1555, uint8* dst_argb, int pix) =
ARGB1555ToARGBRow_C;
void (*ARGB1555ToARGBRow)(const uint8* src_argb1555, uint8* dst_argb,
int pix) = ARGB1555ToARGBRow_C;
#if defined(HAS_ARGB1555TOARGBROW_SSE2)
if (TestCpuFlag(kCpuHasSSE2) && width >= 8 &&
IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
......@@ -585,8 +585,8 @@ int ARGB4444ToARGB(const uint8* src_argb4444, int src_stride_argb4444,
src_argb4444 = src_argb4444 + (height - 1) * src_stride_argb4444;
src_stride_argb4444 = -src_stride_argb4444;
}
void (*ARGB4444ToARGBRow)(const uint8* src_argb4444, uint8* dst_argb, int pix) =
ARGB4444ToARGBRow_C;
void (*ARGB4444ToARGBRow)(const uint8* src_argb4444, uint8* dst_argb,
int pix) = ARGB4444ToARGBRow_C;
#if defined(HAS_ARGB4444TOARGBROW_SSE2)
if (TestCpuFlag(kCpuHasSSE2) && width >= 8 &&
IS_ALIGNED(dst_argb, 16) && IS_ALIGNED(dst_stride_argb, 16)) {
......
......@@ -95,8 +95,7 @@ int ARGBToI422(const uint8* src_argb, int src_stride_argb,
ARGBToUV422Row = ARGBToUV422Row_Any_SSSE3;
if (IS_ALIGNED(width, 16)) {
ARGBToUV422Row = ARGBToUV422Row_Unaligned_SSSE3;
if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16) &&
IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
ARGBToUV422Row = ARGBToUV422Row_SSSE3;
}
}
......@@ -239,6 +238,9 @@ int ARGBToNV12(const uint8* src_argb, int src_stride_argb,
ARGBToYRow = ARGBToYRow_Any_NEON;
if (IS_ALIGNED(width, 8)) {
ARGBToYRow = ARGBToYRow_NEON;
}
if (width >= 16) {
ARGBToUVRow = ARGBToUVRow_Any_NEON;
if (IS_ALIGNED(width, 16)) {
ARGBToUVRow = ARGBToUVRow_NEON;
}
......@@ -345,6 +347,9 @@ int ARGBToNV21(const uint8* src_argb, int src_stride_argb,
ARGBToYRow = ARGBToYRow_Any_NEON;
if (IS_ALIGNED(width, 8)) {
ARGBToYRow = ARGBToYRow_NEON;
}
if (width >= 16) {
ARGBToUVRow = ARGBToUVRow_Any_NEON;
if (IS_ALIGNED(width, 16)) {
ARGBToUVRow = ARGBToUVRow_NEON;
}
......@@ -425,20 +430,27 @@ int ARGBToYUY2(const uint8* src_argb, int src_stride_argb,
dst_yuy2 = dst_yuy2 + (height - 1) * dst_stride_yuy2;
dst_stride_yuy2 = -dst_stride_yuy2;
}
void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
void (*ARGBToUV422Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
int pix) = ARGBToUV422Row_C;
#if defined(HAS_ARGBTOUV422ROW_SSSE3)
if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
ARGBToUV422Row = ARGBToUV422Row_Any_SSSE3;
if (IS_ALIGNED(width, 16)) {
ARGBToUV422Row = ARGBToUV422Row_Unaligned_SSSE3;
if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
ARGBToUV422Row = ARGBToUV422Row_SSSE3;
}
}
}
#endif
void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
ARGBToYRow_C;
#if defined(HAS_ARGBTOYROW_SSSE3)
if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
ARGBToYRow = ARGBToYRow_Any_SSSE3;
if (IS_ALIGNED(width, 16)) {
ARGBToUVRow = ARGBToUVRow_Unaligned_SSSE3;
ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
