普通文本  |  3081行  |  107.22 KB

/*
 *  Copyright 2011 The LibYuv Project Authors. All rights reserved.
 *
 *  Use of this source code is governed by a BSD-style license
 *  that can be found in the LICENSE file in the root of the source
 *  tree. An additional intellectual property rights grant can be found
 *  in the file PATENTS.  All contributing project authors may
 *  be found in the AUTHORS file in the root of the source tree.
 */

#include "libyuv/scale.h"

#include <assert.h>
#include <string.h>
#include <stdlib.h>  // For getenv()

#include "libyuv/cpu_id.h"
#include "libyuv/planar_functions.h"  // For CopyPlane
#include "libyuv/row.h"

#ifdef __cplusplus
namespace libyuv {
extern "C" {
#endif

// Bilinear SSE2 is disabled.
#define SSE2_DISABLED 1

// Note: Some SSE2 reference manuals
// cpuvol1.pdf agner_instruction_tables.pdf 253666.pdf 253667.pdf

// Set the following flag to true to revert to only
// using the reference implementation ScalePlaneBox(), and
// NOT the optimized versions. Useful for debugging and
// when comparing the quality of the resulting YUV planes
// as produced by the optimized and non-optimized versions.
static bool use_reference_impl_ = false;

LIBYUV_API
void SetUseReferenceImpl(bool use) {
  use_reference_impl_ = use;
}

// ScaleRowDown2Int also used by planar functions

/**
 * NEON downscalers with interpolation.
 *
 * Provided by Fritz Koenig
 *
 */

#if !defined(YUV_DISABLE_ASM) && defined(__ARM_NEON__)
#define HAS_SCALEROWDOWN2_NEON
// Note - not static due to reuse in convert for 444 to 420.
void ScaleRowDown2_NEON(const uint8* src_ptr, ptrdiff_t /* src_stride */,
                        uint8* dst, int dst_width);

void ScaleRowDown2Int_NEON(const uint8* src_ptr, ptrdiff_t src_stride,
                           uint8* dst, int dst_width);

#define HAS_SCALEROWDOWN4_NEON
void ScaleRowDown4_NEON(const uint8* src_ptr, ptrdiff_t /* src_stride */,
                        uint8* dst_ptr, int dst_width);
void ScaleRowDown4Int_NEON(const uint8* src_ptr, ptrdiff_t src_stride,
                           uint8* dst_ptr, int dst_width);

#define HAS_SCALEROWDOWN34_NEON
// Down scale from 4 to 3 pixels. Use the neon multilane read/write
//  to load up the every 4th pixel into a 4 different registers.
// Point samples 32 pixels to 24 pixels.
void ScaleRowDown34_NEON(const uint8* src_ptr,
                         ptrdiff_t /* src_stride */,
                         uint8* dst_ptr, int dst_width);
void ScaleRowDown34_0_Int_NEON(const uint8* src_ptr,
                               ptrdiff_t src_stride,
                               uint8* dst_ptr, int dst_width);
void ScaleRowDown34_1_Int_NEON(const uint8* src_ptr,
                               ptrdiff_t src_stride,
                               uint8* dst_ptr, int dst_width);

#define HAS_SCALEROWDOWN38_NEON
// 32 -> 12
void ScaleRowDown38_NEON(const uint8* src_ptr,
                         ptrdiff_t /* src_stride */,
                         uint8* dst_ptr, int dst_width);
// 32x3 -> 12x1
void ScaleRowDown38_3_Int_NEON(const uint8* src_ptr,
                               ptrdiff_t src_stride,
                               uint8* dst_ptr, int dst_width);
// 32x2 -> 12x1
void ScaleRowDown38_2_Int_NEON(const uint8* src_ptr,
                               ptrdiff_t src_stride,
                               uint8* dst_ptr, int dst_width);
// 16x2 -> 16x1
#define HAS_SCALEFILTERROWS_NEON
void ScaleFilterRows_NEON(uint8* dst_ptr,
                          const uint8* src_ptr, ptrdiff_t src_stride,
                          int dst_width, int source_y_fraction);

/**
 * SSE2 downscalers with interpolation.
 *
 * Provided by Frank Barchard (fbarchard@google.com)
 *
 */


// Constants for SSSE3 code
#elif !defined(YUV_DISABLE_ASM) && \
    (defined(_M_IX86) || defined(__i386__) || defined(__x86_64__))

// GCC 4.2 on OSX has link error when passing static or const to inline.
// TODO(fbarchard): Use static const when gcc 4.2 support is dropped.
#ifdef __APPLE__
#define CONST
#else
#define CONST static const
#endif

// Offsets for source bytes 0 to 9
CONST uvec8 kShuf0 =
  { 0, 1, 3, 4, 5, 7, 8, 9, 128, 128, 128, 128, 128, 128, 128, 128 };

// Offsets for source bytes 11 to 20 with 8 subtracted = 3 to 12.
CONST uvec8 kShuf1 =
  { 3, 4, 5, 7, 8, 9, 11, 12, 128, 128, 128, 128, 128, 128, 128, 128 };

// Offsets for source bytes 21 to 31 with 16 subtracted = 5 to 31.
CONST uvec8 kShuf2 =
  { 5, 7, 8, 9, 11, 12, 13, 15, 128, 128, 128, 128, 128, 128, 128, 128 };

// Offsets for source bytes 0 to 10
CONST uvec8 kShuf01 =
  { 0, 1, 1, 2, 2, 3, 4, 5, 5, 6, 6, 7, 8, 9, 9, 10 };

// Offsets for source bytes 10 to 21 with 8 subtracted = 3 to 13.
CONST uvec8 kShuf11 =
  { 2, 3, 4, 5, 5, 6, 6, 7, 8, 9, 9, 10, 10, 11, 12, 13 };

// Offsets for source bytes 21 to 31 with 16 subtracted = 5 to 31.
CONST uvec8 kShuf21 =
  { 5, 6, 6, 7, 8, 9, 9, 10, 10, 11, 12, 13, 13, 14, 14, 15 };

// Coefficients for source bytes 0 to 10
CONST uvec8 kMadd01 =
  { 3, 1, 2, 2, 1, 3, 3, 1, 2, 2, 1, 3, 3, 1, 2, 2 };

// Coefficients for source bytes 10 to 21
CONST uvec8 kMadd11 =
  { 1, 3, 3, 1, 2, 2, 1, 3, 3, 1, 2, 2, 1, 3, 3, 1 };

// Coefficients for source bytes 21 to 31
CONST uvec8 kMadd21 =
  { 2, 2, 1, 3, 3, 1, 2, 2, 1, 3, 3, 1, 2, 2, 1, 3 };

// Coefficients for source bytes 21 to 31
CONST vec16 kRound34 =
  { 2, 2, 2, 2, 2, 2, 2, 2 };

CONST uvec8 kShuf38a =
  { 0, 3, 6, 8, 11, 14, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 };

CONST uvec8 kShuf38b =
  { 128, 128, 128, 128, 128, 128, 0, 3, 6, 8, 11, 14, 128, 128, 128, 128 };

// Arrange words 0,3,6 into 0,1,2
CONST uvec8 kShufAc =
  { 0, 1, 6, 7, 12, 13, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 };

// Arrange words 0,3,6 into 3,4,5
CONST uvec8 kShufAc3 =
  { 128, 128, 128, 128, 128, 128, 0, 1, 6, 7, 12, 13, 128, 128, 128, 128 };

// Scaling values for boxes of 3x3 and 2x3
CONST uvec16 kScaleAc33 =
  { 65536 / 9, 65536 / 9, 65536 / 6, 65536 / 9, 65536 / 9, 65536 / 6, 0, 0 };

// Arrange first value for pixels 0,1,2,3,4,5
CONST uvec8 kShufAb0 =
  { 0, 128, 3, 128, 6, 128, 8, 128, 11, 128, 14, 128, 128, 128, 128, 128 };

// Arrange second value for pixels 0,1,2,3,4,5
CONST uvec8 kShufAb1 =
  { 1, 128, 4, 128, 7, 128, 9, 128, 12, 128, 15, 128, 128, 128, 128, 128 };

// Arrange third value for pixels 0,1,2,3,4,5
CONST uvec8 kShufAb2 =
  { 2, 128, 5, 128, 128, 128, 10, 128, 13, 128, 128, 128, 128, 128, 128, 128 };

// Scaling values for boxes of 3x2 and 2x2
CONST uvec16 kScaleAb2 =
  { 65536 / 3, 65536 / 3, 65536 / 2, 65536 / 3, 65536 / 3, 65536 / 2, 0, 0 };
#endif

#if !defined(YUV_DISABLE_ASM) && defined(_M_IX86)

#define HAS_SCALEROWDOWN2_SSE2
// Reads 32 pixels, throws half away and writes 16 pixels.
// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 16 byte aligned.
__declspec(naked) __declspec(align(16))
static void ScaleRowDown2_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
                               uint8* dst_ptr, int dst_width) {
  __asm {
    mov        eax, [esp + 4]        // src_ptr
                                     // src_stride ignored
    mov        edx, [esp + 12]       // dst_ptr
    mov        ecx, [esp + 16]       // dst_width
    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
    psrlw      xmm5, 8

    align      16
  wloop:
    movdqa     xmm0, [eax]
    movdqa     xmm1, [eax + 16]
    lea        eax,  [eax + 32]
    pand       xmm0, xmm5
    pand       xmm1, xmm5
    packuswb   xmm0, xmm1
    sub        ecx, 16
    movdqa     [edx], xmm0
    lea        edx, [edx + 16]
    jg         wloop

    ret
  }
}
// Blends 32x2 rectangle to 16x1.
// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 16 byte aligned.
__declspec(naked) __declspec(align(16))
void ScaleRowDown2Int_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
                           uint8* dst_ptr, int dst_width) {
  __asm {
    push       esi
    mov        eax, [esp + 4 + 4]    // src_ptr
    mov        esi, [esp + 4 + 8]    // src_stride
    mov        edx, [esp + 4 + 12]   // dst_ptr
    mov        ecx, [esp + 4 + 16]   // dst_width
    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
    psrlw      xmm5, 8

    align      16
  wloop:
    movdqa     xmm0, [eax]
    movdqa     xmm1, [eax + 16]
    movdqa     xmm2, [eax + esi]
    movdqa     xmm3, [eax + esi + 16]
    lea        eax,  [eax + 32]
    pavgb      xmm0, xmm2            // average rows
    pavgb      xmm1, xmm3

    movdqa     xmm2, xmm0            // average columns (32 to 16 pixels)
    psrlw      xmm0, 8
    movdqa     xmm3, xmm1
    psrlw      xmm1, 8
    pand       xmm2, xmm5
    pand       xmm3, xmm5
    pavgw      xmm0, xmm2
    pavgw      xmm1, xmm3
    packuswb   xmm0, xmm1

    sub        ecx, 16
    movdqa     [edx], xmm0
    lea        edx, [edx + 16]
    jg         wloop

    pop        esi
    ret
  }
}

// Reads 32 pixels, throws half away and writes 16 pixels.
// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 16 byte aligned.
__declspec(naked) __declspec(align(16))
static void ScaleRowDown2_Unaligned_SSE2(const uint8* src_ptr,
                                         ptrdiff_t src_stride,
                                         uint8* dst_ptr, int dst_width) {
  __asm {
    mov        eax, [esp + 4]        // src_ptr
                                     // src_stride ignored
    mov        edx, [esp + 12]       // dst_ptr
    mov        ecx, [esp + 16]       // dst_width
    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
    psrlw      xmm5, 8

    align      16
  wloop:
    movdqu     xmm0, [eax]
    movdqu     xmm1, [eax + 16]
    lea        eax,  [eax + 32]
    pand       xmm0, xmm5
    pand       xmm1, xmm5
    packuswb   xmm0, xmm1
    sub        ecx, 16
    movdqu     [edx], xmm0
    lea        edx, [edx + 16]
    jg         wloop

    ret
  }
}
// Blends 32x2 rectangle to 16x1.
// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 16 byte aligned.
__declspec(naked) __declspec(align(16))
static void ScaleRowDown2Int_Unaligned_SSE2(const uint8* src_ptr,
                                            ptrdiff_t src_stride,
                                            uint8* dst_ptr, int dst_width) {
  __asm {
    push       esi
    mov        eax, [esp + 4 + 4]    // src_ptr
    mov        esi, [esp + 4 + 8]    // src_stride
    mov        edx, [esp + 4 + 12]   // dst_ptr
    mov        ecx, [esp + 4 + 16]   // dst_width
    pcmpeqb    xmm5, xmm5            // generate mask 0x00ff00ff
    psrlw      xmm5, 8

    align      16
  wloop:
    movdqu     xmm0, [eax]
    movdqu     xmm1, [eax + 16]
    movdqu     xmm2, [eax + esi]
    movdqu     xmm3, [eax + esi + 16]
    lea        eax,  [eax + 32]
    pavgb      xmm0, xmm2            // average rows
    pavgb      xmm1, xmm3

    movdqa     xmm2, xmm0            // average columns (32 to 16 pixels)
    psrlw      xmm0, 8
    movdqa     xmm3, xmm1
    psrlw      xmm1, 8
    pand       xmm2, xmm5
    pand       xmm3, xmm5
    pavgw      xmm0, xmm2
    pavgw      xmm1, xmm3
    packuswb   xmm0, xmm1

    sub        ecx, 16
    movdqu     [edx], xmm0
    lea        edx, [edx + 16]
    jg         wloop

    pop        esi
    ret
  }
}

#define HAS_SCALEROWDOWN4_SSE2
// Point samples 32 pixels to 8 pixels.
// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
__declspec(naked) __declspec(align(16))
static void ScaleRowDown4_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
                               uint8* dst_ptr, int dst_width) {
  __asm {
    mov        eax, [esp + 4]        // src_ptr
                                     // src_stride ignored
    mov        edx, [esp + 12]       // dst_ptr
    mov        ecx, [esp + 16]       // dst_width
    pcmpeqb    xmm5, xmm5            // generate mask 0x000000ff
    psrld      xmm5, 24

    align      16
  wloop:
    movdqa     xmm0, [eax]
    movdqa     xmm1, [eax + 16]
    lea        eax,  [eax + 32]
    pand       xmm0, xmm5
    pand       xmm1, xmm5
    packuswb   xmm0, xmm1
    packuswb   xmm0, xmm0
    sub        ecx, 8
    movq       qword ptr [edx], xmm0
    lea        edx, [edx + 8]
    jg         wloop

    ret
  }
}

// Blends 32x4 rectangle to 8x1.
// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
__declspec(naked) __declspec(align(16))
static void ScaleRowDown4Int_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
                                  uint8* dst_ptr, int dst_width) {
  __asm {
    push       esi
    push       edi
    mov        eax, [esp + 8 + 4]    // src_ptr
    mov        esi, [esp + 8 + 8]    // src_stride
    mov        edx, [esp + 8 + 12]   // dst_ptr
    mov        ecx, [esp + 8 + 16]   // dst_width
    lea        edi, [esi + esi * 2]  // src_stride * 3
    pcmpeqb    xmm7, xmm7            // generate mask 0x00ff00ff
    psrlw      xmm7, 8

