Skip to content

L
libyuv
Commit fd4b315d authored by fbarchard@google.com's avatar fbarchard@google.com
Browse files
Options

64 bit port of rotation and tweeks to planar 

BUG=none
TEST=local tests
Review URL: http://webrtc-codereview.appspot.com/244008

git-svn-id: http://libyuv.googlecode.com/svn/trunk@38 16f28f9a-4ce2-e073-06de-1de4eb20be90
parent 36ab38a3
Hide whitespace changes
Inline Side-by-side
Showing with 501 additions and 46 deletions
source/planar_functions.cc
View file @ fd4b315d
......@@ -64,6 +64,14 @@ extern "C" TALIGN16(const uint8, kShuffleMaskBG24ToARGB[16]) =
extern "C" TALIGN16(const uint8, kShuffleMaskRAWToARGB[16]) =
{ 2u, 1u, 0u, 12u, 5u, 4u, 3u, 13u, 8u, 7u, 6u, 14u, 11u, 10u, 9u, 15u };
// Constant multiplication table for converting ARGB to I400.
extern "C" TALIGN16(const uint8, kMultiplyMaskARGBToI400[16]) =
{ 13u, 64u, 33u, 0u, 13u, 64u, 33u, 0u,
13u, 64u, 33u, 0u, 13u, 64u, 33u, 0u };
extern "C" TALIGN16(const uint8, kMultiplyMaskARGBToI400_2[16]) =
{ 1u, 1u, 1u, 1u, 1u, 1u, 1u, 1u, 1u, 1u, 1u, 1u, 1u, 1u, 1u, 1u };
#if defined(WIN32) && !defined(COVERAGE_ENABLED)
#define HAS_SPLITUV_SSE2
__declspec(naked)
......@@ -1112,10 +1120,10 @@ __asm {
mov eax, [esp + 4] // src_abgr
mov edx, [esp + 8] // dst_argb
mov ecx, [esp + 12] // pix
movdqa xmm7, byte ptr [_kShuffleMaskABGRToARGB]
movdqa xmm7, _kShuffleMaskABGRToARGB
convertloop :
movdqa xmm0, qword ptr [eax]
movdqa xmm0, [eax]
lea eax, [eax + 16]
pshufb xmm0, xmm7
movdqa [edx], xmm0
......@@ -1134,10 +1142,10 @@ __asm {
mov eax, [esp + 4] // src_bgra
mov edx, [esp + 8] // dst_argb
mov ecx, [esp + 12] // pix
movdqa xmm7, byte ptr [_kShuffleMaskBGRAToARGB]
movdqa xmm7, _kShuffleMaskBGRAToARGB
convertloop :
movdqa xmm0, qword ptr [eax]
movdqa xmm0, [eax]
lea eax, [eax + 16]
pshufb xmm0, xmm7
movdqa [edx], xmm0
......@@ -1158,12 +1166,12 @@ __asm {
mov ecx, [esp + 12] // pix
pcmpeqb xmm7, xmm7 // generate mask 0xff000000
pslld xmm7, 24
movdqa xmm6, byte ptr [_kShuffleMaskBG24ToARGB]
movdqa xmm6, _kShuffleMaskBG24ToARGB
convertloop :
movdqa xmm0, qword ptr [eax]
movdqa xmm1, qword ptr [eax + 16]
movdqa xmm3, qword ptr [eax + 32]
movdqa xmm0, [eax]
movdqa xmm1, [eax + 16]
movdqa xmm3, [eax + 32]
lea eax, [eax + 48]
movdqa xmm2, xmm3
palignr xmm2, xmm1, 8 // xmm2 = { xmm3[0:3] xmm1[8:15]}
......@@ -1198,12 +1206,12 @@ __asm {
mov ecx, [esp + 12] // pix
pcmpeqb xmm7, xmm7 // generate mask 0xff000000
pslld xmm7, 24
movdqa xmm6, byte ptr [_kShuffleMaskRAWToARGB]
movdqa xmm6, _kShuffleMaskRAWToARGB
convertloop :
movdqa xmm0, qword ptr [eax]
movdqa xmm1, qword ptr [eax + 16]
movdqa xmm3, qword ptr [eax + 32]
movdqa xmm0, [eax]
movdqa xmm1, [eax + 16]
movdqa xmm3, [eax + 32]
lea eax, [eax + 48]
movdqa xmm2, xmm3