ARGBToUVRow = ARGBToUVRow_SSSE3;
ARGBToYRow = ARGBToYRow_SSSE3;
}
}
......@@ -448,8 +460,11 @@ int ARGBToYUY2(const uint8* src_argb, int src_stride_argb,
ARGBToYRow = ARGBToYRow_Any_NEON;
if (IS_ALIGNED(width, 8)) {
ARGBToYRow = ARGBToYRow_NEON;
}
if (width >= 16) {
ARGBToUV422Row = ARGBToUV422Row_Any_NEON;
if (IS_ALIGNED(width, 16)) {
ARGBToUVRow = ARGBToUVRow_NEON;
ARGBToUV422Row = ARGBToUV422Row_NEON;
}
}
}
......@@ -479,7 +494,7 @@ int ARGBToYUY2(const uint8* src_argb, int src_stride_argb,
SIMD_ALIGNED(uint8 row_v[kMaxStride / 2]);
for (int y = 0; y < height; ++y) {
ARGBToUVRow(src_argb, 0, row_u, row_v, width);
ARGBToUV422Row(src_argb, row_u, row_v, width);
ARGBToYRow(src_argb, row_y, width);
I422ToYUY2Row(row_y, row_u, row_v, dst_yuy2, width);
src_argb += src_stride_argb;
......@@ -504,20 +519,27 @@ int ARGBToUYVY(const uint8* src_argb, int src_stride_argb,
dst_uyvy = dst_uyvy + (height - 1) * dst_stride_uyvy;
dst_stride_uyvy = -dst_stride_uyvy;
}
void (*ARGBToUVRow)(const uint8* src_argb0, int src_stride_argb,
uint8* dst_u, uint8* dst_v, int width) = ARGBToUVRow_C;
void (*ARGBToUV422Row)(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
int pix) = ARGBToUV422Row_C;
#if defined(HAS_ARGBTOUV422ROW_SSSE3)
if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
ARGBToUV422Row = ARGBToUV422Row_Any_SSSE3;
if (IS_ALIGNED(width, 16)) {
ARGBToUV422Row = ARGBToUV422Row_Unaligned_SSSE3;
if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
ARGBToUV422Row = ARGBToUV422Row_SSSE3;
}
}
}
#endif
void (*ARGBToYRow)(const uint8* src_argb, uint8* dst_y, int pix) =
ARGBToYRow_C;
#if defined(HAS_ARGBTOYROW_SSSE3)
if (TestCpuFlag(kCpuHasSSSE3) && width >= 16) {
ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
ARGBToYRow = ARGBToYRow_Any_SSSE3;
if (IS_ALIGNED(width, 16)) {
ARGBToUVRow = ARGBToUVRow_Unaligned_SSSE3;
ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
if (IS_ALIGNED(src_argb, 16) && IS_ALIGNED(src_stride_argb, 16)) {
ARGBToUVRow = ARGBToUVRow_SSSE3;
ARGBToYRow = ARGBToYRow_SSSE3;
}
}
......@@ -527,8 +549,11 @@ int ARGBToUYVY(const uint8* src_argb, int src_stride_argb,
ARGBToYRow = ARGBToYRow_Any_NEON;
if (IS_ALIGNED(width, 8)) {
ARGBToYRow = ARGBToYRow_NEON;
}
if (width >= 16) {
ARGBToUV422Row = ARGBToUV422Row_Any_NEON;
if (IS_ALIGNED(width, 16)) {
ARGBToUVRow = ARGBToUVRow_NEON;
ARGBToUV422Row = ARGBToUV422Row_NEON;
}
}
}
......@@ -558,7 +583,7 @@ int ARGBToUYVY(const uint8* src_argb, int src_stride_argb,
SIMD_ALIGNED(uint8 row_v[kMaxStride / 2]);
for (int y = 0; y < height; ++y) {
ARGBToUVRow(src_argb, 0, row_u, row_v, width);
ARGBToUV422Row(src_argb, row_u, row_v, width);
ARGBToYRow(src_argb, row_y, width);
I422ToUYVYRow(row_y, row_u, row_v, dst_uyvy, width);
src_argb += src_stride_argb;
......