    align      16
  wloop:
    movdqa     xmm0, [eax]
    movdqa     xmm1, [eax + 16]
    movdqa     xmm2, [eax + esi]
    movdqa     xmm3, [eax + esi + 16]
    pavgb      xmm0, xmm2            // average rows
    pavgb      xmm1, xmm3
    movdqa     xmm2, [eax + esi * 2]
    movdqa     xmm3, [eax + esi * 2 + 16]
    movdqa     xmm4, [eax + edi]
    movdqa     xmm5, [eax + edi + 16]
    lea        eax, [eax + 32]
    pavgb      xmm2, xmm4
    pavgb      xmm3, xmm5
    pavgb      xmm0, xmm2
    pavgb      xmm1, xmm3

    movdqa     xmm2, xmm0            // average columns (32 to 16 pixels)
    psrlw      xmm0, 8
    movdqa     xmm3, xmm1
    psrlw      xmm1, 8
    pand       xmm2, xmm7
    pand       xmm3, xmm7
    pavgw      xmm0, xmm2
    pavgw      xmm1, xmm3
    packuswb   xmm0, xmm1

    movdqa     xmm2, xmm0            // average columns (16 to 8 pixels)
    psrlw      xmm0, 8
    pand       xmm2, xmm7
    pavgw      xmm0, xmm2
    packuswb   xmm0, xmm0

    sub        ecx, 8
    movq       qword ptr [edx], xmm0
    lea        edx, [edx + 8]
    jg         wloop

    pop        edi
    pop        esi
    ret
  }
}

#define HAS_SCALEROWDOWN8_SSE2
// Point samples 32 pixels to 4 pixels.
// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 4 byte aligned.
__declspec(naked) __declspec(align(16))
static void ScaleRowDown8_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
                               uint8* dst_ptr, int dst_width) {
  __asm {
    mov        eax, [esp + 4]        // src_ptr
                                     // src_stride ignored
    mov        edx, [esp + 12]       // dst_ptr
    mov        ecx, [esp + 16]       // dst_width
    pcmpeqb    xmm5, xmm5            // generate mask isolating 1 src 8 bytes
    psrlq      xmm5, 56

    align      16
  wloop:
    movdqa     xmm0, [eax]
    movdqa     xmm1, [eax + 16]
    lea        eax,  [eax + 32]
    pand       xmm0, xmm5
    pand       xmm1, xmm5
    packuswb   xmm0, xmm1  // 32->16
    packuswb   xmm0, xmm0  // 16->8
    packuswb   xmm0, xmm0  // 8->4
    sub        ecx, 4
    movd       dword ptr [edx], xmm0
    lea        edx, [edx + 4]
    jg         wloop

    ret
  }
}

// Blends 32x8 rectangle to 4x1.
// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 4 byte aligned.
__declspec(naked) __declspec(align(16))
static void ScaleRowDown8Int_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
                                  uint8* dst_ptr, int dst_width) {
  __asm {
    push       esi
    push       edi
    push       ebp
    mov        eax, [esp + 12 + 4]   // src_ptr
    mov        esi, [esp + 12 + 8]   // src_stride
    mov        edx, [esp + 12 + 12]  // dst_ptr
    mov        ecx, [esp + 12 + 16]  // dst_width
    lea        edi, [esi + esi * 2]  // src_stride * 3
    pxor       xmm7, xmm7

    align      16
  wloop:
    movdqa     xmm0, [eax]           // average 8 rows to 1
    movdqa     xmm1, [eax + 16]
    movdqa     xmm2, [eax + esi]
    movdqa     xmm3, [eax + esi + 16]
    pavgb      xmm0, xmm2
    pavgb      xmm1, xmm3
    movdqa     xmm2, [eax + esi * 2]
    movdqa     xmm3, [eax + esi * 2 + 16]
    movdqa     xmm4, [eax + edi]
    movdqa     xmm5, [eax + edi + 16]
    lea        ebp, [eax + esi * 4]
    lea        eax, [eax + 32]
    pavgb      xmm2, xmm4
    pavgb      xmm3, xmm5
    pavgb      xmm0, xmm2
    pavgb      xmm1, xmm3

    movdqa     xmm2, [ebp]
    movdqa     xmm3, [ebp + 16]
    movdqa     xmm4, [ebp + esi]
    movdqa     xmm5, [ebp + esi + 16]
    pavgb      xmm2, xmm4
    pavgb      xmm3, xmm5
    movdqa     xmm4, [ebp + esi * 2]
    movdqa     xmm5, [ebp + esi * 2 + 16]
    movdqa     xmm6, [ebp + edi]
    pavgb      xmm4, xmm6
    movdqa     xmm6, [ebp + edi + 16]
    pavgb      xmm5, xmm6
    pavgb      xmm2, xmm4
    pavgb      xmm3, xmm5
    pavgb      xmm0, xmm2
    pavgb      xmm1, xmm3

    psadbw     xmm0, xmm7            // average 32 pixels to 4
    psadbw     xmm1, xmm7
    pshufd     xmm0, xmm0, 0xd8      // x1x0 -> xx01
    pshufd     xmm1, xmm1, 0x8d      // x3x2 -> 32xx
    por        xmm0, xmm1            //      -> 3201
    psrlw      xmm0, 3
    packuswb   xmm0, xmm0
    packuswb   xmm0, xmm0

    sub        ecx, 4
    movd       dword ptr [edx], xmm0
    lea        edx, [edx + 4]
    jg         wloop

    pop        ebp
    pop        edi
    pop        esi
    ret
  }
}

#define HAS_SCALEROWDOWN34_SSSE3
// Point samples 32 pixels to 24 pixels.
// Produces three 8 byte values. For each 8 bytes, 16 bytes are read.
// Then shuffled to do the scaling.

// Note that movdqa+palign may be better than movdqu.
// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
__declspec(naked) __declspec(align(16))
static void ScaleRowDown34_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride,
                                 uint8* dst_ptr, int dst_width) {
  __asm {
    mov        eax, [esp + 4]        // src_ptr
                                     // src_stride ignored
    mov        edx, [esp + 12]       // dst_ptr
    mov        ecx, [esp + 16]       // dst_width
    movdqa     xmm3, kShuf0
    movdqa     xmm4, kShuf1
    movdqa     xmm5, kShuf2

    align      16
  wloop:
    movdqa     xmm0, [eax]
    movdqa     xmm1, [eax + 16]
    lea        eax,  [eax + 32]
    movdqa     xmm2, xmm1
    palignr    xmm1, xmm0, 8
    pshufb     xmm0, xmm3
    pshufb     xmm1, xmm4
    pshufb     xmm2, xmm5
    movq       qword ptr [edx], xmm0
    movq       qword ptr [edx + 8], xmm1
    movq       qword ptr [edx + 16], xmm2
    lea        edx, [edx + 24]
    sub        ecx, 24
    jg         wloop

    ret
  }
}

// Blends 32x2 rectangle to 24x1
// Produces three 8 byte values. For each 8 bytes, 16 bytes are read.
// Then shuffled to do the scaling.

// Register usage:
// xmm0 src_row 0
// xmm1 src_row 1
// xmm2 shuf 0
// xmm3 shuf 1
// xmm4 shuf 2
// xmm5 madd 0
// xmm6 madd 1
// xmm7 kRound34

// Note that movdqa+palign may be better than movdqu.
// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
__declspec(naked) __declspec(align(16))
static void ScaleRowDown34_1_Int_SSSE3(const uint8* src_ptr,
                                       ptrdiff_t src_stride,
                                       uint8* dst_ptr, int dst_width) {
  __asm {
    push       esi
    mov        eax, [esp + 4 + 4]    // src_ptr
    mov        esi, [esp + 4 + 8]    // src_stride
    mov        edx, [esp + 4 + 12]   // dst_ptr
    mov        ecx, [esp + 4 + 16]   // dst_width
    movdqa     xmm2, kShuf01
    movdqa     xmm3, kShuf11
    movdqa     xmm4, kShuf21
    movdqa     xmm5, kMadd01
    movdqa     xmm6, kMadd11
    movdqa     xmm7, kRound34

    align      16
  wloop:
    movdqa     xmm0, [eax]           // pixels 0..7
    movdqa     xmm1, [eax + esi]
    pavgb      xmm0, xmm1
    pshufb     xmm0, xmm2
    pmaddubsw  xmm0, xmm5
    paddsw     xmm0, xmm7
    psrlw      xmm0, 2
    packuswb   xmm0, xmm0
    movq       qword ptr [edx], xmm0
    movdqu     xmm0, [eax + 8]       // pixels 8..15
    movdqu     xmm1, [eax + esi + 8]
    pavgb      xmm0, xmm1
    pshufb     xmm0, xmm3
    pmaddubsw  xmm0, xmm6
    paddsw     xmm0, xmm7
    psrlw      xmm0, 2
    packuswb   xmm0, xmm0
    movq       qword ptr [edx + 8], xmm0
    movdqa     xmm0, [eax + 16]      // pixels 16..23
    movdqa     xmm1, [eax + esi + 16]
    lea        eax, [eax + 32]
    pavgb      xmm0, xmm1
    pshufb     xmm0, xmm4
    movdqa     xmm1, kMadd21
    pmaddubsw  xmm0, xmm1
    paddsw     xmm0, xmm7
    psrlw      xmm0, 2
    packuswb   xmm0, xmm0
    sub        ecx, 24
    movq       qword ptr [edx + 16], xmm0
    lea        edx, [edx + 24]
    jg         wloop

    pop        esi
    ret
  }
}

// Note that movdqa+palign may be better than movdqu.
// Alignment requirement: src_ptr 16 byte aligned, dst_ptr 8 byte aligned.
__declspec(naked) __declspec(align(16))
static void ScaleRowDown34_0_Int_SSSE3(const uint8* src_ptr,
                                       ptrdiff_t src_stride,
                                       uint8* dst_ptr, int dst_width) {
  __asm {
    push       esi
    mov        eax, [esp + 4 + 4]    // src_ptr
    mov        esi, [esp + 4 + 8]    // src_stride
    mov        edx, [esp + 4 + 12]   // dst_ptr
    mov        ecx, [esp + 4 + 16]   // dst_width
    movdqa     xmm2, kShuf01
    movdqa     xmm3, kShuf11
    movdqa     xmm4, kShuf21
    movdqa     xmm5, kMadd01
    movdqa     xmm6, kMadd11
    movdqa     xmm7, kRound34

    align      16
  wloop:
    movdqa     xmm0, [eax]           // pixels 0..7
    movdqa     xmm1, [eax + esi]
    pavgb      xmm1, xmm0
    pavgb      xmm0, xmm1
    pshufb     xmm0, xmm2
    pmaddubsw  xmm0, xmm5
    paddsw     xmm0, xmm7
    psrlw      xmm0, 2
    packuswb   xmm0, xmm0
    movq       qword ptr [edx], xmm0
    movdqu     xmm0, [eax + 8]       // pixels 8..15
    movdqu     xmm1, [eax + esi + 8]
    pavgb      xmm1, xmm0
    pavgb      xmm0, xmm1
    pshufb     xmm0, xmm3
    pmaddubsw  xmm0, xmm6
    paddsw     xmm0, xmm7
    psrlw      xmm0, 2
    packuswb   xmm0, xmm0
    movq       qword ptr [edx + 8], xmm0
    movdqa     xmm0, [eax + 16]      // pixels 16..23
    movdqa     xmm1, [eax + esi + 16]
    lea        eax, [eax + 32]
    pavgb      xmm1, xmm0
    pavgb      xmm0, xmm1
    pshufb     xmm0, xmm4
    movdqa     xmm1, kMadd21
    pmaddubsw  xmm0, xmm1
    paddsw     xmm0, xmm7
    psrlw      xmm0, 2
    packuswb   xmm0, xmm0
    sub        ecx, 24
    movq       qword ptr [edx + 16], xmm0
    lea        edx, [edx+24]
    jg         wloop

    pop        esi
    ret
  }
}

#define HAS_SCALEROWDOWN38_SSSE3
// 3/8 point sampler

// Scale 32 pixels to 12
__declspec(naked) __declspec(align(16))
static void ScaleRowDown38_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride,
                                 uint8* dst_ptr, int dst_width) {
  __asm {
    mov        eax, [esp + 4]        // src_ptr
                                     // src_stride ignored
    mov        edx, [esp + 12]       // dst_ptr
    mov        ecx, [esp + 16]       // dst_width
    movdqa     xmm4, kShuf38a
    movdqa     xmm5, kShuf38b

    align      16
  xloop:
    movdqa     xmm0, [eax]           // 16 pixels -> 0,1,2,3,4,5
    movdqa     xmm1, [eax + 16]      // 16 pixels -> 6,7,8,9,10,11
    lea        eax, [eax + 32]
    pshufb     xmm0, xmm4
    pshufb     xmm1, xmm5
    paddusb    xmm0, xmm1

    sub        ecx, 12
    movq       qword ptr [edx], xmm0 // write 12 pixels
    movhlps    xmm1, xmm0
    movd       [edx + 8], xmm1
    lea        edx, [edx + 12]
    jg         xloop

    ret
  }
}

// Scale 16x3 pixels to 6x1 with interpolation
__declspec(naked) __declspec(align(16))
static void ScaleRowDown38_3_Int_SSSE3(const uint8* src_ptr,
                                       ptrdiff_t src_stride,
                                       uint8* dst_ptr, int dst_width) {
  __asm {
    push       esi
    mov        eax, [esp + 4 + 4]    // src_ptr
    mov        esi, [esp + 4 + 8]    // src_stride
    mov        edx, [esp + 4 + 12]   // dst_ptr
    mov        ecx, [esp + 4 + 16]   // dst_width
    movdqa     xmm2, kShufAc
    movdqa     xmm3, kShufAc3
    movdqa     xmm4, kScaleAc33
    pxor       xmm5, xmm5

    align      16
  xloop:
    movdqa     xmm0, [eax]           // sum up 3 rows into xmm0/1
    movdqa     xmm6, [eax + esi]
    movhlps    xmm1, xmm0
    movhlps    xmm7, xmm6
    punpcklbw  xmm0, xmm5
    punpcklbw  xmm1, xmm5
    punpcklbw  xmm6, xmm5
    punpcklbw  xmm7, xmm5
    paddusw    xmm0, xmm6
    paddusw    xmm1, xmm7
    movdqa     xmm6, [eax + esi * 2]
    lea        eax, [eax + 16]
    movhlps    xmm7, xmm6
    punpcklbw  xmm6, xmm5
    punpcklbw  xmm7, xmm5
    paddusw    xmm0, xmm6
    paddusw    xmm1, xmm7

    movdqa     xmm6, xmm0            // 8 pixels -> 0,1,2 of xmm6
    psrldq     xmm0, 2
    paddusw    xmm6, xmm0
    psrldq     xmm0, 2
    paddusw    xmm6, xmm0
    pshufb     xmm6, xmm2

    movdqa     xmm7, xmm1            // 8 pixels -> 3,4,5 of xmm6
    psrldq     xmm1, 2
    paddusw    xmm7, xmm1
    psrldq     xmm1, 2
    paddusw    xmm7, xmm1
    pshufb     xmm7, xmm3
    paddusw    xmm6, xmm7

    pmulhuw    xmm6, xmm4            // divide by 9,9,6, 9,9,6
    packuswb   xmm6, xmm6

    sub        ecx, 6
    movd       [edx], xmm6           // write 6 pixels
    psrlq      xmm6, 16
    movd       [edx + 2], xmm6
    lea        edx, [edx + 6]
    jg         xloop

    pop        esi
    ret
  }
}

// Scale 16x2 pixels to 6x1 with interpolation
__declspec(naked) __declspec(align(16))
static void ScaleRowDown38_2_Int_SSSE3(const uint8* src_ptr,
                                       ptrdiff_t src_stride,
                                       uint8* dst_ptr, int dst_width) {
  __asm {
    push       esi
    mov        eax, [esp + 4 + 4]    // src_ptr
    mov        esi, [esp + 4 + 8]    // src_stride
    mov        edx, [esp + 4 + 12]   // dst_ptr
    mov        ecx, [esp + 4 + 16]   // dst_width
    movdqa     xmm2, kShufAb0
    movdqa     xmm3, kShufAb1
    movdqa     xmm4, kShufAb2
    movdqa     xmm5, kScaleAb2

    align      16
  xloop:
    movdqa     xmm0, [eax]           // average 2 rows into xmm0
    pavgb      xmm0, [eax + esi]
    lea        eax, [eax + 16]

    movdqa     xmm1, xmm0            // 16 pixels -> 0,1,2,3,4,5 of xmm1
    pshufb     xmm1, xmm2
    movdqa     xmm6, xmm0
    pshufb     xmm6, xmm3
    paddusw    xmm1, xmm6
    pshufb     xmm0, xmm4
    paddusw    xmm1, xmm0

    pmulhuw    xmm1, xmm5            // divide by 3,3,2, 3,3,2
    packuswb   xmm1, xmm1

    sub        ecx, 6
    movd       [edx], xmm1           // write 6 pixels
    psrlq      xmm1, 16
    movd       [edx + 2], xmm1
    lea        edx, [edx + 6]
    jg         xloop

    pop        esi
    ret
  }
}

#define HAS_SCALEADDROWS_SSE2

// Reads 16xN bytes and produces 16 shorts at a time.
__declspec(naked) __declspec(align(16))
static void ScaleAddRows_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
                              uint16* dst_ptr, int src_width,
                              int src_height) {
  __asm {
    push       esi
    push       edi
    push       ebx
    push       ebp
    mov        esi, [esp + 16 + 4]   // src_ptr
    mov        edx, [esp + 16 + 8]   // src_stride
    mov        edi, [esp + 16 + 12]  // dst_ptr
    mov        ecx, [esp + 16 + 16]  // dst_width
    mov        ebx, [esp + 16 + 20]  // height
    pxor       xmm4, xmm4
    dec        ebx

    align      16
  xloop:
    // first row
    movdqa     xmm0, [esi]
    lea        eax, [esi + edx]
    movdqa     xmm1, xmm0
    punpcklbw  xmm0, xmm4
    punpckhbw  xmm1, xmm4
    lea        esi, [esi + 16]
    mov        ebp, ebx
    test       ebp, ebp
    je         ydone

    // sum remaining rows
    align      16
  yloop:
    movdqa     xmm2, [eax]       // read 16 pixels
    lea        eax, [eax + edx]  // advance to next row
    movdqa     xmm3, xmm2
    punpcklbw  xmm2, xmm4
    punpckhbw  xmm3, xmm4
    paddusw    xmm0, xmm2        // sum 16 words
    paddusw    xmm1, xmm3
    sub        ebp, 1
    jg         yloop
  ydone:
    movdqa     [edi], xmm0
    movdqa     [edi + 16], xmm1
    lea        edi, [edi + 32]

    sub        ecx, 16
    jg         xloop

    pop        ebp
    pop        ebx
    pop        edi
    pop        esi
    ret
  }
}

#ifndef SSE2_DISABLED
// Bilinear row filtering combines 16x2 -> 16x1. SSE2 version.
// Normal formula for bilinear interpolation is:
//   source_y_fraction * row1 + (1 - source_y_fraction) row0
// SSE2 version using the a single multiply of difference:
//   source_y_fraction * (row1 - row0) + row0
#define HAS_SCALEFILTERROWS_SSE2_DISABLED
__declspec(naked) __declspec(align(16))
static void ScaleFilterRows_SSE2(uint8* dst_ptr, const uint8* src_ptr,
                                 ptrdiff_t src_stride, int dst_width,
                                 int source_y_fraction) {
  __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
    cmp        eax, 0
    je         xloop1
    cmp        eax, 128
    je         xloop2

    movd       xmm5, eax            // xmm5 = y fraction
    punpcklbw  xmm5, xmm5
    punpcklwd  xmm5, xmm5
    pshufd     xmm5, xmm5, 0
    pxor       xmm4, xmm4

    align      16
  xloop:
    movdqa     xmm0, [esi]  // row0
    movdqa     xmm2, [esi + edx]  // row1
    movdqa     xmm1, xmm0
    movdqa     xmm3, xmm2
    punpcklbw  xmm2, xmm4
    punpckhbw  xmm3, xmm4
    punpcklbw  xmm0, xmm4
    punpckhbw  xmm1, xmm4
    psubw      xmm2, xmm0  // row1 - row0
    psubw      xmm3, xmm1
    pmulhw     xmm2, xmm5  // scale diff
    pmulhw     xmm3, xmm5
    paddw      xmm0, xmm2  // sum rows
    paddw      xmm1, xmm3
    packuswb   xmm0, xmm1
    sub        ecx, 16
    movdqa     [esi + edi], xmm0
    lea        esi, [esi + 16]
    jg         xloop

    punpckhbw  xmm0, xmm0           // duplicate last pixel for filtering
    pshufhw    xmm0, xmm0, 0xff
    punpckhqdq xmm0, xmm0
    movdqa     [esi + edi], xmm0
    pop        edi
    pop        esi
    ret

    align      16
  xloop1:
    movdqa     xmm0, [esi]
    sub        ecx, 16
    movdqa     [esi + edi], xmm0
    lea        esi, [esi + 16]
    jg         xloop1

    punpckhbw  xmm0, xmm0           // duplicate last pixel for filtering
    pshufhw    xmm0, xmm0, 0xff
    punpckhqdq xmm0, xmm0
    movdqa     [esi + edi], xmm0
    pop        edi
    pop        esi
    ret

    align      16
  xloop2:
    movdqa     xmm0, [esi]
    pavgb      xmm0, [esi + edx]
    sub        ecx, 16
    movdqa     [esi + edi], xmm0
    lea        esi, [esi + 16]
    jg         xloop2

    punpckhbw  xmm0, xmm0           // duplicate last pixel for filtering
    pshufhw    xmm0, xmm0, 0xff
    punpckhqdq xmm0, xmm0
    movdqa     [esi + edi], xmm0
    pop        edi
    pop        esi
    ret
  }
}
#endif  // SSE2_DISABLED
// Bilinear row filtering combines 16x2 -> 16x1. SSSE3 version.
#define HAS_SCALEFILTERROWS_SSSE3
__declspec(naked) __declspec(align(16))
static void ScaleFilterRows_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
                                  ptrdiff_t src_stride, int dst_width,
                                  int source_y_fraction) {
  __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
    cmp        eax, 0
    je         xloop1
    cmp        eax, 64
    je         xloop2
    movd       xmm0, eax  // high fraction 0..127
    neg        eax
    add        eax, 128
    movd       xmm5, eax  // low fraction 128..1
    punpcklbw  xmm5, xmm0
    punpcklwd  xmm5, xmm5
    pshufd     xmm5, xmm5, 0

    align      16
  xloop:
    movdqa     xmm0, [esi]
    movdqa     xmm2, [esi + edx]
    movdqa     xmm1, xmm0
    punpcklbw  xmm0, xmm2
    punpckhbw  xmm1, xmm2
    pmaddubsw  xmm0, xmm5
    pmaddubsw  xmm1, xmm5
    psrlw      xmm0, 7
    psrlw      xmm1, 7
    packuswb   xmm0, xmm1
    sub        ecx, 16
    movdqa     [esi + edi], xmm0
    lea        esi, [esi + 16]
    jg         xloop

    punpckhbw  xmm0, xmm0           // duplicate last pixel for filtering
    pshufhw    xmm0, xmm0, 0xff
    punpckhqdq xmm0, xmm0
    movdqa     [esi + edi], xmm0

    pop        edi
    pop        esi
    ret

    align      16
  xloop1:
    movdqa     xmm0, [esi]
    sub        ecx, 16
    movdqa     [esi + edi], xmm0
    lea        esi, [esi + 16]
    jg         xloop1

    punpckhbw  xmm0, xmm0
    pshufhw    xmm0, xmm0, 0xff
    punpckhqdq xmm0, xmm0
    movdqa     [esi + edi], xmm0
    pop        edi
    pop        esi
    ret

    align      16
  xloop2:
    movdqa     xmm0, [esi]
    pavgb      xmm0, [esi + edx]
    sub        ecx, 16
    movdqa     [esi + edi], xmm0
    lea        esi, [esi + 16]
    jg         xloop2

    punpckhbw  xmm0, xmm0
    pshufhw    xmm0, xmm0, 0xff
    punpckhqdq xmm0, xmm0
    movdqa     [esi + edi], xmm0
    pop        edi
    pop        esi
    ret
  }
}

#elif !defined(YUV_DISABLE_ASM) && (defined(__x86_64__) || defined(__i386__))

// GCC versions of row functions are verbatim conversions from Visual C.
// Generated using gcc disassembly on Visual C object file:
// objdump -D yuvscaler.obj >yuvscaler.txt
#define HAS_SCALEROWDOWN2_SSE2
static void ScaleRowDown2_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
                               uint8* dst_ptr, int dst_width) {
  asm volatile (
    "pcmpeqb   %%xmm5,%%xmm5                   \n"
    "psrlw     $0x8,%%xmm5                     \n"
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqa    (%0),%%xmm0                     \n"
    "movdqa    0x10(%0),%%xmm1                 \n"
    "lea       0x20(%0),%0                     \n"
    "pand      %%xmm5,%%xmm0                   \n"
    "pand      %%xmm5,%%xmm1                   \n"
    "packuswb  %%xmm1,%%xmm0                   \n"
    "movdqa    %%xmm0,(%1)                     \n"
    "lea       0x10(%1),%1                     \n"
    "sub       $0x10,%2                        \n"
    "jg        1b                              \n"
  : "+r"(src_ptr),    // %0
    "+r"(dst_ptr),    // %1
    "+r"(dst_width)   // %2
  :
  : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm5"
#endif
  );
}

void ScaleRowDown2Int_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
                           uint8* dst_ptr, int dst_width) {
  asm volatile (
    "pcmpeqb   %%xmm5,%%xmm5                   \n"
    "psrlw     $0x8,%%xmm5                     \n"
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqa    (%0),%%xmm0                     \n"
    "movdqa    0x10(%0),%%xmm1                 \n"
    "movdqa    (%0,%3,1),%%xmm2                \n"
    "movdqa    0x10(%0,%3,1),%%xmm3            \n"
    "lea       0x20(%0),%0                     \n"
    "pavgb     %%xmm2,%%xmm0                   \n"
    "pavgb     %%xmm3,%%xmm1                   \n"
    "movdqa    %%xmm0,%%xmm2                   \n"
    "psrlw     $0x8,%%xmm0                     \n"
    "movdqa    %%xmm1,%%xmm3                   \n"
    "psrlw     $0x8,%%xmm1                     \n"
    "pand      %%xmm5,%%xmm2                   \n"
    "pand      %%xmm5,%%xmm3                   \n"
    "pavgw     %%xmm2,%%xmm0                   \n"
    "pavgw     %%xmm3,%%xmm1                   \n"
    "packuswb  %%xmm1,%%xmm0                   \n"
    "movdqa    %%xmm0,(%1)                     \n"
    "lea       0x10(%1),%1                     \n"
    "sub       $0x10,%2                        \n"
    "jg        1b                              \n"
  : "+r"(src_ptr),    // %0
    "+r"(dst_ptr),    // %1
    "+r"(dst_width)   // %2
  : "r"(static_cast<intptr_t>(src_stride))   // %3
  : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
#endif
  );
}
static void ScaleRowDown2_Unaligned_SSE2(const uint8* src_ptr,
                                         ptrdiff_t src_stride,
                                         uint8* dst_ptr, int dst_width) {
  asm volatile (
    "pcmpeqb   %%xmm5,%%xmm5                   \n"
    "psrlw     $0x8,%%xmm5                     \n"
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqu    (%0),%%xmm0                     \n"
    "movdqu    0x10(%0),%%xmm1                 \n"
    "lea       0x20(%0),%0                     \n"
    "pand      %%xmm5,%%xmm0                   \n"
    "pand      %%xmm5,%%xmm1                   \n"
    "packuswb  %%xmm1,%%xmm0                   \n"
    "movdqu    %%xmm0,(%1)                     \n"
    "lea       0x10(%1),%1                     \n"
    "sub       $0x10,%2                        \n"
    "jg        1b                              \n"
  : "+r"(src_ptr),    // %0
    "+r"(dst_ptr),    // %1
    "+r"(dst_width)   // %2
  :
  : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm5"
#endif
  );
}

static void ScaleRowDown2Int_Unaligned_SSE2(const uint8* src_ptr,
                                            ptrdiff_t src_stride,
                                            uint8* dst_ptr, int dst_width) {
  asm volatile (
    "pcmpeqb   %%xmm5,%%xmm5                   \n"
    "psrlw     $0x8,%%xmm5                     \n"
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqu    (%0),%%xmm0                     \n"
    "movdqu    0x10(%0),%%xmm1                 \n"
    "movdqu    (%0,%3,1),%%xmm2                \n"
    "movdqu    0x10(%0,%3,1),%%xmm3            \n"
    "lea       0x20(%0),%0                     \n"
    "pavgb     %%xmm2,%%xmm0                   \n"
    "pavgb     %%xmm3,%%xmm1                   \n"
    "movdqa    %%xmm0,%%xmm2                   \n"
    "psrlw     $0x8,%%xmm0                     \n"
    "movdqa    %%xmm1,%%xmm3                   \n"
    "psrlw     $0x8,%%xmm1                     \n"
    "pand      %%xmm5,%%xmm2                   \n"
    "pand      %%xmm5,%%xmm3                   \n"
    "pavgw     %%xmm2,%%xmm0                   \n"
    "pavgw     %%xmm3,%%xmm1                   \n"
    "packuswb  %%xmm1,%%xmm0                   \n"
    "movdqu    %%xmm0,(%1)                     \n"
    "lea       0x10(%1),%1                     \n"
    "sub       $0x10,%2                        \n"
    "jg        1b                              \n"
  : "+r"(src_ptr),    // %0
    "+r"(dst_ptr),    // %1
    "+r"(dst_width)   // %2
  : "r"(static_cast<intptr_t>(src_stride))   // %3
  : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
#endif
  );
}

#define HAS_SCALEROWDOWN4_SSE2
static void ScaleRowDown4_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
                               uint8* dst_ptr, int dst_width) {
  asm volatile (
    "pcmpeqb   %%xmm5,%%xmm5                   \n"
    "psrld     $0x18,%%xmm5                    \n"
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqa    (%0),%%xmm0                     \n"
    "movdqa    0x10(%0),%%xmm1                 \n"
    "lea       0x20(%0),%0                     \n"
    "pand      %%xmm5,%%xmm0                   \n"
    "pand      %%xmm5,%%xmm1                   \n"
    "packuswb  %%xmm1,%%xmm0                   \n"
    "packuswb  %%xmm0,%%xmm0                   \n"
    "movq      %%xmm0,(%1)                     \n"
    "lea       0x8(%1),%1                      \n"
    "sub       $0x8,%2                         \n"
    "jg        1b                              \n"
  : "+r"(src_ptr),    // %0
    "+r"(dst_ptr),    // %1
    "+r"(dst_width)   // %2
  :
  : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm5"
#endif
  );
}

static void ScaleRowDown4Int_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
                                  uint8* dst_ptr, int dst_width) {
  intptr_t stridex3 = 0;
  asm volatile (
    "pcmpeqb   %%xmm7,%%xmm7                   \n"
    "psrlw     $0x8,%%xmm7                     \n"
    "lea       (%4,%4,2),%3                    \n"
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqa    (%0),%%xmm0                     \n"
    "movdqa    0x10(%0),%%xmm1                 \n"
    "movdqa    (%0,%4,1),%%xmm2                \n"
    "movdqa    0x10(%0,%4,1),%%xmm3            \n"
    "pavgb     %%xmm2,%%xmm0                   \n"
    "pavgb     %%xmm3,%%xmm1                   \n"
    "movdqa    (%0,%4,2),%%xmm2                \n"
    "movdqa    0x10(%0,%4,2),%%xmm3            \n"
    "movdqa    (%0,%3,1),%%xmm4                \n"
    "movdqa    0x10(%0,%3,1),%%xmm5            \n"
    "lea       0x20(%0),%0                     \n"
    "pavgb     %%xmm4,%%xmm2                   \n"
    "pavgb     %%xmm2,%%xmm0                   \n"
    "pavgb     %%xmm5,%%xmm3                   \n"
    "pavgb     %%xmm3,%%xmm1                   \n"
    "movdqa    %%xmm0,%%xmm2                   \n"
    "psrlw     $0x8,%%xmm0                     \n"
    "movdqa    %%xmm1,%%xmm3                   \n"
    "psrlw     $0x8,%%xmm1                     \n"
    "pand      %%xmm7,%%xmm2                   \n"
    "pand      %%xmm7,%%xmm3                   \n"
    "pavgw     %%xmm2,%%xmm0                   \n"
    "pavgw     %%xmm3,%%xmm1                   \n"
    "packuswb  %%xmm1,%%xmm0                   \n"
    "movdqa    %%xmm0,%%xmm2                   \n"
    "psrlw     $0x8,%%xmm0                     \n"
    "pand      %%xmm7,%%xmm2                   \n"
    "pavgw     %%xmm2,%%xmm0                   \n"
    "packuswb  %%xmm0,%%xmm0                   \n"
    "movq      %%xmm0,(%1)                     \n"
    "lea       0x8(%1),%1                      \n"
    "sub       $0x8,%2                         \n"
    "jg        1b                              \n"
  : "+r"(src_ptr),     // %0
    "+r"(dst_ptr),     // %1
    "+r"(dst_width),   // %2
    "+r"(stridex3)     // %3
  : "r"(static_cast<intptr_t>(src_stride))    // %4
  : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm7"
#endif
  );
}

#define HAS_SCALEROWDOWN8_SSE2
static void ScaleRowDown8_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
                               uint8* dst_ptr, int dst_width) {
  asm volatile (
    "pcmpeqb   %%xmm5,%%xmm5                   \n"
    "psrlq     $0x38,%%xmm5                    \n"
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqa    (%0),%%xmm0                     \n"
    "movdqa    0x10(%0),%%xmm1                 \n"
    "lea       0x20(%0),%0                     \n"
    "pand      %%xmm5,%%xmm0                   \n"
    "pand      %%xmm5,%%xmm1                   \n"
    "packuswb  %%xmm1,%%xmm0                   \n"
    "packuswb  %%xmm0,%%xmm0                   \n"
    "packuswb  %%xmm0,%%xmm0                   \n"
    "movd      %%xmm0,(%1)                     \n"
    "lea       0x4(%1),%1                      \n"
    "sub       $0x4,%2                         \n"
    "jg        1b                              \n"
  : "+r"(src_ptr),    // %0
    "+r"(dst_ptr),    // %1
    "+r"(dst_width)   // %2
  :
  : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm5"
#endif
  );
}

static void ScaleRowDown8Int_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
                                  uint8* dst_ptr, int dst_width) {
  intptr_t stridex3 = 0;
  intptr_t row4 = 0;
  asm volatile (
    "lea       (%5,%5,2),%3                    \n"
    "pxor      %%xmm7,%%xmm7                   \n"
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqa    (%0),%%xmm0                     \n"
    "movdqa    0x10(%0),%%xmm1                 \n"
    "movdqa    (%0,%5,1),%%xmm2                \n"
    "movdqa    0x10(%0,%5,1),%%xmm3            \n"
    "pavgb     %%xmm2,%%xmm0                   \n"
    "pavgb     %%xmm3,%%xmm1                   \n"
    "movdqa    (%0,%5,2),%%xmm2                \n"
    "movdqa    0x10(%0,%5,2),%%xmm3            \n"
    "movdqa    (%0,%3,1),%%xmm4                \n"
    "movdqa    0x10(%0,%3,1),%%xmm5            \n"
    "lea       (%0,%5,4),%4                    \n"
    "lea       0x20(%0),%0                     \n"
    "pavgb     %%xmm4,%%xmm2                   \n"
    "pavgb     %%xmm5,%%xmm3                   \n"
    "pavgb     %%xmm2,%%xmm0                   \n"
    "pavgb     %%xmm3,%%xmm1                   \n"
    "movdqa    0x0(%4),%%xmm2                  \n"
    "movdqa    0x10(%4),%%xmm3                 \n"
    "movdqa    0x0(%4,%5,1),%%xmm4             \n"
    "movdqa    0x10(%4,%5,1),%%xmm5            \n"
    "pavgb     %%xmm4,%%xmm2                   \n"
    "pavgb     %%xmm5,%%xmm3                   \n"
    "movdqa    0x0(%4,%5,2),%%xmm4             \n"
    "movdqa    0x10(%4,%5,2),%%xmm5            \n"
    "movdqa    0x0(%4,%3,1),%%xmm6             \n"
    "pavgb     %%xmm6,%%xmm4                   \n"
    "movdqa    0x10(%4,%3,1),%%xmm6            \n"
    "pavgb     %%xmm6,%%xmm5                   \n"
    "pavgb     %%xmm4,%%xmm2                   \n"
    "pavgb     %%xmm5,%%xmm3                   \n"
    "pavgb     %%xmm2,%%xmm0                   \n"
    "pavgb     %%xmm3,%%xmm1                   \n"
    "psadbw    %%xmm7,%%xmm0                   \n"
    "psadbw    %%xmm7,%%xmm1                   \n"
    "pshufd    $0xd8,%%xmm0,%%xmm0             \n"
    "pshufd    $0x8d,%%xmm1,%%xmm1             \n"
    "por       %%xmm1,%%xmm0                   \n"
    "psrlw     $0x3,%%xmm0                     \n"
    "packuswb  %%xmm0,%%xmm0                   \n"
    "packuswb  %%xmm0,%%xmm0                   \n"
    "movd      %%xmm0,(%1)                     \n"
    "lea       0x4(%1),%1                      \n"
    "sub       $0x4,%2                         \n"
    "jg        1b                              \n"
  : "+r"(src_ptr),     // %0
    "+r"(dst_ptr),     // %1
    "+rm"(dst_width),  // %2
    "+r"(stridex3),    // %3
    "+r"(row4)         // %4
  : "r"(static_cast<intptr_t>(src_stride))  // %5
  : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
#endif
  );
}

#define HAS_SCALEROWDOWN34_SSSE3
static void ScaleRowDown34_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride,
                                 uint8* dst_ptr, int dst_width) {
  asm volatile (
    "movdqa    %0,%%xmm3                       \n"
    "movdqa    %1,%%xmm4                       \n"
    "movdqa    %2,%%xmm5                       \n"
  :
  : "m"(kShuf0),  // %0
    "m"(kShuf1),  // %1
    "m"(kShuf2)   // %2
  );
  asm volatile (
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqa    (%0),%%xmm0                     \n"
    "movdqa    0x10(%0),%%xmm2                 \n"
    "lea       0x20(%0),%0                     \n"
    "movdqa    %%xmm2,%%xmm1                   \n"
    "palignr   $0x8,%%xmm0,%%xmm1              \n"
    "pshufb    %%xmm3,%%xmm0                   \n"
    "pshufb    %%xmm4,%%xmm1                   \n"
    "pshufb    %%xmm5,%%xmm2                   \n"
    "movq      %%xmm0,(%1)                     \n"
    "movq      %%xmm1,0x8(%1)                  \n"
    "movq      %%xmm2,0x10(%1)                 \n"
    "lea       0x18(%1),%1                     \n"
    "sub       $0x18,%2                        \n"
    "jg        1b                              \n"
  : "+r"(src_ptr),   // %0
    "+r"(dst_ptr),   // %1
    "+r"(dst_width)  // %2
  :
  : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
#endif
  );
}

static void ScaleRowDown34_1_Int_SSSE3(const uint8* src_ptr,
                                       ptrdiff_t src_stride,
                                       uint8* dst_ptr, int dst_width) {
  asm volatile (
    "movdqa    %0,%%xmm2                       \n"  // kShuf01
    "movdqa    %1,%%xmm3                       \n"  // kShuf11
    "movdqa    %2,%%xmm4                       \n"  // kShuf21
  :
  : "m"(kShuf01),  // %0
    "m"(kShuf11),  // %1
    "m"(kShuf21)   // %2
  );
  asm volatile (
    "movdqa    %0,%%xmm5                       \n"  // kMadd01
    "movdqa    %1,%%xmm0                       \n"  // kMadd11
    "movdqa    %2,%%xmm1                       \n"  // kRound34
  :
  : "m"(kMadd01),  // %0
    "m"(kMadd11),  // %1
    "m"(kRound34)  // %2
  );
  asm volatile (
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqa    (%0),%%xmm6                     \n"
    "movdqa    (%0,%3),%%xmm7                  \n"
    "pavgb     %%xmm7,%%xmm6                   \n"
    "pshufb    %%xmm2,%%xmm6                   \n"
    "pmaddubsw %%xmm5,%%xmm6                   \n"
    "paddsw    %%xmm1,%%xmm6                   \n"
    "psrlw     $0x2,%%xmm6                     \n"
    "packuswb  %%xmm6,%%xmm6                   \n"
    "movq      %%xmm6,(%1)                     \n"
    "movdqu    0x8(%0),%%xmm6                  \n"
    "movdqu    0x8(%0,%3),%%xmm7               \n"
    "pavgb     %%xmm7,%%xmm6                   \n"
    "pshufb    %%xmm3,%%xmm6                   \n"
    "pmaddubsw %%xmm0,%%xmm6                   \n"
    "paddsw    %%xmm1,%%xmm6                   \n"
    "psrlw     $0x2,%%xmm6                     \n"
    "packuswb  %%xmm6,%%xmm6                   \n"
    "movq      %%xmm6,0x8(%1)                  \n"
    "movdqa    0x10(%0),%%xmm6                 \n"
    "movdqa    0x10(%0,%3),%%xmm7              \n"
    "lea       0x20(%0),%0                     \n"
    "pavgb     %%xmm7,%%xmm6                   \n"
    "pshufb    %%xmm4,%%xmm6                   \n"
    "pmaddubsw %4,%%xmm6                       \n"
    "paddsw    %%xmm1,%%xmm6                   \n"
    "psrlw     $0x2,%%xmm6                     \n"
    "packuswb  %%xmm6,%%xmm6                   \n"
    "movq      %%xmm6,0x10(%1)                 \n"
    "lea       0x18(%1),%1                     \n"
    "sub       $0x18,%2                        \n"
    "jg        1b                              \n"
  : "+r"(src_ptr),   // %0
    "+r"(dst_ptr),   // %1
    "+r"(dst_width)  // %2
  : "r"(static_cast<intptr_t>(src_stride)),  // %3
    "m"(kMadd21)     // %4
  : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
#endif
  );
}

static void ScaleRowDown34_0_Int_SSSE3(const uint8* src_ptr,
                                       ptrdiff_t src_stride,
                                       uint8* dst_ptr, int dst_width) {
  asm volatile (
    "movdqa    %0,%%xmm2                       \n"  // kShuf01
    "movdqa    %1,%%xmm3                       \n"  // kShuf11
    "movdqa    %2,%%xmm4                       \n"  // kShuf21
  :
  : "m"(kShuf01),  // %0
    "m"(kShuf11),  // %1
    "m"(kShuf21)   // %2
  );
  asm volatile (
    "movdqa    %0,%%xmm5                       \n"  // kMadd01
    "movdqa    %1,%%xmm0                       \n"  // kMadd11
    "movdqa    %2,%%xmm1                       \n"  // kRound34
  :
  : "m"(kMadd01),  // %0
    "m"(kMadd11),  // %1
    "m"(kRound34)  // %2
  );

  asm volatile (
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqa    (%0),%%xmm6                     \n"
    "movdqa    (%0,%3,1),%%xmm7                \n"
    "pavgb     %%xmm6,%%xmm7                   \n"
    "pavgb     %%xmm7,%%xmm6                   \n"
    "pshufb    %%xmm2,%%xmm6                   \n"
    "pmaddubsw %%xmm5,%%xmm6                   \n"
    "paddsw    %%xmm1,%%xmm6                   \n"
    "psrlw     $0x2,%%xmm6                     \n"
    "packuswb  %%xmm6,%%xmm6                   \n"
    "movq      %%xmm6,(%1)                     \n"
    "movdqu    0x8(%0),%%xmm6                  \n"
    "movdqu    0x8(%0,%3,1),%%xmm7             \n"
    "pavgb     %%xmm6,%%xmm7                   \n"
    "pavgb     %%xmm7,%%xmm6                   \n"
    "pshufb    %%xmm3,%%xmm6                   \n"
    "pmaddubsw %%xmm0,%%xmm6                   \n"
    "paddsw    %%xmm1,%%xmm6                   \n"
    "psrlw     $0x2,%%xmm6                     \n"
    "packuswb  %%xmm6,%%xmm6                   \n"
    "movq      %%xmm6,0x8(%1)                  \n"
    "movdqa    0x10(%0),%%xmm6                 \n"
    "movdqa    0x10(%0,%3,1),%%xmm7            \n"
    "lea       0x20(%0),%0                     \n"
    "pavgb     %%xmm6,%%xmm7                   \n"
    "pavgb     %%xmm7,%%xmm6                   \n"
    "pshufb    %%xmm4,%%xmm6                   \n"
    "pmaddubsw %4,%%xmm6                       \n"
    "paddsw    %%xmm1,%%xmm6                   \n"
    "psrlw     $0x2,%%xmm6                     \n"
    "packuswb  %%xmm6,%%xmm6                   \n"
    "movq      %%xmm6,0x10(%1)                 \n"
    "lea       0x18(%1),%1                     \n"
    "sub       $0x18,%2                        \n"
    "jg        1b                              \n"
    : "+r"(src_ptr),   // %0
      "+r"(dst_ptr),   // %1
      "+r"(dst_width)  // %2
    : "r"(static_cast<intptr_t>(src_stride)),  // %3
      "m"(kMadd21)     // %4
    : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
#endif
  );
}

#define HAS_SCALEROWDOWN38_SSSE3
static void ScaleRowDown38_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride,
                                 uint8* dst_ptr, int dst_width) {
  asm volatile (
    "movdqa    %3,%%xmm4                       \n"
    "movdqa    %4,%%xmm5                       \n"
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqa    (%0),%%xmm0                     \n"
    "movdqa    0x10(%0),%%xmm1                 \n"
    "lea       0x20(%0),%0                     \n"
    "pshufb    %%xmm4,%%xmm0                   \n"
    "pshufb    %%xmm5,%%xmm1                   \n"
    "paddusb   %%xmm1,%%xmm0                   \n"
    "movq      %%xmm0,(%1)                     \n"
    "movhlps   %%xmm0,%%xmm1                   \n"
    "movd      %%xmm1,0x8(%1)                  \n"
    "lea       0xc(%1),%1                      \n"
    "sub       $0xc,%2                         \n"
    "jg        1b                              \n"
  : "+r"(src_ptr),   // %0
    "+r"(dst_ptr),   // %1
    "+r"(dst_width)  // %2
  : "m"(kShuf38a),   // %3
    "m"(kShuf38b)    // %4
  : "memory", "cc"
#if defined(__SSE2__)
      , "xmm0", "xmm1", "xmm4", "xmm5"
#endif
  );
}

static void ScaleRowDown38_2_Int_SSSE3(const uint8* src_ptr,
                                       ptrdiff_t src_stride,
                                       uint8* dst_ptr, int dst_width) {
  asm volatile (
    "movdqa    %0,%%xmm2                       \n"
    "movdqa    %1,%%xmm3                       \n"
    "movdqa    %2,%%xmm4                       \n"
    "movdqa    %3,%%xmm5                       \n"
  :
  : "m"(kShufAb0),   // %0
    "m"(kShufAb1),   // %1
    "m"(kShufAb2),   // %2
    "m"(kScaleAb2)   // %3
  );
  asm volatile (
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqa    (%0),%%xmm0                     \n"
    "pavgb     (%0,%3,1),%%xmm0                \n"
    "lea       0x10(%0),%0                     \n"
    "movdqa    %%xmm0,%%xmm1                   \n"
    "pshufb    %%xmm2,%%xmm1                   \n"
    "movdqa    %%xmm0,%%xmm6                   \n"
    "pshufb    %%xmm3,%%xmm6                   \n"
    "paddusw   %%xmm6,%%xmm1                   \n"
    "pshufb    %%xmm4,%%xmm0                   \n"
    "paddusw   %%xmm0,%%xmm1                   \n"
    "pmulhuw   %%xmm5,%%xmm1                   \n"
    "packuswb  %%xmm1,%%xmm1                   \n"
    "sub       $0x6,%2                         \n"
    "movd      %%xmm1,(%1)                     \n"
    "psrlq     $0x10,%%xmm1                    \n"
    "movd      %%xmm1,0x2(%1)                  \n"
    "lea       0x6(%1),%1                      \n"
    "jg        1b                              \n"
  : "+r"(src_ptr),     // %0
    "+r"(dst_ptr),     // %1
    "+r"(dst_width)    // %2
  : "r"(static_cast<intptr_t>(src_stride))  // %3
  : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6"
#endif
  );
}

static void ScaleRowDown38_3_Int_SSSE3(const uint8* src_ptr,
                                       ptrdiff_t src_stride,
                                       uint8* dst_ptr, int dst_width) {
  asm volatile (
    "movdqa    %0,%%xmm2                       \n"
    "movdqa    %1,%%xmm3                       \n"
    "movdqa    %2,%%xmm4                       \n"
    "pxor      %%xmm5,%%xmm5                   \n"
  :
  : "m"(kShufAc),    // %0
    "m"(kShufAc3),   // %1
    "m"(kScaleAc33)  // %2
  );
  asm volatile (
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqa    (%0),%%xmm0                     \n"
    "movdqa    (%0,%3,1),%%xmm6                \n"
    "movhlps   %%xmm0,%%xmm1                   \n"
    "movhlps   %%xmm6,%%xmm7                   \n"
    "punpcklbw %%xmm5,%%xmm0                   \n"
    "punpcklbw %%xmm5,%%xmm1                   \n"
    "punpcklbw %%xmm5,%%xmm6                   \n"
    "punpcklbw %%xmm5,%%xmm7                   \n"
    "paddusw   %%xmm6,%%xmm0                   \n"
    "paddusw   %%xmm7,%%xmm1                   \n"
    "movdqa    (%0,%3,2),%%xmm6                \n"
    "lea       0x10(%0),%0                     \n"
    "movhlps   %%xmm6,%%xmm7                   \n"
    "punpcklbw %%xmm5,%%xmm6                   \n"
    "punpcklbw %%xmm5,%%xmm7                   \n"
    "paddusw   %%xmm6,%%xmm0                   \n"
    "paddusw   %%xmm7,%%xmm1                   \n"
    "movdqa    %%xmm0,%%xmm6                   \n"
    "psrldq    $0x2,%%xmm0                     \n"
    "paddusw   %%xmm0,%%xmm6                   \n"
    "psrldq    $0x2,%%xmm0                     \n"
    "paddusw   %%xmm0,%%xmm6                   \n"
    "pshufb    %%xmm2,%%xmm6                   \n"
    "movdqa    %%xmm1,%%xmm7                   \n"
    "psrldq    $0x2,%%xmm1                     \n"
    "paddusw   %%xmm1,%%xmm7                   \n"
    "psrldq    $0x2,%%xmm1                     \n"
    "paddusw   %%xmm1,%%xmm7                   \n"
    "pshufb    %%xmm3,%%xmm7                   \n"
    "paddusw   %%xmm7,%%xmm6                   \n"
    "pmulhuw   %%xmm4,%%xmm6                   \n"
    "packuswb  %%xmm6,%%xmm6                   \n"
    "sub       $0x6,%2                         \n"
    "movd      %%xmm6,(%1)                     \n"
    "psrlq     $0x10,%%xmm6                    \n"
    "movd      %%xmm6,0x2(%1)                  \n"
    "lea       0x6(%1),%1                      \n"
    "jg        1b                              \n"
  : "+r"(src_ptr),    // %0
    "+r"(dst_ptr),    // %1
    "+r"(dst_width)   // %2
  : "r"(static_cast<intptr_t>(src_stride))   // %3
  : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
#endif
  );
}

#define HAS_SCALEADDROWS_SSE2
static void ScaleAddRows_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
                              uint16* dst_ptr, int src_width, int src_height) {
  int tmp_height = 0;
  intptr_t tmp_src = 0;
  asm volatile (
    "pxor      %%xmm4,%%xmm4                   \n"
    "sub       $0x1,%5                         \n"
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqa    (%0),%%xmm0                     \n"
    "mov       %0,%3                           \n"
    "add       %6,%0                           \n"
    "movdqa    %%xmm0,%%xmm1                   \n"
    "punpcklbw %%xmm4,%%xmm0                   \n"
    "punpckhbw %%xmm4,%%xmm1                   \n"
    "mov       %5,%2                           \n"
    "test      %2,%2                           \n"
    "je        3f                              \n"
  "2:                                          \n"
    "movdqa    (%0),%%xmm2                     \n"
    "add       %6,%0                           \n"
    "movdqa    %%xmm2,%%xmm3                   \n"
    "punpcklbw %%xmm4,%%xmm2                   \n"
    "punpckhbw %%xmm4,%%xmm3                   \n"
    "paddusw   %%xmm2,%%xmm0                   \n"
    "paddusw   %%xmm3,%%xmm1                   \n"
    "sub       $0x1,%2                         \n"
    "jg        2b                              \n"
  "3:                                          \n"
    "movdqa    %%xmm0,(%1)                     \n"
    "movdqa    %%xmm1,0x10(%1)                 \n"
    "lea       0x10(%3),%0                     \n"
    "lea       0x20(%1),%1                     \n"
    "sub       $0x10,%4                        \n"
    "jg        1b                              \n"
  : "+r"(src_ptr),     // %0
    "+r"(dst_ptr),     // %1
    "+r"(tmp_height),  // %2
    "+r"(tmp_src),     // %3
    "+r"(src_width),   // %4
    "+rm"(src_height)  // %5
  : "rm"(static_cast<intptr_t>(src_stride))  // %6
  : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4"
#endif
  );
}

#ifndef SSE2_DISABLED
// Bilinear row filtering combines 16x2 -> 16x1. SSE2 version
#define HAS_SCALEFILTERROWS_SSE2_DISABLED
static void ScaleFilterRows_SSE2(uint8* dst_ptr,
                                 const uint8* src_ptr, ptrdiff_t src_stride,
                                 int dst_width, int source_y_fraction) {
  asm volatile (
    "sub       %1,%0                           \n"
    "cmp       $0x0,%3                         \n"
    "je        2f                              \n"
    "cmp       $0x80,%3                        \n"
    "je        3f                              \n"
    "movd      %3,%%xmm5                       \n"
    "punpcklbw %%xmm5,%%xmm5                   \n"
    "punpcklwd %%xmm5,%%xmm5                   \n"
    "pshufd    $0x0,%%xmm5,%%xmm5              \n"
    "pxor      %%xmm4,%%xmm4                   \n"
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqa    (%1),%%xmm0                     \n"
    "movdqa    (%1,%4,1),%%xmm2                \n"
    "movdqa    %%xmm0,%%xmm1                   \n"
    "movdqa    %%xmm2,%%xmm3                   \n"
    "punpcklbw %%xmm4,%%xmm2                   \n"
    "punpckhbw %%xmm4,%%xmm3                   \n"
    "punpcklbw %%xmm4,%%xmm0                   \n"
    "punpckhbw %%xmm4,%%xmm1                   \n"
    "psubw     %%xmm0,%%xmm2                   \n"
    "psubw     %%xmm1,%%xmm3                   \n"
    "pmulhw    %%xmm5,%%xmm2                   \n"
    "pmulhw    %%xmm5,%%xmm3                   \n"
    "paddw     %%xmm2,%%xmm0                   \n"
    "paddw     %%xmm3,%%xmm1                   \n"
    "packuswb  %%xmm1,%%xmm0                   \n"
    "sub       $0x10,%2                        \n"
    "movdqa    %%xmm0,(%1,%0,1)                \n"
    "lea       0x10(%1),%1                     \n"
    "jg        1b                              \n"
    "jmp       4f                              \n"
    ".p2align  4                               \n"
  "2:                                          \n"
    "movdqa    (%1),%%xmm0                     \n"
    "sub       $0x10,%2                        \n"
    "movdqa    %%xmm0,(%1,%0,1)                \n"
    "lea       0x10(%1),%1                     \n"
    "jg        2b                              \n"
    "jmp       4f                              \n"
    ".p2align  4                               \n"
  "3:                                          \n"
    "movdqa    (%1),%%xmm0                     \n"
    "pavgb     (%1,%4,1),%%xmm0                \n"
    "sub       $0x10,%2                        \n"
    "movdqa    %%xmm0,(%1,%0,1)                \n"
    "lea       0x10(%1),%1                     \n"
    "jg        3b                              \n"
    ".p2align  4                               \n"
  "4:                                          \n"
    "punpckhbw %%xmm0,%%xmm0                   \n"
    "pshufhw   $0xff,%%xmm0,%%xmm0             \n"
    "punpckhqdq %%xmm0,%%xmm0                  \n"
    "movdqa    %%xmm0,(%1,%0,1)                \n"
  : "+r"(dst_ptr),    // %0
    "+r"(src_ptr),    // %1
    "+r"(dst_width),  // %2
    "+r"(source_y_fraction)  // %3
  : "r"(static_cast<intptr_t>(src_stride))  // %4
  : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
#endif
  );
}
#endif  // SSE2_DISABLED

// Bilinear row filtering combines 16x2 -> 16x1. SSSE3 version
#define HAS_SCALEFILTERROWS_SSSE3
static void ScaleFilterRows_SSSE3(uint8* dst_ptr,
                                  const uint8* src_ptr, ptrdiff_t src_stride,
                                  int dst_width, int source_y_fraction) {
  asm volatile (
    "sub       %1,%0                           \n"
    "shr       %3                              \n"
    "cmp       $0x0,%3                         \n"
    "je        2f                              \n"
    "cmp       $0x40,%3                        \n"
    "je        3f                              \n"
    "movd      %3,%%xmm0                       \n"
    "neg       %3                              \n"
    "add       $0x80,%3                        \n"
    "movd      %3,%%xmm5                       \n"
    "punpcklbw %%xmm0,%%xmm5                   \n"
    "punpcklwd %%xmm5,%%xmm5                   \n"
    "pshufd    $0x0,%%xmm5,%%xmm5              \n"
    ".p2align  4                               \n"
  "1:                                          \n"
    "movdqa    (%1),%%xmm0                     \n"
    "movdqa    (%1,%4,1),%%xmm2                \n"
    "movdqa    %%xmm0,%%xmm1                   \n"
    "punpcklbw %%xmm2,%%xmm0                   \n"
    "punpckhbw %%xmm2,%%xmm1                   \n"
    "pmaddubsw %%xmm5,%%xmm0                   \n"
    "pmaddubsw %%xmm5,%%xmm1                   \n"
    "psrlw     $0x7,%%xmm0                     \n"
    "psrlw     $0x7,%%xmm1                     \n"
    "packuswb  %%xmm1,%%xmm0                   \n"
    "sub       $0x10,%2                        \n"
    "movdqa    %%xmm0,(%1,%0,1)                \n"
    "lea       0x10(%1),%1                     \n"
    "jg        1b                              \n"
    "jmp       4f                              \n"
    ".p2align  4                               \n"
  "2:                                          \n"
    "movdqa    (%1),%%xmm0                     \n"
    "sub       $0x10,%2                        \n"
    "movdqa    %%xmm0,(%1,%0,1)                \n"
    "lea       0x10(%1),%1                     \n"
    "jg        2b                              \n"
    "jmp       4f                              \n"
    ".p2align  4                               \n"
  "3:                                          \n"
    "movdqa    (%1),%%xmm0                     \n"
    "pavgb     (%1,%4,1),%%xmm0                \n"
    "sub       $0x10,%2                        \n"
    "movdqa    %%xmm0,(%1,%0,1)                \n"
    "lea       0x10(%1),%1                     \n"
    "jg        3b                              \n"
    ".p2align  4                               \n"
  "4:                                          \n"
    "punpckhbw %%xmm0,%%xmm0                   \n"
    "pshufhw   $0xff,%%xmm0,%%xmm0             \n"
    "punpckhqdq %%xmm0,%%xmm0                  \n"
    "movdqa    %%xmm0,(%1,%0,1)                \n"
  : "+r"(dst_ptr),    // %0
    "+r"(src_ptr),    // %1
    "+r"(dst_width),  // %2
    "+r"(source_y_fraction)  // %3
  : "r"(static_cast<intptr_t>(src_stride))  // %4
  : "memory", "cc"
#if defined(__SSE2__)
    , "xmm0", "xmm1", "xmm2", "xmm5"
#endif
  );
}
#endif  // defined(__x86_64__) || defined(__i386__)

// CPU agnostic row functions
static void ScaleRowDown2_C(const uint8* src_ptr, ptrdiff_t /* src_stride */,
                            uint8* dst, int dst_width) {
  uint8* dend = dst + dst_width - 1;
  do {
    dst[0] = src_ptr[0];
    dst[1] = src_ptr[2];
    dst += 2;
    src_ptr += 4;
  } while (dst < dend);
  if (dst_width & 1) {
    dst[0] = src_ptr[0];
  }
}

void ScaleRowDown2Int_C(const uint8* src_ptr, ptrdiff_t src_stride,
                        uint8* dst, int dst_width) {
  const uint8* s = src_ptr;
  const uint8* t = src_ptr + src_stride;
  uint8* dend = dst + dst_width - 1;
  do {
    dst[0] = (s[0] + s[1] + t[0] + t[1] + 2) >> 2;
    dst[1] = (s[2] + s[3] + t[2] + t[3] + 2) >> 2;
    dst += 2;
    s += 4;
    t += 4;
  } while (dst < dend);
  if (dst_width & 1) {
    dst[0] = (s[0] + s[1] + t[0] + t[1] + 2) >> 2;
  }
}

static void ScaleRowDown4_C(const uint8* src_ptr, ptrdiff_t /* src_stride */,
                            uint8* dst, int dst_width) {
  uint8* dend = dst + dst_width - 1;
  do {
    dst[0] = src_ptr[0];
    dst[1] = src_ptr[4];
    dst += 2;
    src_ptr += 8;
  } while (dst < dend);
  if (dst_width & 1) {
    dst[0] = src_ptr[0];
  }
}

static void ScaleRowDown4Int_C(const uint8* src_ptr, ptrdiff_t src_stride,
                               uint8* dst, int dst_width) {
  intptr_t stride = src_stride;
  uint8* dend = dst + dst_width - 1;
  do {
    dst[0] = (src_ptr[0] + src_ptr[1] + src_ptr[2] + src_ptr[3] +
             src_ptr[stride + 0] + src_ptr[stride + 1] +
             src_ptr[stride + 2] + src_ptr[stride + 3] +
             src_ptr[stride * 2 + 0] + src_ptr[stride * 2 + 1] +
             src_ptr[stride * 2 + 2] + src_ptr[stride * 2 + 3] +
             src_ptr[stride * 3 + 0] + src_ptr[stride * 3 + 1] +
             src_ptr[stride * 3 + 2] + src_ptr[stride * 3 + 3] +
             8) >> 4;
    dst[1] = (src_ptr[4] + src_ptr[5] + src_ptr[6] + src_ptr[7] +
             src_ptr[stride + 4] + src_ptr[stride + 5] +
             src_ptr[stride + 6] + src_ptr[stride + 7] +
             src_ptr[stride * 2 + 4] + src_ptr[stride * 2 + 5] +
             src_ptr[stride * 2 + 6] + src_ptr[stride * 2 + 7] +
             src_ptr[stride * 3 + 4] + src_ptr[stride * 3 + 5] +
             src_ptr[stride * 3 + 6] + src_ptr[stride * 3 + 7] +
             8) >> 4;
    dst += 2;
    src_ptr += 8;
  } while (dst < dend);
  if (dst_width & 1) {
    dst[0] = (src_ptr[0] + src_ptr[1] + src_ptr[2] + src_ptr[3] +
             src_ptr[stride + 0] + src_ptr[stride + 1] +
             src_ptr[stride + 2] + src_ptr[stride + 3] +
             src_ptr[stride * 2 + 0] + src_ptr[stride * 2 + 1] +
             src_ptr[stride * 2 + 2] + src_ptr[stride * 2 + 3] +
             src_ptr[stride * 3 + 0] + src_ptr[stride * 3 + 1] +
             src_ptr[stride * 3 + 2] + src_ptr[stride * 3 + 3] +
             8) >> 4;
  }
}

// 640 output pixels is enough to allow 5120 input pixels with 1/8 scale down.
// Keeping the total buffer under 4096 bytes avoids a stackcheck, saving 4% cpu.
static const int kMaxOutputWidth = 640;
static const int kMaxRow12 = kMaxOutputWidth * 2;

static void ScaleRowDown8_C(const uint8* src_ptr, ptrdiff_t /* src_stride */,
                            uint8* dst, int dst_width) {
  uint8* dend = dst + dst_width - 1;
  do {
    dst[0] = src_ptr[0];
    dst[1] = src_ptr[8];
    dst += 2;
    src_ptr += 16;
  } while (dst < dend);
  if (dst_width & 1) {
    dst[0] = src_ptr[0];
  }
}

// Note calling code checks width is less than max and if not
// uses ScaleRowDown8_C instead.
static void ScaleRowDown8Int_C(const uint8* src_ptr, ptrdiff_t src_stride,
                               uint8* dst, int dst_width) {
  SIMD_ALIGNED(uint8 src_row[kMaxRow12 * 2]);
  assert(dst_width <= kMaxOutputWidth);
  ScaleRowDown4Int_C(src_ptr, src_stride, src_row, dst_width * 2);
  ScaleRowDown4Int_C(src_ptr + src_stride * 4, src_stride,
                     src_row + kMaxOutputWidth,
                     dst_width * 2);
  ScaleRowDown2Int_C(src_row, kMaxOutputWidth, dst, dst_width);
}

static void ScaleRowDown34_C(const uint8* src_ptr, ptrdiff_t /* src_stride */,
                             uint8* dst, int dst_width) {
  assert((dst_width % 3 == 0) && (dst_width > 0));
  uint8* dend = dst + dst_width;
  do {
    dst[0] = src_ptr[0];
    dst[1] = src_ptr[1];
    dst[2] = src_ptr[3];
    dst += 3;
    src_ptr += 4;
  } while (dst < dend);
}

// Filter rows 0 and 1 together, 3 : 1
static void ScaleRowDown34_0_Int_C(const uint8* src_ptr, ptrdiff_t src_stride,
                                   uint8* d, int dst_width) {
  assert((dst_width % 3 == 0) && (dst_width > 0));
  const uint8* s = src_ptr;
  const uint8* t = src_ptr + src_stride;
  uint8* dend = d + dst_width;
  do {
    uint8 a0 = (s[0] * 3 + s[1] * 1 + 2) >> 2;
    uint8 a1 = (s[1] * 1 + s[2] * 1 + 1) >> 1;
    uint8 a2 = (s[2] * 1 + s[3] * 3 + 2) >> 2;
    uint8 b0 = (t[0] * 3 + t[1] * 1 + 2) >> 2;
    uint8 b1 = (t[1] * 1 + t[2] * 1 + 1) >> 1;
    uint8 b2 = (t[2] * 1 + t[3] * 3 + 2) >> 2;
    d[0] = (a0 * 3 + b0 + 2) >> 2;
    d[1] = (a1 * 3 + b1 + 2) >> 2;
    d[2] = (a2 * 3 + b2 + 2) >> 2;
    d += 3;
    s += 4;
    t += 4;
  } while (d < dend);
}

// Filter rows 1 and 2 together, 1 : 1
static void ScaleRowDown34_1_Int_C(const uint8* src_ptr, ptrdiff_t src_stride,
                                   uint8* d, int dst_width) {
  assert((dst_width % 3 == 0) && (dst_width > 0));
  const uint8* s = src_ptr;
  const uint8* t = src_ptr + src_stride;
  uint8* dend = d + dst_width;
  do {
    uint8 a0 = (s[0] * 3 + s[1] * 1 + 2) >> 2;
    uint8 a1 = (s[1] * 1 + s[2] * 1 + 1) >> 1;
    uint8 a2 = (s[2] * 1 + s[3] * 3 + 2) >> 2;
    uint8 b0 = (t[0] * 3 + t[1] * 1 + 2) >> 2;
    uint8 b1 = (t[1] * 1 + t[2] * 1 + 1) >> 1;
    uint8 b2 = (t[2] * 1 + t[3] * 3 + 2) >> 2;
    d[0] = (a0 + b0 + 1) >> 1;
    d[1] = (a1 + b1 + 1) >> 1;
    d[2] = (a2 + b2 + 1) >> 1;
    d += 3;
    s += 4;
    t += 4;
  } while (d < dend);
}

// (1-f)a + fb can be replaced with a + f(b-a)
#define BLENDER(a, b, f) (static_cast<int>(a) + \
    ((f) * (static_cast<int>(b) - static_cast<int>(a)) >> 16))

static void ScaleFilterCols_C(uint8* dst_ptr, const uint8* src_ptr,
                              int dst_width, int x, int dx) {
  for (int j = 0; j < dst_width - 1; j += 2) {
    int xi = x >> 16;
    int a = src_ptr[xi];
    int b = src_ptr[xi + 1];
    dst_ptr[0] = BLENDER(a, b, x & 0xffff);
    x += dx;
    xi = x >> 16;
    a = src_ptr[xi];
    b = src_ptr[xi + 1];
    dst_ptr[1] = BLENDER(a, b, x & 0xffff);
    x += dx;
    dst_ptr += 2;
  }
  if (dst_width & 1) {
    int xi = x >> 16;
    int a = src_ptr[xi];
    int b = src_ptr[xi + 1];
    dst_ptr[0] = BLENDER(a, b, x & 0xffff);
  }
}

static const int kMaxInputWidth = 2560;

#if defined(HAS_SCALEFILTERROWS_SSE2)
// Filter row to 3/4
static void ScaleFilterCols34_C(uint8* dst_ptr, const uint8* src_ptr,
                                int dst_width) {
  assert((dst_width % 3 == 0) && (dst_width > 0));
  const uint8* s = src_ptr;
  uint8* dend = dst_ptr + dst_width;
  do {
    dst_ptr[0] = (s[0] * 3 + s[1] * 1 + 2) >> 2;
    dst_ptr[1] = (s[1] * 1 + s[2] * 1 + 1) >> 1;
    dst_ptr[2] = (s[2] * 1 + s[3] * 3 + 2) >> 2;
    dst_ptr += 3;
    s += 4;
  } while (dst_ptr < dend);
}

#define HAS_SCALEROWDOWN34_SSE2_DISABLED
// Filter rows 0 and 1 together, 3 : 1
static void ScaleRowDown34_0_Int_SSE2(const uint8* src_ptr,
                                      ptrdiff_t src_stride,
                                      uint8* dst_ptr, int dst_width) {
  assert((dst_width % 3 == 0) && (dst_width > 0));
  SIMD_ALIGNED(uint8 row[kMaxInputWidth]);
  ScaleFilterRows_SSE2(row, src_ptr, src_stride, dst_width * 4 / 3, 256 / 4);
  ScaleFilterCols34_C(dst_ptr, row, dst_width);
}

// Filter rows 1 and 2 together, 1 : 1
static void ScaleRowDown34_1_Int_SSE2(const uint8* src_ptr,
                                      ptrdiff_t src_stride,
                                      uint8* dst_ptr, int dst_width) {
  assert((dst_width % 3 == 0) && (dst_width > 0));
  SIMD_ALIGNED(uint8 row[kMaxInputWidth]);
  ScaleFilterRows_SSE2(row, src_ptr, src_stride, dst_width * 4 / 3, 256 / 2);
  ScaleFilterCols34_C(dst_ptr, row, dst_width);
}
#endif

static void ScaleRowDown38_C(const uint8* src_ptr, ptrdiff_t /* src_stride */,
                             uint8* dst, int dst_width) {
  assert(dst_width % 3 == 0);
  for (int x = 0; x < dst_width; x += 3) {
    dst[0] = src_ptr[0];
    dst[1] = src_ptr[3];
    dst[2] = src_ptr[6];
    dst += 3;
    src_ptr += 8;
  }
}

// 8x3 -> 3x1
static void ScaleRowDown38_3_Int_C(const uint8* src_ptr,
                                   ptrdiff_t src_stride,
                                   uint8* dst_ptr, int dst_width) {
  assert((dst_width % 3 == 0) && (dst_width > 0));
  intptr_t stride = src_stride;
  for (int i = 0; i < dst_width; i += 3) {
    dst_ptr[0] = (src_ptr[0] + src_ptr[1] + src_ptr[2] +
        src_ptr[stride + 0] + src_ptr[stride + 1] +
        src_ptr[stride + 2] + src_ptr[stride * 2 + 0] +
        src_ptr[stride * 2 + 1] + src_ptr[stride * 2 + 2]) *
        (65536 / 9) >> 16;
    dst_ptr[1] = (src_ptr[3] + src_ptr[4] + src_ptr[5] +
        src_ptr[stride + 3] + src_ptr[stride + 4] +
        src_ptr[stride + 5] + src_ptr[stride * 2 + 3] +
        src_ptr[stride * 2 + 4] + src_ptr[stride * 2 + 5]) *
        (65536 / 9) >> 16;
    dst_ptr[2] = (src_ptr[6] + src_ptr[7] +
        src_ptr[stride + 6] + src_ptr[stride + 7] +
        src_ptr[stride * 2 + 6] + src_ptr[stride * 2 + 7]) *
        (65536 / 6) >> 16;
    src_ptr += 8;
    dst_ptr += 3;
  }
}

// 8x2 -> 3x1
static void ScaleRowDown38_2_Int_C(const uint8* src_ptr, ptrdiff_t src_stride,
                                   uint8* dst_ptr, int dst_width) {
  assert((dst_width % 3 == 0) && (dst_width > 0));
  intptr_t stride = src_stride;
  for (int i = 0; i < dst_width; i += 3) {
    dst_ptr[0] = (src_ptr[0] + src_ptr[1] + src_ptr[2] +
        src_ptr[stride + 0] + src_ptr[stride + 1] +
        src_ptr[stride + 2]) * (65536 / 6) >> 16;
    dst_ptr[1] = (src_ptr[3] + src_ptr[4] + src_ptr[5] +
        src_ptr[stride + 3] + src_ptr[stride + 4] +
        src_ptr[stride + 5]) * (65536 / 6) >> 16;
    dst_ptr[2] = (src_ptr[6] + src_ptr[7] +
        src_ptr[stride + 6] + src_ptr[stride + 7]) *
        (65536 / 4) >> 16;
    src_ptr += 8;
    dst_ptr += 3;
  }
}

// C version 8x2 -> 8x1
static void ScaleFilterRows_C(uint8* dst_ptr,
                              const uint8* src_ptr, ptrdiff_t src_stride,
                              int dst_width, int source_y_fraction) {
  assert(dst_width > 0);
  int y1_fraction = source_y_fraction;
  int y0_fraction = 256 - y1_fraction;
  const uint8* src_ptr1 = src_ptr + src_stride;
  uint8* end = dst_ptr + dst_width;
  do {
    dst_ptr[0] = (src_ptr[0] * y0_fraction + src_ptr1[0] * y1_fraction) >> 8;
    dst_ptr[1] = (src_ptr[1] * y0_fraction + src_ptr1[1] * y1_fraction) >> 8;
    dst_ptr[2] = (src_ptr[2] * y0_fraction + src_ptr1[2] * y1_fraction) >> 8;
    dst_ptr[3] = (src_ptr[3] * y0_fraction + src_ptr1[3] * y1_fraction) >> 8;
    dst_ptr[4] = (src_ptr[4] * y0_fraction + src_ptr1[4] * y1_fraction) >> 8;
    dst_ptr[5] = (src_ptr[5] * y0_fraction + src_ptr1[5] * y1_fraction) >> 8;
    dst_ptr[6] = (src_ptr[6] * y0_fraction + src_ptr1[6] * y1_fraction) >> 8;
    dst_ptr[7] = (src_ptr[7] * y0_fraction + src_ptr1[7] * y1_fraction) >> 8;
    src_ptr += 8;
    src_ptr1 += 8;
    dst_ptr += 8;
  } while (dst_ptr < end);
  dst_ptr[0] = dst_ptr[-1];
}

void ScaleAddRows_C(const uint8* src_ptr, ptrdiff_t src_stride,
                    uint16* dst_ptr, int src_width, int src_height) {
  assert(src_width > 0);
  assert(src_height > 0);
  for (int x = 0; x < src_width; ++x) {
    const uint8* s = src_ptr + x;
    int sum = 0;
    for (int y = 0; y < src_height; ++y) {
      sum += s[0];
      s += src_stride;
    }
    dst_ptr[x] = sum;
  }
}

/**
 * Scale plane, 1/2
 *
 * This is an optimized version for scaling down a plane to 1/2 of
 * its original size.
 *
 */
static void ScalePlaneDown2(int /* src_width */, int /* src_height */,
                            int dst_width, int dst_height,
                            int src_stride, int dst_stride,
                            const uint8* src_ptr, uint8* dst_ptr,
                            FilterMode filtering) {
  void (*ScaleRowDown2)(const uint8* src_ptr, ptrdiff_t src_stride,
                        uint8* dst_ptr, int dst_width) =
      filtering ? ScaleRowDown2Int_C : ScaleRowDown2_C;
#if defined(HAS_SCALEROWDOWN2_NEON)
  if (TestCpuFlag(kCpuHasNEON) &&
      IS_ALIGNED(dst_width, 16)) {
    ScaleRowDown2 = filtering ? ScaleRowDown2Int_NEON : ScaleRowDown2_NEON;
  }
#elif defined(HAS_SCALEROWDOWN2_SSE2)
  if (TestCpuFlag(kCpuHasSSE2) && IS_ALIGNED(dst_width, 16)) {
    ScaleRowDown2 = filtering ? ScaleRowDown2Int_Unaligned_SSE2 :
        ScaleRowDown2_Unaligned_SSE2;
    if (IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16) &&
        IS_ALIGNED(dst_ptr, 16) && IS_ALIGNED(dst_stride, 16)) {
      ScaleRowDown2 = filtering ? ScaleRowDown2Int_SSE2 : ScaleRowDown2_SSE2;
    }
  }
#endif

  // TODO(fbarchard): Loop through source height to allow odd height.
  for (int y = 0; y < dst_height; ++y) {
    ScaleRowDown2(src_ptr, src_stride, dst_ptr, dst_width);
    src_ptr += (src_stride << 1);
    dst_ptr += dst_stride;
  }
}

/**
 * Scale plane, 1/4
 *
 * This is an optimized version for scaling down a plane to 1/4 of
 * its original size.
 */
static void ScalePlaneDown4(int /* src_width */, int /* src_height */,
                            int dst_width, int dst_height,
                            int src_stride, int dst_stride,
                            const uint8* src_ptr, uint8* dst_ptr,
                            FilterMode filtering) {
  void (*ScaleRowDown4)(const uint8* src_ptr, ptrdiff_t src_stride,
                        uint8* dst_ptr, int dst_width) =
      filtering ? ScaleRowDown4Int_C : ScaleRowDown4_C;
#if defined(HAS_SCALEROWDOWN4_NEON)
  if (TestCpuFlag(kCpuHasNEON) &&
      IS_ALIGNED(dst_width, 4)) {
    ScaleRowDown4 = filtering ? ScaleRowDown4Int_NEON : ScaleRowDown4_NEON;
  }
#elif defined(HAS_SCALEROWDOWN4_SSE2)
  if (TestCpuFlag(kCpuHasSSE2) &&
      IS_ALIGNED(dst_width, 8) &&
      IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16)) {
    ScaleRowDown4 = filtering ? ScaleRowDown4Int_SSE2 : ScaleRowDown4_SSE2;
  }
#endif

  for (int y = 0; y < dst_height; ++y) {
    ScaleRowDown4(src_ptr, src_stride, dst_ptr, dst_width);
    src_ptr += (src_stride << 2);
    dst_ptr += dst_stride;
  }
}

/**
 * Scale plane, 1/8
 *
 * This is an optimized version for scaling down a plane to 1/8
 * of its original size.
 *
 */
static void ScalePlaneDown8(int /* src_width */, int /* src_height */,
                            int dst_width, int dst_height,
                            int src_stride, int dst_stride,
                            const uint8* src_ptr, uint8* dst_ptr,
                            FilterMode filtering) {
  void (*ScaleRowDown8)(const uint8* src_ptr, ptrdiff_t src_stride,
                        uint8* dst_ptr, int dst_width) =
      filtering && (dst_width <= kMaxOutputWidth) ?
      ScaleRowDown8Int_C : ScaleRowDown8_C;
#if defined(HAS_SCALEROWDOWN8_SSE2)
  if (TestCpuFlag(kCpuHasSSE2) &&
      IS_ALIGNED(dst_width, 4) &&
      IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16)) {
    ScaleRowDown8 = filtering ? ScaleRowDown8Int_SSE2 : ScaleRowDown8_SSE2;
  }
#endif

  for (int y = 0; y < dst_height; ++y) {
    ScaleRowDown8(src_ptr, src_stride, dst_ptr, dst_width);
    src_ptr += (src_stride << 3);
    dst_ptr += dst_stride;
  }
}

/**
 * Scale plane down, 3/4
 *
 * Provided by Frank Barchard (fbarchard@google.com)
 *
 */
static void ScalePlaneDown34(int /* src_width */, int /* src_height */,
                             int dst_width, int dst_height,
                             int src_stride, int dst_stride,
                             const uint8* src_ptr, uint8* dst_ptr,
                             FilterMode filtering) {
  assert(dst_width % 3 == 0);
  void (*ScaleRowDown34_0)(const uint8* src_ptr, ptrdiff_t src_stride,
                           uint8* dst_ptr, int dst_width);
  void (*ScaleRowDown34_1)(const uint8* src_ptr, ptrdiff_t src_stride,
                           uint8* dst_ptr, int dst_width);
  if (!filtering) {
    ScaleRowDown34_0 = ScaleRowDown34_C;
    ScaleRowDown34_1 = ScaleRowDown34_C;
  } else {
    ScaleRowDown34_0 = ScaleRowDown34_0_Int_C;
    ScaleRowDown34_1 = ScaleRowDown34_1_Int_C;
  }
#if defined(HAS_SCALEROWDOWN34_NEON)
  if (TestCpuFlag(kCpuHasNEON) && (dst_width % 24 == 0)) {
    if (!filtering) {
      ScaleRowDown34_0 = ScaleRowDown34_NEON;
      ScaleRowDown34_1 = ScaleRowDown34_NEON;
    } else {
      ScaleRowDown34_0 = ScaleRowDown34_0_Int_NEON;
      ScaleRowDown34_1 = ScaleRowDown34_1_Int_NEON;
    }
  }
#endif
#if defined(HAS_SCALEROWDOWN34_SSE2)
  if (TestCpuFlag(kCpuHasSSE2) && (dst_width % 24 == 0) &&
      IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16) && filtering) {
    ScaleRowDown34_0 = ScaleRowDown34_0_Int_SSE2;
    ScaleRowDown34_1 = ScaleRowDown34_1_Int_SSE2;
  }
#endif
#if defined(HAS_SCALEROWDOWN34_SSSE3)
  if (TestCpuFlag(kCpuHasSSSE3) && (dst_width % 24 == 0) &&
      IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16)) {
    if (!filtering) {
      ScaleRowDown34_0 = ScaleRowDown34_SSSE3;
      ScaleRowDown34_1 = ScaleRowDown34_SSSE3;
    } else {
      ScaleRowDown34_0 = ScaleRowDown34_0_Int_SSSE3;
      ScaleRowDown34_1 = ScaleRowDown34_1_Int_SSSE3;
    }
  }
#endif

  for (int y = 0; y < dst_height - 2; y += 3) {
    ScaleRowDown34_0(src_ptr, src_stride, dst_ptr, dst_width);
    src_ptr += src_stride;
    dst_ptr += dst_stride;
    ScaleRowDown34_1(src_ptr, src_stride, dst_ptr, dst_width);
    src_ptr += src_stride;
    dst_ptr += dst_stride;
    ScaleRowDown34_0(src_ptr + src_stride, -src_stride,
                     dst_ptr, dst_width);
    src_ptr += src_stride * 2;
    dst_ptr += dst_stride;
  }

  // Remainder 1 or 2 rows with last row vertically unfiltered
  if ((dst_height % 3) == 2) {
    ScaleRowDown34_0(src_ptr, src_stride, dst_ptr, dst_width);
    src_ptr += src_stride;
    dst_ptr += dst_stride;
    ScaleRowDown34_1(src_ptr, 0, dst_ptr, dst_width);
  } else if ((dst_height % 3) == 1) {
    ScaleRowDown34_0(src_ptr, 0, dst_ptr, dst_width);
  }
}

/**
 * Scale plane, 3/8
 *
 * This is an optimized version for scaling down a plane to 3/8
 * of its original size.
 *
 * Uses box filter arranges like this
 * aaabbbcc -> abc
 * aaabbbcc    def
 * aaabbbcc    ghi
 * dddeeeff
 * dddeeeff
 * dddeeeff
 * ggghhhii
 * ggghhhii
 * Boxes are 3x3, 2x3, 3x2 and 2x2
 */
static void ScalePlaneDown38(int /* src_width */, int /* src_height */,
                             int dst_width, int dst_height,
                             int src_stride, int dst_stride,
                             const uint8* src_ptr, uint8* dst_ptr,
                             FilterMode filtering) {
  assert(dst_width % 3 == 0);
  void (*ScaleRowDown38_3)(const uint8* src_ptr, ptrdiff_t src_stride,
                           uint8* dst_ptr, int dst_width);
  void (*ScaleRowDown38_2)(const uint8* src_ptr, ptrdiff_t src_stride,
                           uint8* dst_ptr, int dst_width);
  if (!filtering) {
    ScaleRowDown38_3 = ScaleRowDown38_C;
    ScaleRowDown38_2 = ScaleRowDown38_C;
  } else {
    ScaleRowDown38_3 = ScaleRowDown38_3_Int_C;
    ScaleRowDown38_2 = ScaleRowDown38_2_Int_C;
  }
#if defined(HAS_SCALEROWDOWN38_NEON)
  if (TestCpuFlag(kCpuHasNEON) && (dst_width % 12 == 0)) {
    if (!filtering) {
      ScaleRowDown38_3 = ScaleRowDown38_NEON;
      ScaleRowDown38_2 = ScaleRowDown38_NEON;
    } else {
      ScaleRowDown38_3 = ScaleRowDown38_3_Int_NEON;
      ScaleRowDown38_2 = ScaleRowDown38_2_Int_NEON;
    }
  }
#elif defined(HAS_SCALEROWDOWN38_SSSE3)
  if (TestCpuFlag(kCpuHasSSSE3) && (dst_width % 24 == 0) &&
      IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16)) {
    if (!filtering) {
      ScaleRowDown38_3 = ScaleRowDown38_SSSE3;
      ScaleRowDown38_2 = ScaleRowDown38_SSSE3;
    } else {
      ScaleRowDown38_3 = ScaleRowDown38_3_Int_SSSE3;
      ScaleRowDown38_2 = ScaleRowDown38_2_Int_SSSE3;
    }
  }
#endif

  for (int y = 0; y < dst_height - 2; y += 3) {
    ScaleRowDown38_3(src_ptr, src_stride, dst_ptr, dst_width);
    src_ptr += src_stride * 3;
    dst_ptr += dst_stride;
    ScaleRowDown38_3(src_ptr, src_stride, dst_ptr, dst_width);
    src_ptr += src_stride * 3;
    dst_ptr += dst_stride;
    ScaleRowDown38_2(src_ptr, src_stride, dst_ptr, dst_width);
    src_ptr += src_stride * 2;
    dst_ptr += dst_stride;
  }

  // Remainder 1 or 2 rows with last row vertically unfiltered
  if ((dst_height % 3) == 2) {
    ScaleRowDown38_3(src_ptr, src_stride, dst_ptr, dst_width);
    src_ptr += src_stride * 3;
    dst_ptr += dst_stride;
    ScaleRowDown38_3(src_ptr, 0, dst_ptr, dst_width);
  } else if ((dst_height % 3) == 1) {
    ScaleRowDown38_3(src_ptr, 0, dst_ptr, dst_width);
  }
}

static __inline uint32 SumBox(int iboxwidth, int iboxheight,
                              ptrdiff_t src_stride, const uint8* src_ptr) {
  assert(iboxwidth > 0);
  assert(iboxheight > 0);
  uint32 sum = 0u;
  for (int y = 0; y < iboxheight; ++y) {
    for (int x = 0; x < iboxwidth; ++x) {
      sum += src_ptr[x];
    }
    src_ptr += src_stride;
  }
  return sum;
}

static void ScalePlaneBoxRow_C(int dst_width, int boxheight,
                               int x, int dx, ptrdiff_t src_stride,
                               const uint8* src_ptr, uint8* dst_ptr) {
  for (int i = 0; i < dst_width; ++i) {
    int ix = x >> 16;
    x += dx;
    int boxwidth = (x >> 16) - ix;
    *dst_ptr++ = SumBox(boxwidth, boxheight, src_stride, src_ptr + ix) /
        (boxwidth * boxheight);
  }
}

static __inline uint32 SumPixels(int iboxwidth, const uint16* src_ptr) {
  assert(iboxwidth > 0);
  uint32 sum = 0u;
  for (int x = 0; x < iboxwidth; ++x) {
    sum += src_ptr[x];
  }
  return sum;
}

static void ScaleAddCols2_C(int dst_width, int boxheight, int x, int dx,
                            const uint16* src_ptr, uint8* dst_ptr) {
  int scaletbl[2];
  int minboxwidth = (dx >> 16);
  scaletbl[0] = 65536 / (minboxwidth * boxheight);
  scaletbl[1] = 65536 / ((minboxwidth + 1) * boxheight);
  int *scaleptr = scaletbl - minboxwidth;
  for (int i = 0; i < dst_width; ++i) {
    int ix = x >> 16;
    x += dx;
    int boxwidth = (x >> 16) - ix;
    *dst_ptr++ = SumPixels(boxwidth, src_ptr + ix) * scaleptr[boxwidth] >> 16;
  }
}

static void ScaleAddCols1_C(int dst_width, int boxheight, int x, int dx,
                            const uint16* src_ptr, uint8* dst_ptr) {
  int boxwidth = (dx >> 16);
  int scaleval = 65536 / (boxwidth * boxheight);
  for (int i = 0; i < dst_width; ++i) {
    *dst_ptr++ = SumPixels(boxwidth, src_ptr + x) * scaleval >> 16;
    x += boxwidth;
  }
}