palignr xmm2, xmm1, 8 // xmm2 = { xmm3[0:3] xmm1[8:15]}
......@@ -1228,6 +1236,39 @@ __asm {
}
}
#define HAS_ARGBTOI400ROW_SSSE3
__declspec(naked)
static void ARGBToI400Row_SSSE3(const uint8* src_argb, uint8* dst_y,
int pix) {
__asm {
mov eax, [esp + 4] // src_argb
mov edx, [esp + 8] // dst_y
mov ecx, [esp + 12] // pix
movdqa xmm7, _kMultiplyMaskARGBToI400
movdqa xmm6, _kMultiplyMaskARGBToI400_2
movdqa xmm5, xmm6
psllw xmm5, 4 // Generate a mask of 0x10 on each byte.
convertloop :
movdqa xmm0, [eax]
pmaddubsw xmm0, xmm7
movdqa xmm1, [eax + 16]
psrlw xmm0, 7
pmaddubsw xmm1, xmm7
lea eax, [eax + 32]
psrlw xmm1, 7
packuswb xmm0, xmm1
pmaddubsw xmm0, xmm6
packuswb xmm0, xmm0
paddb xmm0, xmm5
movq qword ptr [edx], xmm0
lea edx, [edx + 8]
sub ecx, 8
ja convertloop
ret
}
}
#elif (defined(__x86_64__) || defined(__i386__)) && \
!defined(COVERAGE_ENABLED) && !defined(TARGET_IPHONE_SIMULATOR)
......@@ -1378,6 +1419,40 @@ static void RAWToARGBRow_SSSE3(const uint8* src_raw, uint8* dst_argb,
: "memory"
);
}
#define HAS_ARGBTOI400ROW_SSSE3
static void ARGBToI400Row_SSSE3(const uint8* src_argb, uint8* dst_y,
int pix) {
asm volatile(
"movdqa (%3),%%xmm7\n"
"movdqa (%4),%%xmm6\n"
"movdqa %%xmm6,%%xmm5\n"
"psllw $0x4,%%xmm5\n" // Generate a mask of 0x10 on each byte.
"1:"
"movdqa (%0),%%xmm0\n"
"pmaddubsw %%xmm7,%%xmm0\n"
"movdqa 0x10(%0),%%xmm1\n"
"psrlw $0x7,%%xmm0\n"
"pmaddubsw %%xmm7,%%xmm1\n"
"lea 0x20(%0),%0\n"
"psrlw $0x7,%%xmm1\n"
"packuswb %%xmm1,%%xmm0\n"
"pmaddubsw %%xmm6,%%xmm0\n"
"packuswb %%xmm0,%%xmm0\n"
"paddb %%xmm5,%%xmm0\n"
"movq %%xmm0,(%1)\n"
"lea 0x8(%1),%1\n"
"sub $0x8,%2\n"
"ja 1b\n"
: "+r"(src_argb), // %0
"+r"(dst_y), // %1
"+r"(pix) // %2
: "r"(kMultiplyMaskARGBToI400), // %3
"r"(kMultiplyMaskARGBToI400_2) // %4
: "memory"
);
}
#endif
static void I400ToARGBRow_C(const uint8* src_y, uint8* dst_argb, int pix) {
......@@ -1605,10 +1680,10 @@ int BGRAToARGB(const uint8* src_bgra, int src_stride_bgra,
static void ARGBToI400Row_C(const uint8* src_argb, uint8* dst_y, int pix) {
for (int x = 0; x < pix; ++x) {
uint32 b = static_cast<uint32>(src_argb[0] * 25u);
uint32 g = static_cast<uint32>(src_argb[1] * 129u);
uint32 r = static_cast<uint32>(src_argb[2] * 66u);
*(dst_y++) = static_cast<uint8>(((b + g + r) >> 8) + 16u);
uint32 b = static_cast<uint32>(src_argb[0] * 13u);
uint32 g = static_cast<uint32>(src_argb[1] * 64u);
uint32 r = static_cast<uint32>(src_argb[2] * 33u);
*(dst_y++) = static_cast<uint8>(((b + g + r) >> 7) + 16u);
src_argb += 4;
}
}
......@@ -1622,8 +1697,21 @@ int ARGBToI400(const uint8* src_argb, int src_stride_argb,
src_argb = src_argb + (height - 1) * src_stride_argb;
src_stride_argb = -src_stride_argb;
}
void (*ARGBToI400Row)(const uint8* src_argb, uint8* dst_y, int pix);
#if defined(HAS_ARGBTOI400ROW_SSSE3)