......@@ -315,8 +315,8 @@ int BayerToI420(const uint8* src_bayer, int src_stride_bayer,
ARGBToUVRow = ARGBToUVRow_Any_SSSE3;
ARGBToYRow = ARGBToYRow_Any_SSSE3;
if (IS_ALIGNED(width, 16)) {
ARGBToUVRow = ARGBToUVRow_SSSE3;
ARGBToYRow = ARGBToYRow_Unaligned_SSSE3;
ARGBToUVRow = ARGBToUVRow_SSSE3;
if (IS_ALIGNED(dst_y, 16) && IS_ALIGNED(dst_stride_y, 16)) {
ARGBToYRow = ARGBToYRow_SSSE3;
}
......@@ -327,6 +327,9 @@ int BayerToI420(const uint8* src_bayer, int src_stride_bayer,
ARGBToYRow = ARGBToYRow_Any_NEON;
if (IS_ALIGNED(width, 8)) {
ARGBToYRow = ARGBToYRow_NEON;
}
if (width >= 16) {
ARGBToUVRow = ARGBToUVRow_Any_NEON;
if (IS_ALIGNED(width, 16)) {
ARGBToUVRow = ARGBToUVRow_NEON;
}
......
......@@ -248,6 +248,11 @@ UVANY(UYVYToUVRow_Any_SSE2, UYVYToUVRow_Unaligned_SSE2, UYVYToUVRow_C, 2)
#endif
#ifdef HAS_ARGBTOUVROW_NEON
UVANY(ARGBToUVRow_Any_NEON, ARGBToUVRow_NEON, ARGBToUVRow_C, 4)
UVANY(BGRAToUVRow_Any_NEON, BGRAToUVRow_NEON, BGRAToUVRow_C, 4)
UVANY(ABGRToUVRow_Any_NEON, ABGRToUVRow_NEON, ABGRToUVRow_C, 4)
UVANY(RGBAToUVRow_Any_NEON, RGBAToUVRow_NEON, RGBAToUVRow_C, 4)
UVANY(RGB24ToUVRow_Any_NEON, RGB24ToUVRow_NEON, RGB24ToUVRow_C, 3)
UVANY(RAWToUVRow_Any_NEON, RAWToUVRow_NEON, RAWToUVRow_C, 3)
UVANY(RGB565ToUVRow_Any_NEON, RGB565ToUVRow_NEON, RGB565ToUVRow_C, 2)
UVANY(ARGB1555ToUVRow_Any_NEON, ARGB1555ToUVRow_NEON, ARGB1555ToUVRow_C, 2)
UVANY(ARGB4444ToUVRow_Any_NEON, ARGB4444ToUVRow_NEON, ARGB4444ToUVRow_C, 2)
......
......@@ -109,7 +109,8 @@ void RGB565ToARGBRow_C(const uint8* src_rgb565, uint8* dst_argb, int width) {
}
}
void ARGB1555ToARGBRow_C(const uint8* src_argb1555, uint8* dst_argb, int width) {
void ARGB1555ToARGBRow_C(const uint8* src_argb1555, uint8* dst_argb,
int width) {
for (int x = 0; x < width; ++x) {
uint8 b = src_argb1555[0] & 0x1f;
uint8 g = (src_argb1555[0] >> 5) | ((src_argb1555[1] & 0x03) << 3);
......@@ -124,7 +125,8 @@ void ARGB1555ToARGBRow_C(const uint8* src_argb1555, uint8* dst_argb, int width)
}
}
void ARGB4444ToARGBRow_C(const uint8* src_argb4444, uint8* dst_argb, int width) {
void ARGB4444ToARGBRow_C(const uint8* src_argb4444, uint8* dst_argb,
int width) {
for (int x = 0; x < width; ++x) {
uint8 b = src_argb4444[0] & 0x0f;
uint8 g = src_argb4444[0] >> 4;
......
This diff is collapsed.
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment