Skip to content

L
libyuv
Commit 3f4d8605 authored by Frank Barchard's avatar Frank Barchard
Browse files
Options

avx2 interpolate use 8 bit 

BUG=libyuv:535
R=dhrosa@google.com

Review URL: https://codereview.chromium.org/1535833003 .
parent f4447745
Show whitespace changes
Inline Side-by-side
Showing with 86 additions and 92 deletions
README.chromium
View file @ 3f4d8605
Name: libyuv
URL: http://code.google.com/p/libyuv/
Version: 1557
Version: 1558
License: BSD
License File: LICENSE
......
include/libyuv/version.h
View file @ 3f4d8605
......@@ -11,6 +11,6 @@
#ifndef INCLUDE_LIBYUV_VERSION_H_ // NOLINT
#define INCLUDE_LIBYUV_VERSION_H_
#define LIBYUV_VERSION 1557
#define LIBYUV_VERSION 1558
#endif // INCLUDE_LIBYUV_VERSION_H_ NOLINT
source/row_common.cc
View file @ 3f4d8605
......@@ -2211,30 +2211,30 @@ static void HalfRow_16_C(const uint16* src_uv, int src_uv_stride,
void InterpolateRow_C(uint8* dst_ptr, const uint8* src_ptr,
ptrdiff_t src_stride,
int width, int source_y_fraction) {
int y1_fraction = source_y_fraction >> 1;
int y0_fraction = 128 - y1_fraction;
int y1_fraction = source_y_fraction ;
int y0_fraction = 256 - y1_fraction;
const uint8* src_ptr1 = src_ptr + src_stride;
int x;
if (y1_fraction == 0) {
memcpy(dst_ptr, src_ptr, width);
return;
}
if (y1_fraction == 64) {
if (y1_fraction == 128) {
HalfRow_C(src_ptr, (int)(src_stride), dst_ptr, width);
return;
}
for (x = 0; x < width - 1; x += 2) {
dst_ptr[0] =
(src_ptr[0] * y0_fraction + src_ptr1[0] * y1_fraction + 64) >> 7;
(src_ptr[0] * y0_fraction + src_ptr1[0] * y1_fraction + 128) >> 8;
dst_ptr[1] =
(src_ptr[1] * y0_fraction + src_ptr1[1] * y1_fraction + 64) >> 7;
(src_ptr[1] * y0_fraction + src_ptr1[1] * y1_fraction + 128) >> 8;
src_ptr += 2;
src_ptr1 += 2;
dst_ptr += 2;
}
if (width & 1) {
dst_ptr[0] =
(src_ptr[0] * y0_fraction + src_ptr1[0] * y1_fraction + 64) >> 7;
(src_ptr[0] * y0_fraction + src_ptr1[0] * y1_fraction + 128) >> 8;
}
}
......
source/row_gcc.cc
View file @ 3f4d8605
......@@ -4791,20 +4791,19 @@ void InterpolateRow_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
int source_y_fraction) {
asm volatile (
"sub %1,%0 \n"
"shr %3 \n"
"cmp $0x0,%3 \n"
"je 100f \n"
"cmp $0x40,%3 \n"
"cmp $0x80,%3 \n"
"je 50f \n"
"movd %3,%%xmm0 \n"
"neg %3 \n"
"add $0x80,%3 \n"
"add $0x100,%3 \n"
"movd %3,%%xmm5 \n"
"punpcklbw %%xmm0,%%xmm5 \n"
"punpcklwd %%xmm5,%%xmm5 \n"
"pshufd $0x0,%%xmm5,%%xmm5 \n"
"mov $0x400040,%%eax \n"
"mov $0x80808080,%%eax \n"
"movd %%eax,%%xmm4 \n"
"pshufd $0x0,%%xmm4,%%xmm4 \n"
......@@ -4816,14 +4815,18 @@ void InterpolateRow_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
"movdqa %%xmm0,%%xmm1 \n"
"punpcklbw %%xmm2,%%xmm0 \n"
"punpckhbw %%xmm2,%%xmm1 \n"
"pmaddubsw %%xmm5,%%xmm0 \n"
"pmaddubsw %%xmm5,%%xmm1 \n"
"paddw %%xmm4,%%xmm0 \n"
"paddw %%xmm4,%%xmm1 \n"
"psrlw $0x7,%%xmm0 \n"
"psrlw $0x7,%%xmm1 \n"
"packuswb %%xmm1,%%xmm0 \n"
MEMOPMEM(movdqu,xmm0,0x00,1,0,1)
"psubb %%xmm4,%%xmm0 \n"
"psubb %%xmm4,%%xmm1 \n"
"movdqa %%xmm5,%%xmm2 \n"
"movdqa %%xmm5,%%xmm3 \n"
"pmaddubsw %%xmm0,%%xmm2 \n"
"pmaddubsw %%xmm1,%%xmm3 \n"
"paddw %%xmm4,%%xmm2 \n"
"paddw %%xmm4,%%xmm3 \n"
"psrlw $0x8,%%xmm2 \n"
"psrlw $0x8,%%xmm3 \n"
"packuswb %%xmm3,%%xmm2 \n"
MEMOPMEM(movdqu,xmm2,0x00,1,0,1)
"lea " MEMLEA(0x10,1) ",%1 \n"
"sub $0x10,%2 \n"
"jg 1b \n"
......@@ -4857,7 +4860,7 @@ void InterpolateRow_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
"+r"(source_y_fraction) // %3
: "r"((intptr_t)(src_stride)) // %4
: "memory", "cc", "eax", NACL_R14
"xmm0", "xmm1", "xmm2", "xmm4", "xmm5"
"xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
);
}
#endif // HAS_INTERPOLATEROW_SSSE3
......@@ -4881,9 +4884,8 @@ void InterpolateRow_AVX2(uint8* dst_ptr, const uint8* src_ptr,
"vmovd %3,%%xmm5 \n"
"vpunpcklbw %%xmm0,%%xmm5,%%xmm5 \n"
"vpunpcklwd %%xmm5,%%xmm5,%%xmm5 \n"
"vpxor %%ymm0,%%ymm0,%%ymm0 \n"
"vpermd %%ymm5,%%ymm0,%%ymm5 \n"
"mov $0x400040,%%eax \n"
"vbroadcastss %%xmm5,%%ymm5 \n"
"mov $0x80808080,%%eax \n"
"vmovd %%eax,%%xmm4 \n"
"vbroadcastss %%xmm4,%%ymm4 \n"
......@@ -4894,12 +4896,14 @@ void InterpolateRow_AVX2(uint8* dst_ptr, const uint8* src_ptr,
MEMOPREG(vmovdqu,0x00,1,4,1,ymm2)
"vpunpckhbw %%ymm2,%%ymm0,%%ymm1 \n"
"vpunpcklbw %%ymm2,%%ymm0,%%ymm0 \n"
"vpmaddubsw %%ymm5,%%ymm0,%%ymm0 \n"
"vpmaddubsw %%ymm5,%%ymm1,%%ymm1 \n"
"vpaddw %%ymm4,%%ymm0,%%ymm0 \n"
"vpsubb %%ymm4,%%ymm1,%%ymm1 \n"
"vpsubb %%ymm4,%%ymm0,%%ymm0 \n"
"vpmaddubsw %%ymm1,%%ymm5,%%ymm1 \n"
"vpmaddubsw %%ymm0,%%ymm5,%%ymm0 \n"
"vpaddw %%ymm4,%%ymm1,%%ymm1 \n"
"vpsrlw $0x7,%%ymm0,%%ymm0 \n"
"vpsrlw $0x7,%%ymm1,%%ymm1 \n"
"vpaddw %%ymm4,%%ymm0,%%ymm0 \n"
"vpsrlw $0x8,%%ymm1,%%ymm1 \n"
"vpsrlw $0x8,%%ymm0,%%ymm0 \n"
"vpackuswb %%ymm1,%%ymm0,%%ymm0 \n"
MEMOPMEM(vmovdqu,ymm0,0x00,1,0,1)
"lea " MEMLEA(0x20,1) ",%1 \n"
......
source/row_neon.cc
View file @ 3f4d8605
......@@ -2259,16 +2259,16 @@ void RAWToYRow_NEON(const uint8* src_raw, uint8* dst_y, int width) {
void InterpolateRow_NEON(uint8* dst_ptr,
const uint8* src_ptr, ptrdiff_t src_stride,
int dst_width, int source_y_fraction) {
int y1_fraction = source_y_fraction >> 1;
int y1_fraction = source_y_fraction;
asm volatile (
"cmp %4, #0 \n"
"beq 100f \n"
"add %2, %1 \n"
"cmp %4, #64 \n"
"cmp %4, #128 \n"
"beq 50f \n"
"vdup.8 d5, %4 \n"
"rsb %4, #128 \n"
"rsb %4, #256 \n"
"vdup.8 d4, %4 \n"
// General purpose row blend.
"1: \n"
......@@ -2281,8 +2281,8 @@ void InterpolateRow_NEON(uint8* dst_ptr,
"vmull.u8 q14, d1, d4 \n"
"vmlal.u8 q13, d2, d5 \n"
"vmlal.u8 q14, d3, d5 \n"
"vrshrn.u16 d0, q13, #7 \n"
"vrshrn.u16 d1, q14, #7 \n"
"vrshrn.u16 d0, q13, #8 \n"
"vrshrn.u16 d1, q14, #8 \n"
MEMACCESS(0)
"vst1.8 {q0}, [%0]! \n"
"bgt 1b \n"
......
source/row_neon64.cc
View file @ 3f4d8605
......@@ -2336,13 +2336,13 @@ void RAWToYRow_NEON(const uint8* src_raw, uint8* dst_y, int width) {
void InterpolateRow_NEON(uint8* dst_ptr,
const uint8* src_ptr, ptrdiff_t src_stride,
int dst_width, int source_y_fraction) {
int y1_fraction = source_y_fraction >> 1;
int y0_fraction = 128 - y1_fraction;
int y1_fraction = source_y_fraction;
int y0_fraction = 256 - y1_fraction;
const uint8* src_ptr1 = src_ptr + src_stride;
asm volatile (
"cmp %w4, #0 \n"
"b.eq 100f \n"
"cmp %w4, #64 \n"
"cmp %w4, #128 \n"
"b.eq 50f \n"
"dup v5.16b, %w4 \n"
......@@ -2358,8 +2358,8 @@ void InterpolateRow_NEON(uint8* dst_ptr,
"umull2 v3.8h, v0.16b, v4.16b \n"
"umlal v2.8h, v1.8b, v5.8b \n"
"umlal2 v3.8h, v1.16b, v5.16b \n"
"rshrn v0.8b, v2.8h, #7 \n"
"rshrn2 v0.16b, v3.8h, #7 \n"
"rshrn v0.8b, v2.8h, #8 \n"
"rshrn2 v0.16b, v3.8h, #8 \n"
MEMACCESS(0)
"st1 {v0.16b}, [%0], #16 \n"
"b.gt 1b \n"
......
source/row_win.cc
View file @ 3f4d8605
......@@ -5566,24 +5566,22 @@ void InterpolateRow_AVX2(uint8* dst_ptr, const uint8* src_ptr,
mov edx, [esp + 8 + 12] // src_stride
mov ecx, [esp + 8 + 16] // dst_width
mov eax, [esp + 8 + 20] // source_y_fraction (0..255)
shr eax, 1
// Dispatch to specialized filters if applicable.
cmp eax, 0
je xloop100 // 0 / 128. Blend 100 / 0.
je xloop100 // 0 / 256. Blend 100 / 0.
sub edi, esi
cmp eax, 64
je xloop50 // 64 / 128 is 0.50. Blend 50 / 50.
cmp eax, 128
je xloop50 // 128 /256 is 0.50. Blend 50 / 50.
vmovd xmm0, eax // high fraction 0..127
vmovd xmm0, eax // high fraction 0..255
neg eax
add eax, 128
vmovd xmm5, eax // low fraction 128..1
add eax, 256
vmovd xmm5, eax // low fraction 256..1
vpunpcklbw xmm5, xmm5, xmm0
vpunpcklwd xmm5, xmm5, xmm5
vpxor ymm0, ymm0, ymm0
vpermd ymm5, ymm0, ymm5
vbroadcastss ymm5, xmm5
mov eax, 0x00400040 // 64 for rounding.
mov eax, 0x80808080 // 128b for bias and rounding.
vmovd xmm4, eax
vbroadcastss ymm4, xmm4
......@@ -5591,13 +5589,15 @@ void InterpolateRow_AVX2(uint8* dst_ptr, const uint8* src_ptr,
vmovdqu ymm0, [esi]
vmovdqu ymm2, [esi + edx]
vpunpckhbw ymm1, ymm0, ymm2 // mutates
vpunpcklbw ymm0, ymm0, ymm2 // mutates
vpmaddubsw ymm0, ymm0, ymm5
vpmaddubsw ymm1, ymm1, ymm5
vpunpcklbw ymm0, ymm0, ymm2
vpsubb ymm1, ymm1, ymm4 // bias to signed image
vpsubb ymm0, ymm0, ymm4
vpmaddubsw ymm1, ymm5, ymm1
vpmaddubsw ymm0, ymm5, ymm0
vpaddw ymm1, ymm1, ymm4 // unbias and round
vpaddw ymm0, ymm0, ymm4
vpaddw ymm1, ymm1, ymm4
vpsrlw ymm0, ymm0, 7
vpsrlw ymm1, ymm1, 7
vpsrlw ymm1, ymm1, 8
vpsrlw ymm0, ymm0, 8
vpackuswb ymm0, ymm0, ymm1 // unmutates
vmovdqu [esi + edi], ymm0
lea esi, [esi + 32]
......@@ -5629,6 +5629,7 @@ void InterpolateRow_AVX2(uint8* dst_ptr, const uint8* src_ptr,
#endif // HAS_INTERPOLATEROW_AVX2
// Bilinear filter 16x2 -> 16x1
// TODO(fbarchard): Consider allowing 256 using memcpy.
__declspec(naked)
void InterpolateRow_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
ptrdiff_t src_stride, int dst_width,
......@@ -5636,28 +5637,27 @@ void InterpolateRow_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
__asm {
push esi
push edi
mov edi, [esp + 8 + 4] // dst_ptr
mov esi, [esp + 8 + 8] // src_ptr
mov edx, [esp + 8 + 12] // src_stride
mov ecx, [esp + 8 + 16] // dst_width
mov eax, [esp + 8 + 20] // source_y_fraction (0..255)
sub edi, esi
shr eax, 1
// Dispatch to specialized filters if applicable.
cmp eax, 0
je xloop100 // 0 / 128. Blend 100 / 0.
cmp eax, 64
je xloop50 // 64 / 128 is 0.50. Blend 50 / 50.
je xloop100 // 0 /256. Blend 100 / 0.
cmp eax, 128
je xloop50 // 128 / 256 is 0.50. Blend 50 / 50.
movd xmm0, eax // high fraction 0..127
movd xmm0, eax // high fraction 0..255
neg eax
add eax, 128
movd xmm5, eax // low fraction 128..1
add eax, 256
movd xmm5, eax // low fraction 255..1
punpcklbw xmm5, xmm0
punpcklwd xmm5, xmm5
pshufd xmm5, xmm5, 0
mov eax, 0x00400040 // 64 for rounding.
mov eax, 0x80808080 // 128 for biasing image to signed.
movd xmm4, eax
pshufd xmm4, xmm4, 0x00
......@@ -5667,14 +5667,18 @@ void InterpolateRow_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
movdqu xmm1, xmm0
punpcklbw xmm0, xmm2
punpckhbw xmm1, xmm2
pmaddubsw xmm0, xmm5
pmaddubsw xmm1, xmm5
paddw xmm0, xmm4
paddw xmm1, xmm4
psrlw xmm0, 7
psrlw xmm1, 7
packuswb xmm0, xmm1
movdqu [esi + edi], xmm0
psubb xmm0, xmm4 // bias image by -128
psubb xmm1, xmm4
movdqa xmm2, xmm5
movdqa xmm3, xmm5
pmaddubsw xmm2, xmm0
pmaddubsw xmm3, xmm1
paddw xmm2, xmm4
paddw xmm3, xmm4
psrlw xmm2, 8
psrlw xmm3, 8
packuswb xmm2, xmm3
movdqu [esi + edi], xmm2
lea esi, [esi + 16]
sub ecx, 16
jg xloop
......
unit_test/planar_test.cc
View file @ 3f4d8605
......@@ -940,7 +940,6 @@ TEST_F(LibYUVPlanarTest, TestARGBInterpolate) {
}
}
TEST_F(LibYUVPlanarTest, TestInterpolatePlane) {
SIMD_ALIGNED(uint8 orig_pixels_0[1280]);
SIMD_ALIGNED(uint8 orig_pixels_1[1280]);
......@@ -1024,7 +1023,7 @@ TEST_F(LibYUVPlanarTest, TestInterpolatePlane) {
#define TESTTERP(FMT_A, BPP_A, STRIDE_A, \
FMT_B, BPP_B, STRIDE_B, \
W1280, TERP, DIFF, N, NEG, OFF) \
W1280, TERP, N, NEG, OFF) \
TEST_F(LibYUVPlanarTest, ARGBInterpolate##TERP##N) { \
const int kWidth = ((W1280) > 0) ? (W1280) : 1; \
const int kHeight = benchmark_height_; \
......@@ -1050,16 +1049,9 @@ TEST_F(LibYUVPlanarTest, ARGBInterpolate##TERP##N) { \
dst_argb_opt, kStrideB, \
kWidth, NEG kHeight, TERP); \
} \
int max_diff = 0; \
for (int i = 0; i < kStrideB * kHeight; ++i) { \
int abs_diff = \
abs(static_cast<int>(dst_argb_c[i]) - \
static_cast<int>(dst_argb_opt[i])); \
if (abs_diff > max_diff) { \
max_diff = abs_diff; \
} \
EXPECT_EQ(dst_argb_c[i], dst_argb_opt[i]); \
} \
EXPECT_LE(max_diff, DIFF); \
free_aligned_buffer_64(src_argb_a); \
free_aligned_buffer_64(src_argb_b); \
free_aligned_buffer_64(dst_argb_c); \
......@@ -1067,16 +1059,10 @@ TEST_F(LibYUVPlanarTest, ARGBInterpolate##TERP##N) { \
}
#define TESTINTERPOLATE(TERP) \
TESTTERP(ARGB, 4, 1, ARGB, 4, 1, \
benchmark_width_ - 1, TERP, 1, _Any, +, 0) \
TESTTERP(ARGB, 4, 1, ARGB, 4, 1, \
benchmark_width_, TERP, 1, _Unaligned, +, 1) \
TESTTERP(ARGB, 4, 1, ARGB, 4, 1, \
benchmark_width_, TERP, 1, _Invert, -, 0) \
TESTTERP(ARGB, 4, 1, ARGB, 4, 1, \
benchmark_width_, TERP, 1, _Opt, +, 0) \
TESTTERP(ARGB, 4, 1, ARGB, 4, 1, \
benchmark_width_ - 1, TERP, 1, _Any_Invert, -, 0)
TESTTERP(ARGB, 4, 1, ARGB, 4, 1, benchmark_width_ - 1, TERP, _Any, +, 0) \
TESTTERP(ARGB, 4, 1, ARGB, 4, 1, benchmark_width_, TERP, _Unaligned, +, 1) \
TESTTERP(ARGB, 4, 1, ARGB, 4, 1, benchmark_width_, TERP, _Invert, -, 0) \
TESTTERP(ARGB, 4, 1, ARGB, 4, 1, benchmark_width_, TERP, _Opt, +, 0)
TESTINTERPOLATE(0)
TESTINTERPOLATE(64)
......
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