if (libyuv::TestCpuFlag(libyuv::kCpuHasSSSE3) &&
(width % 4 == 0) &&
IS_ALIGNED(src_argb, 16) && (src_stride_argb % 16 == 0) &&
IS_ALIGNED(dst_y, 16) && (dst_stride_y % 16 == 0)) {
ARGBToI400Row = ARGBToI400Row_SSSE3;
} else
#endif
{
ARGBToI400Row = ARGBToI400Row_C;
}
for (int y = 0; y < height; ++y) {
ARGBToI400Row_C(src_argb, dst_y, width);
ARGBToI400Row(src_argb, dst_y, width);
src_argb += src_stride_argb;
dst_y += dst_stride_y;
}
......
source/rotate.cc
View file @ fd4b315d
......@@ -12,8 +12,25 @@
#include "libyuv/rotate.h"
#include "rotate_priv.h"
#include "libyuv/cpu_id.h"
namespace libyuv {
#if (defined(WIN32) || defined(__x86_64__) || defined(__i386__)) \
&& !defined(COVERAGE_ENABLED) && !defined(TARGET_IPHONE_SIMULATOR)
#if defined(_MSC_VER)
#define TALIGN16(t, var) static __declspec(align(16)) t _ ## var
#else
#define TALIGN16(t, var) t var __attribute__((aligned(16)))
#endif
// Shuffle table for reversing the bytes.
extern "C" TALIGN16(const uint8, kShuffleReverse[16]) =
{ 15u, 14u, 13u, 12u, 11u, 10u, 9u, 8u, 7u, 6u, 5u, 4u, 3u, 2u, 1u, 0u };
// Shuffle table for reversing the bytes of UV channels.
extern "C" TALIGN16(const uint8, kShuffleReverseUV[16]) =
{ 14u, 12u, 10u, 8u, 6u, 4u, 2u, 0u, 15u, 13u, 11u, 9u, 7u, 5u, 3u, 1u };
#endif
typedef void (*reverse_uv_func)(const uint8*, uint8*, uint8*, int);
typedef void (*reverse_func)(const uint8*, uint8*, int);
typedef void (*rotate_uv_wx8_func)(const uint8*, int,
......@@ -29,12 +46,16 @@ typedef void (*rotate_wxh_func)(const uint8*, int, uint8*, int, int, int);
extern "C" {
void RestoreRegisters_NEON(unsigned long long *restore);
void SaveRegisters_NEON(unsigned long long *store);
#define HAS_REVERSE_LINE_NEON
void ReverseLine_NEON(const uint8* src, uint8* dst, int width);
#define HAS_REVERSE_LINE_UV_NEON
void ReverseLineUV_NEON(const uint8* src,
uint8* dst_a, uint8* dst_b,
int width);
#define HAS_TRANSPOSE_WX8_NEON
void TransposeWx8_NEON(const uint8* src, int src_stride,
uint8* dst, int dst_stride, int width);
#define HAS_TRANSPOSE_UVWX8_NEON
void TransposeUVWx8_NEON(const uint8* src, int src_stride,
uint8* dst_a, int dst_stride_a,
uint8* dst_b, int dst_stride_b,
......@@ -42,6 +63,185 @@ void TransposeUVWx8_NEON(const uint8* src, int src_stride,
} // extern "C"
#endif
#if defined(WIN32) && !defined(COVERAGE_ENABLED)
#define HAS_TRANSPOSE_WX8_SSSE3
__declspec(naked)
static void TransposeWx8_SSSE3(const uint8* src, int src_stride,
uint8* dst, int dst_stride, int width) {
__asm {
push edi
push esi
push ebp
mov eax, [esp + 12 + 4] // src
mov edi, [esp + 12 + 8] // src_stride
mov edx, [esp + 12 + 12] // dst
mov esi, [esp + 12 + 16] // dst_stride
mov ecx, [esp + 12 + 20] // width
convertloop :
// Read in the data from the source pointer.
// First round of bit swap.
movq xmm0, qword ptr [eax]
lea ebp, [eax + 8]
movq xmm1, qword ptr [eax + edi]
lea eax, [eax + 2 * edi]
punpcklbw xmm0, xmm1
movq xmm2, qword ptr [eax]
movdqa xmm1, xmm0
palignr xmm1, xmm1, 8
movq xmm3, qword ptr [eax + edi]
lea eax, [eax + 2 * edi]
punpcklbw xmm2, xmm3
movdqa xmm3, xmm2
movq xmm4, qword ptr [eax]
palignr xmm3, xmm3, 8
movq xmm5, qword ptr [eax + edi]
punpcklbw xmm4, xmm5
lea eax, [eax + 2 * edi]
movdqa xmm5, xmm4
movq xmm6, qword ptr [eax]
palignr xmm5, xmm5, 8
movq xmm7, qword ptr [eax + edi]
punpcklbw xmm6, xmm7
mov eax, ebp
movdqa xmm7, xmm6
palignr xmm7, xmm7, 8
// Second round of bit swap.
punpcklwd xmm0, xmm2
punpcklwd xmm1, xmm3
movdqa xmm2, xmm0
movdqa xmm3, xmm1
palignr xmm2, xmm2, 8
palignr xmm3, xmm3, 8
punpcklwd xmm4, xmm6
punpcklwd xmm5, xmm7
movdqa xmm6, xmm4
movdqa xmm7, xmm5
palignr xmm6, xmm6, 8
palignr xmm7, xmm7, 8
// Third round of bit swap.
// Write to the destination pointer.
punpckldq xmm0, xmm4
movq qword ptr [edx], xmm0
movdqa xmm4, xmm0
palignr xmm4, xmm4, 8
movq qword ptr [edx + esi], xmm4
lea edx, [edx + 2 * esi]
punpckldq xmm2, xmm6
movdqa xmm6, xmm2
palignr xmm6, xmm6, 8
movq qword ptr [edx], xmm2
punpckldq xmm1, xmm5
movq qword ptr [edx + esi], xmm6
lea edx, [edx + 2 * esi]
movdqa xmm5, xmm1
movq qword ptr [edx], xmm1
palignr xmm5, xmm5, 8
punpckldq xmm3, xmm7
movq qword ptr [edx + esi], xmm5
lea edx, [edx + 2 * esi]
movq qword ptr [edx], xmm3
movdqa xmm7, xmm3
palignr xmm7, xmm7, 8
movq qword ptr [edx + esi], xmm7
lea edx, [edx + 2 * esi]
sub ecx, 8
ja convertloop
pop ebp
pop esi
pop edi
ret
}
}
#elif (defined(__i386__) || defined(__x86_64__)) && \
!defined(COVERAGE_ENABLED) && !defined(TARGET_IPHONE_SIMULATOR)
#define HAS_TRANSPOSE_WX8_SSSE3
static void TransposeWx8_SSSE3(const uint8* src, int src_stride,
uint8* dst, int dst_stride, int width) {
asm volatile(
"1:"
// Read in the data from the source pointer.
// First round of bit swap.
"movq (%0),%%xmm0\n"
"movq (%0,%3),%%xmm1\n"
"lea (%0,%3,2),%0\n"
"punpcklbw %%xmm1,%%xmm0\n"
"movq (%0),%%xmm2\n"
"movdqa %%xmm0,%%xmm1\n"
"palignr $0x8,%%xmm1,%%xmm1\n"
"movq (%0,%3),%%xmm3\n"
"lea (%0,%3,2),%0\n"
"punpcklbw %%xmm3,%%xmm2\n"
"movdqa %%xmm2,%%xmm3\n"
"movq (%0),%%xmm4\n"
"palignr $0x8,%%xmm3,%%xmm3\n"
"movq (%0,%3),%%xmm5\n"
"lea (%0,%3,2),%0\n"
"punpcklbw %%xmm5,%%xmm4\n"
"movdqa %%xmm4,%%xmm5\n"
"movq (%0),%%xmm6\n"
"palignr $0x8,%%xmm5,%%xmm5\n"
"movq (%0,%3),%%xmm7\n"
"lea (%0,%3,2),%0\n"
"punpcklbw %%xmm7,%%xmm6\n"
"neg %3\n"
"movdqa %%xmm6,%%xmm7\n"
"lea 0x8(%0,%3,8),%0\n"
"palignr $0x8,%%xmm7,%%xmm7\n"
"neg %3\n"
// Second round of bit swap.
"punpcklwd %%xmm2,%%xmm0\n"
"punpcklwd %%xmm3,%%xmm1\n"
"movdqa %%xmm0,%%xmm2\n"
"movdqa %%xmm1,%%xmm3\n"
"palignr $0x8,%%xmm2,%%xmm2\n"
"palignr $0x8,%%xmm3,%%xmm3\n"
"punpcklwd %%xmm6,%%xmm4\n"
"punpcklwd %%xmm7,%%xmm5\n"
"movdqa %%xmm4,%%xmm6\n"
"movdqa %%xmm5,%%xmm7\n"
"palignr $0x8,%%xmm6,%%xmm6\n"
"palignr $0x8,%%xmm7,%%xmm7\n"
// Third round of bit swap.
// Write to the destination pointer.
"punpckldq %%xmm4,%%xmm0\n"
"movq %%xmm0,(%1)\n"
"movdqa %%xmm0,%%xmm4\n"
"palignr $0x8,%%xmm4,%%xmm4\n"
"movq %%xmm4,(%1,%4)\n"
"lea (%1,%4,2),%1\n"
"punpckldq %%xmm6,%%xmm2\n"
"movdqa %%xmm2,%%xmm6\n"
"movq %%xmm2,(%1)\n"
"palignr $0x8,%%xmm6,%%xmm6\n"
"punpckldq %%xmm5,%%xmm1\n"
"movq %%xmm6,(%1,%4)\n"
"lea (%1,%4,2),%1\n"
"movdqa %%xmm1,%%xmm5\n"
"movq %%xmm1,(%1)\n"
"palignr $0x8,%%xmm5,%%xmm5\n"
"movq %%xmm5,(%1,%4)\n"
"lea (%1,%4,2),%1\n"
"punpckldq %%xmm7,%%xmm3\n"
"movq %%xmm3,(%1)\n"
"movdqa %%xmm3,%%xmm7\n"
"palignr $0x8,%%xmm7,%%xmm7\n"
"movq %%xmm7,(%1,%4)\n"
"lea (%1,%4,2),%1\n"
"sub $0x8,%2\n"
"ja 1b\n"
: "+r"(src), // %0
"+r"(dst), // %1
"+r"(width) // %2
: "r"(static_cast<intptr_t>(src_stride)), // %3
"r"(static_cast<intptr_t>(dst_stride)) // %4
: "memory"
);
}
#endif
static void TransposeWx8_C(const uint8* src, int src_stride,
uint8* dst, int dst_stride,
int w) {
......@@ -67,14 +267,28 @@ void TransposePlane(const uint8* src, int src_stride,
rotate_wx8_func TransposeWx8;
rotate_wxh_func TransposeWxH;
// do processor detection here.
#ifdef __ARM_NEON__
TransposeWx8 = TransposeWx8_NEON;
TransposeWxH = TransposeWxH_C;
#else
TransposeWx8 = TransposeWx8_C;
TransposeWxH = TransposeWxH_C;
#if defined(HAS_TRANSPOSE_WX8_NEON)
if (libyuv::TestCpuFlag(libyuv::kCpuHasNEON) &&
(width % 8 == 0) &&
IS_ALIGNED(src, 16) && (src_stride % 8 == 0) &&
IS_ALIGNED(dst, 16) && (dst_stride % 8 == 0)) {
TransposeWx8 = TransposeWx8_NEON;
TransposeWxH = TransposeWxH_C;
} else
#endif
#if defined(HAS_TRANSPOSE_WX8_SSSE3)
if (libyuv::TestCpuFlag(libyuv::kCpuHasSSSE3) &&
(width % 8 == 0) &&
IS_ALIGNED(src, 16) && (src_stride % 8 == 0) &&
IS_ALIGNED(dst, 16) && (dst_stride % 8 == 0)) {
TransposeWx8 = TransposeWx8_SSSE3;
TransposeWxH = TransposeWxH_C;
} else
#endif
{
TransposeWx8 = TransposeWx8_C;
TransposeWxH = TransposeWxH_C;
}
// work across the source in 8x8 tiles
while (i >= 8) {
......@@ -112,7 +326,7 @@ void RotatePlane270(const uint8* src, int src_stride,
TransposePlane(src, src_stride, dst, dst_stride, width, height);
}
void ReverseLine_C(const uint8* src, uint8* dst, int width) {
static void ReverseLine_C(const uint8* src, uint8* dst, int width) {
int i;
src += width;
for (i = 0; i < width; ++i) {
......@@ -121,19 +335,78 @@ void ReverseLine_C(const uint8* src, uint8* dst, int width) {
}
}
#if defined(WIN32) && !defined(COVERAGE_ENABLED)
#define HAS_REVERSE_LINE_SSSE3
__declspec(naked)
static void ReverseLine_SSSE3(const uint8* src, uint8* dst, int width) {
__asm {
mov eax, [esp + 4] // src
mov edx, [esp + 8] // dst
mov ecx, [esp + 12] // width
movdqa xmm7, _kShuffleReverse
lea eax, [eax + ecx - 16]
convertloop :
movdqa xmm0, [eax]
lea eax, [eax - 16]
pshufb xmm0, xmm7
movdqa [edx], xmm0
lea edx, [edx + 16]
sub ecx, 16
ja convertloop
ret
}
}
#elif (defined(__i386__) || defined(__x86_64__)) && \
!defined(COVERAGE_ENABLED) && !defined(TARGET_IPHONE_SIMULATOR)
#define HAS_REVERSE_LINE_SSSE3
static void ReverseLine_SSSE3(const uint8* src, uint8* dst, int width) {
intptr_t temp_width = static_cast<intptr_t>(width);
asm volatile(
"movdqa (%3),%%xmm7\n"
"lea -0x10(%0,%2,1),%0\n"
"1:"
"movdqa (%0),%%xmm0\n"
"lea -0x10(%0),%0\n"
"pshufb %%xmm7,%%xmm0\n"
"movdqa %%xmm0,(%1)\n"
"lea 0x10(%1),%1\n"
"sub $0x10,%2\n"
"ja 1b\n"
: "+r"(src), // %0
"+r"(dst), // %1
"+r"(temp_width) // %2
: "r"(kShuffleReverse) // %3
: "memory"
);
}
#endif
void RotatePlane180(const uint8* src, int src_stride,
uint8* dst, int dst_stride,
int width, int height) {
int i;
reverse_func ReverseLine;
// TODO(frkoenig): do processor detection here.
#ifdef __ARM_NEON__
ReverseLine = ReverseLine_NEON;
#else
ReverseLine = ReverseLine_C;
#if defined(HAS_REVERSE_LINE_NEON)
if (libyuv::TestCpuFlag(libyuv::kCpuHasNEON) &&
(width % 16 == 0) &&
IS_ALIGNED(src, 16) && (src_stride % 16 == 0) &&
IS_ALIGNED(dst, 16) && (dst_stride % 16 == 0)) {
ReverseLine = ReverseLine_NEON;
} else
#endif
#if defined(HAS_REVERSE_LINE_SSSE3)
if (libyuv::TestCpuFlag(libyuv::kCpuHasSSSE3) &&
(width % 16 == 0) &&
IS_ALIGNED(src, 16) && (src_stride % 16 == 0) &&
IS_ALIGNED(dst, 16) && (dst_stride % 16 == 0)) {
ReverseLine = ReverseLine_SSSE3;
} else
#endif
{
ReverseLine = ReverseLine_C;
}
// Rotate by 180 is a mirror with the destination
// written in reverse.
dst += dst_stride * (height - 1);
......@@ -178,16 +451,32 @@ void TransposeUV(const uint8* src, int src_stride,
rotate_uv_wx8_func TransposeWx8;
rotate_uv_wxh_func TransposeWxH;
// do processor detection here.
#ifdef __ARM_NEON__
unsigned long long store_reg[8];
SaveRegisters_NEON(store_reg);
TransposeWx8 = TransposeUVWx8_NEON;
TransposeWxH = TransposeUVWxH_C;
#else
TransposeWx8 = TransposeUVWx8_C;
TransposeWxH = TransposeUVWxH_C;
#if defined(HAS_TRANSPOSE_UVWX8_NEON)
if (libyuv::TestCpuFlag(libyuv::kCpuHasNEON) &&
(width % 8 == 0) &&
IS_ALIGNED(src, 16) && (src_stride % 8 == 0) &&
IS_ALIGNED(dst_a, 16) && (dst_stride_a % 8 == 0) &&
IS_ALIGNED(dst_b, 16) && (dst_stride_b % 8 == 0)) {
unsigned long long store_reg[8];
SaveRegisters_NEON(store_reg);
TransposeWx8 = TransposeUVWx8_NEON;
TransposeWxH = TransposeUVWxH_C;
} else
#endif
#if defined(HAS_TRANSPOSE_UVWX8_SSE2)
if (libyuv::TestCpuFlag(libyuv::kCpuHasSSE2) &&
(width % 8 == 0) &&
IS_ALIGNED(src, 16) && (src_stride % 8 == 0) &&
IS_ALIGNED(dst_a, 16) && (dst_stride_a % 8 == 0) &&
IS_ALIGNED(dst_b, 16) && (dst_stride_b % 8 == 0)) {
TransposeWx8 = TransposeUVWx8_SSE2;
TransposeWxH = TransposeUVWxH_C;
} else
#endif
{
TransposeWx8 = TransposeUVWx8_C;
TransposeWxH = TransposeUVWxH_C;
}
// work through the source in 8x8 tiles
while (i >= 8) {
......@@ -207,8 +496,10 @@ void TransposeUV(const uint8* src, int src_stride,
dst_b, dst_stride_b,
width, i);
#ifdef __ARM_NEON__
RestoreRegisters_NEON(store_reg);
#if defined(HAS_TRANSPOSE_UVWX8_NEON)
if (libyuv::TestCpuFlag(libyuv::kCpuHasNEON)) {
RestoreRegisters_NEON(store_reg);
}
#endif
}
......@@ -240,6 +531,66 @@ void RotateUV270(const uint8* src, int src_stride,
width, height);
}
#if defined(WIN32) && !defined(COVERAGE_ENABLED)
#define HAS_REVERSE_LINE_UV_SSSE3
__declspec(naked)
void ReverseLineUV_SSSE3(const uint8* src,
uint8* dst_a, uint8* dst_b,
int width) {
__asm {
push edi
mov eax, [esp + 4 + 4] // src
mov edx, [esp + 4 + 8] // dst_a
mov edi, [esp + 4 + 12] // dst_b
mov ecx, [esp + 4 + 16] // width
movdqa xmm7, _kShuffleReverseUV
lea eax, [eax + 2 * ecx - 16]
convertloop :
movdqa xmm0, [eax]
lea eax, [eax - 16]
pshufb xmm0, xmm7
movlpd qword ptr [edx], xmm0
lea edx, [edx + 8]
movhpd qword ptr [edi], xmm0
lea edi, [edi + 8]
sub ecx, 8
ja convertloop
pop edi
ret
}
}
#elif (defined(__i386__) || defined(__x86_64__)) && \
!defined(COVERAGE_ENABLED) && !defined(TARGET_IPHONE_SIMULATOR)
#define HAS_REVERSE_LINE_UV_SSSE3
void ReverseLineUV_SSSE3(const uint8* src,
uint8* dst_a, uint8* dst_b,
int width) {
intptr_t temp_width = static_cast<intptr_t>(width);
asm volatile(
"movdqa (%4),%%xmm7\n"
"lea -0x10(%0,%3,2),%0\n"
"1:"
"movdqa (%0),%%xmm0\n"
"lea -0x10(%0),%0\n"
"pshufb %%xmm7,%%xmm0\n"
"movlpd %%xmm0,(%1)\n"
"lea 0x8(%1),%1\n"
"movhpd %%xmm0,(%2)\n"
"lea 0x8(%2),%2\n"
"sub $0x8,%3\n"
"ja 1b\n"
: "+r"(src), // %0
"+r"(dst_a), // %1
"+r"(dst_b), // %2
"+r"(temp_width) // %3
: "r"(kShuffleReverseUV) // %4
: "memory"
);
}
#endif
static void ReverseLineUV_C(const uint8* src,
uint8* dst_a, uint8* dst_b,
int width) {
......@@ -260,11 +611,27 @@ void RotateUV180(const uint8* src, int src_stride,
reverse_uv_func ReverseLine;
// TODO(frkoenig) : do processor detection here.
#ifdef __ARM_NEON__
ReverseLine = ReverseLineUV_NEON;
#else
ReverseLine = ReverseLineUV_C;
#if defined(HAS_REVERSE_LINE_UV_NEON)
if (libyuv::TestCpuFlag(libyuv::kCpuHasNEON) &&
(width % 16 == 0) &&
IS_ALIGNED(src, 16) && (src_stride % 16 == 0) &&
IS_ALIGNED(dst_a, 16) && (dst_stride_a % 8 == 0) &&
IS_ALIGNED(dst_b, 16) && (dst_stride_b % 8 == 0) ) {
ReverseLine = ReverseLineUV_NEON;
} else
#endif
#if defined(HAS_REVERSE_LINE_UV_SSSE3)
if (libyuv::TestCpuFlag(libyuv::kCpuHasSSSE3) &&
(width % 16 == 0) &&
IS_ALIGNED(src, 16) && (src_stride % 16 == 0) &&
IS_ALIGNED(dst_a, 16) && (dst_stride_a % 8 == 0) &&
IS_ALIGNED(dst_b, 16) && (dst_stride_b % 8 == 0) ) {
ReverseLine = ReverseLineUV_SSSE3;
} else
#endif
{
ReverseLine = ReverseLineUV_C;
}
dst_a += dst_stride_a * (height - 1);
dst_b += dst_stride_b * (height - 1);
......
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