/**
 * Scale plane down to any dimensions, with interpolation.
 * (boxfilter).
 *
 * Same method as SimpleScale, which is fixed point, outputting
 * one pixel of destination using fixed point (16.16) to step
 * through source, sampling a box of pixel with simple
 * averaging.
 */
static void ScalePlaneBox(int src_width, int src_height,
                          int dst_width, int dst_height,
                          int src_stride, int dst_stride,
                          const uint8* src_ptr, uint8* dst_ptr) {
  assert(dst_width > 0);
  assert(dst_height > 0);
  int dx = (src_width << 16) / dst_width;
  int dy = (src_height << 16) / dst_height;
  int x = (dx >= 65536) ? ((dx >> 1) - 32768) : (dx >> 1);
  int y = (dy >= 65536) ? ((dy >> 1) - 32768) : (dy >> 1);
  int maxy = (src_height << 16);
  if (!IS_ALIGNED(src_width, 16) || (src_width > kMaxInputWidth) ||
      dst_height * 2 > src_height) {
    uint8* dst = dst_ptr;
    for (int j = 0; j < dst_height; ++j) {
      int iy = y >> 16;
      const uint8* src = src_ptr + iy * src_stride;
      y += dy;
      if (y > maxy) {
        y = maxy;
      }
      int boxheight = (y >> 16) - iy;
      ScalePlaneBoxRow_C(dst_width, boxheight,
                         x, dx, src_stride,
                         src, dst);
      dst += dst_stride;
    }
  } else {
    SIMD_ALIGNED(uint16 row[kMaxInputWidth]);
    void (*ScaleAddRows)(const uint8* src_ptr, ptrdiff_t src_stride,
                         uint16* dst_ptr, int src_width, int src_height)=
        ScaleAddRows_C;
    void (*ScaleAddCols)(int dst_width, int boxheight, int x, int dx,
                         const uint16* src_ptr, uint8* dst_ptr);
    if (dx & 0xffff) {
      ScaleAddCols = ScaleAddCols2_C;
    } else {
      ScaleAddCols = ScaleAddCols1_C;
    }
#if defined(HAS_SCALEADDROWS_SSE2)
    if (TestCpuFlag(kCpuHasSSE2) &&
        IS_ALIGNED(src_stride, 16) && IS_ALIGNED(src_ptr, 16)) {
      ScaleAddRows = ScaleAddRows_SSE2;
    }
#endif

    for (int j = 0; j < dst_height; ++j) {
      int iy = y >> 16;
      const uint8* src = src_ptr + iy * src_stride;
      y += dy;
      if (y > (src_height << 16)) {
        y = (src_height << 16);
      }
      int boxheight = (y >> 16) - iy;
      ScaleAddRows(src, src_stride, row, src_width, boxheight);
      ScaleAddCols(dst_width, boxheight, x, dx, row, dst_ptr);
      dst_ptr += dst_stride;
    }
  }
}

/**
 * Scale plane to/from any dimensions, with interpolation.
 */
static void ScalePlaneBilinearSimple(int src_width, int src_height,
                                     int dst_width, int dst_height,
                                     int src_stride, int dst_stride,
                                     const uint8* src_ptr, uint8* dst_ptr) {
  int dx = (src_width << 16) / dst_width;
  int dy = (src_height << 16) / dst_height;
  int y = (dy >= 65536) ? ((dy >> 1) - 32768) : (dy >> 1);
  int maxx = (src_width > 1) ? ((src_width - 1) << 16) - 1 : 0;
  int maxy = (src_height > 1) ? ((src_height - 1) << 16) - 1 : 0;
  for (int i = 0; i < dst_height; ++i) {
    int x = (dx >= 65536) ? ((dx >> 1) - 32768) : (dx >> 1);
    int yi = y >> 16;
    int yf = y & 0xffff;
    const uint8* src0 = src_ptr + yi * src_stride;
    const uint8* src1 = (yi < src_height - 1) ? src0 + src_stride : src0;
    uint8* dst = dst_ptr;
    for (int j = 0; j < dst_width; ++j) {
      int xi = x >> 16;
      int xf = x & 0xffff;
      int x1 = (xi < src_width - 1) ? xi + 1 : xi;
      int a = src0[xi];
      int b = src0[x1];
      int r0 = BLENDER(a, b, xf);
      a = src1[xi];
      b = src1[x1];
      int r1 = BLENDER(a, b, xf);
      *dst++ = BLENDER(r0, r1, yf);
      x += dx;
      if (x > maxx)
        x = maxx;
    }
    dst_ptr += dst_stride;
    y += dy;
    if (y > maxy)
      y = maxy;
  }
}

/**
 * Scale plane to/from any dimensions, with bilinear
 * interpolation.
 */
void ScalePlaneBilinear(int src_width, int src_height,
                        int dst_width, int dst_height,
                        int src_stride, int dst_stride,
                        const uint8* src_ptr, uint8* dst_ptr) {
  assert(dst_width > 0);
  assert(dst_height > 0);
  if (!IS_ALIGNED(src_width, 8) || (src_width > kMaxInputWidth)) {
    ScalePlaneBilinearSimple(src_width, src_height, dst_width, dst_height,
                             src_stride, dst_stride, src_ptr, dst_ptr);

  } else {
    SIMD_ALIGNED(uint8 row[kMaxInputWidth + 16]);
    void (*ScaleFilterRows)(uint8* dst_ptr, const uint8* src_ptr,
                            ptrdiff_t src_stride,
                            int dst_width, int source_y_fraction) =
        ScaleFilterRows_C;
#if defined(HAS_SCALEFILTERROWS_NEON)
    if (TestCpuFlag(kCpuHasNEON)) {
      ScaleFilterRows = ScaleFilterRows_NEON;
    }
#endif
#if defined(HAS_SCALEFILTERROWS_SSE2)
    if (TestCpuFlag(kCpuHasSSE2) &&
        IS_ALIGNED(src_stride, 16) && IS_ALIGNED(src_ptr, 16)) {
      ScaleFilterRows = ScaleFilterRows_SSE2;
    }
#endif
#if defined(HAS_SCALEFILTERROWS_SSSE3)
    if (TestCpuFlag(kCpuHasSSSE3) &&
        IS_ALIGNED(src_stride, 16) && IS_ALIGNED(src_ptr, 16)) {
      ScaleFilterRows = ScaleFilterRows_SSSE3;
    }
#endif

    int dx = (src_width << 16) / dst_width;
    int dy = (src_height << 16) / dst_height;
    int x = (dx >= 65536) ? ((dx >> 1) - 32768) : (dx >> 1);
    int y = (dy >= 65536) ? ((dy >> 1) - 32768) : (dy >> 1);
    int maxy = (src_height > 1) ? ((src_height - 1) << 16) - 1 : 0;
    for (int j = 0; j < dst_height; ++j) {
      int yi = y >> 16;
      int yf = (y >> 8) & 255;
      const uint8* src = src_ptr + yi * src_stride;
      ScaleFilterRows(row, src, src_stride, src_width, yf);
      ScaleFilterCols_C(dst_ptr, row, dst_width, x, dx);
      dst_ptr += dst_stride;
      y += dy;
      if (y > maxy) {
        y = maxy;
      }
    }
  }
}

/**
 * Scale plane to/from any dimensions, without interpolation.
 * Fixed point math is used for performance: The upper 16 bits
 * of x and dx is the integer part of the source position and
 * the lower 16 bits are the fixed decimal part.
 */
static void ScalePlaneSimple(int src_width, int src_height,
                             int dst_width, int dst_height,
                             int src_stride, int dst_stride,
                             const uint8* src_ptr, uint8* dst_ptr) {
  int dx = (src_width << 16) / dst_width;
  int dy = (src_height << 16) / dst_height;
  int y = (dy >= 65536) ? ((dy >> 1) - 32768) : (dy >> 1);
  for (int j = 0; j < dst_height; ++j) {
    int x = (dx >= 65536) ? ((dx >> 1) - 32768) : (dx >> 1);
    int yi = y >> 16;
    const uint8* src = src_ptr + yi * src_stride;
    uint8* dst = dst_ptr;
    for (int i = 0; i < dst_width; ++i) {
      *dst++ = src[x >> 16];
      x += dx;
    }
    dst_ptr += dst_stride;
    y += dy;
  }
}

/**
 * Scale plane to/from any dimensions.
 */
static void ScalePlaneAnySize(int src_width, int src_height,
                              int dst_width, int dst_height,
                              int src_stride, int dst_stride,
                              const uint8* src_ptr, uint8* dst_ptr,
                              FilterMode filtering) {
  if (!filtering) {
    ScalePlaneSimple(src_width, src_height, dst_width, dst_height,
                     src_stride, dst_stride, src_ptr, dst_ptr);
  } else {
    // fall back to non-optimized version
    ScalePlaneBilinear(src_width, src_height, dst_width, dst_height,
                       src_stride, dst_stride, src_ptr, dst_ptr);
  }
}

/**
 * Scale plane down, any size
 *
 * This is an optimized version for scaling down a plane to any size.
 * The current implementation is ~10 times faster compared to the
 * reference implementation for e.g. XGA->LowResPAL
 *
 */
static void ScalePlaneDown(int src_width, int src_height,
                           int dst_width, int dst_height,
                           int src_stride, int dst_stride,
                           const uint8* src_ptr, uint8* dst_ptr,
                           FilterMode filtering) {
  if (!filtering) {
    ScalePlaneSimple(src_width, src_height, dst_width, dst_height,
                     src_stride, dst_stride, src_ptr, dst_ptr);
  } else if (filtering == kFilterBilinear || src_height * 2 > dst_height) {
    // between 1/2x and 1x use bilinear
    ScalePlaneBilinear(src_width, src_height, dst_width, dst_height,
                       src_stride, dst_stride, src_ptr, dst_ptr);
  } else {
    ScalePlaneBox(src_width, src_height, dst_width, dst_height,
                  src_stride, dst_stride, src_ptr, dst_ptr);
  }
}

// Scale a plane.
// This function in turn calls a scaling function suitable for handling
// the desired resolutions.

LIBYUV_API
void ScalePlane(const uint8* src, int src_stride,
                int src_width, int src_height,
                uint8* dst, int dst_stride,
                int dst_width, int dst_height,
                FilterMode filtering) {
#ifdef CPU_X86
  // environment variable overrides for testing.
  char *filter_override = getenv("LIBYUV_FILTER");
  if (filter_override) {
    filtering = (FilterMode)atoi(filter_override);  // NOLINT
  }
#endif
  // Use specialized scales to improve performance for common resolutions.
  // For example, all the 1/2 scalings will use ScalePlaneDown2()
  if (dst_width == src_width && dst_height == src_height) {
    // Straight copy.
    CopyPlane(src, src_stride, dst, dst_stride, dst_width, dst_height);
  } else if (dst_width <= src_width && dst_height <= src_height) {
    // Scale down.
    if (use_reference_impl_) {
      // For testing, allow the optimized versions to be disabled.
      ScalePlaneDown(src_width, src_height, dst_width, dst_height,
                     src_stride, dst_stride, src, dst, filtering);
    } else if (4 * dst_width == 3 * src_width &&
               4 * dst_height == 3 * src_height) {
      // optimized, 3/4
      ScalePlaneDown34(src_width, src_height, dst_width, dst_height,
                       src_stride, dst_stride, src, dst, filtering);
    } else if (2 * dst_width == src_width && 2 * dst_height == src_height) {
      // optimized, 1/2
      ScalePlaneDown2(src_width, src_height, dst_width, dst_height,
                      src_stride, dst_stride, src, dst, filtering);
    // 3/8 rounded up for odd sized chroma height.
    } else if (8 * dst_width == 3 * src_width &&
               dst_height == ((src_height * 3 + 7) / 8)) {
      // optimized, 3/8
      ScalePlaneDown38(src_width, src_height, dst_width, dst_height,
                       src_stride, dst_stride, src, dst, filtering);
    } else if (4 * dst_width == src_width && 4 * dst_height == src_height &&
               filtering != kFilterBilinear) {
      // optimized, 1/4
      ScalePlaneDown4(src_width, src_height, dst_width, dst_height,
                      src_stride, dst_stride, src, dst, filtering);
    } else if (8 * dst_width == src_width && 8 * dst_height == src_height &&
               filtering != kFilterBilinear) {
      // optimized, 1/8
      ScalePlaneDown8(src_width, src_height, dst_width, dst_height,
                      src_stride, dst_stride, src, dst, filtering);
    } else {
      // Arbitrary downsample
      ScalePlaneDown(src_width, src_height, dst_width, dst_height,
                     src_stride, dst_stride, src, dst, filtering);
    }
  } else {
    // Arbitrary scale up and/or down.
    ScalePlaneAnySize(src_width, src_height, dst_width, dst_height,
                      src_stride, dst_stride, src, dst, filtering);
  }
}

// Scale an I420 image.
// This function in turn calls a scaling function for each plane.

#define UNDER_ALLOCATED_HACK 1

LIBYUV_API
int I420Scale(const uint8* src_y, int src_stride_y,
              const uint8* src_u, int src_stride_u,
              const uint8* src_v, int src_stride_v,
              int src_width, int src_height,
              uint8* dst_y, int dst_stride_y,
              uint8* dst_u, int dst_stride_u,
              uint8* dst_v, int dst_stride_v,
              int dst_width, int dst_height,
              FilterMode filtering) {
  if (!src_y || !src_u || !src_v || src_width <= 0 || src_height == 0 ||
      !dst_y || !dst_u || !dst_v || dst_width <= 0 || dst_height <= 0) {
    return -1;
  }
  // Negative height means invert the image.
  if (src_height < 0) {
    src_height = -src_height;
    int halfheight = (src_height + 1) >> 1;
    src_y = src_y + (src_height - 1) * src_stride_y;
    src_u = src_u + (halfheight - 1) * src_stride_u;
    src_v = src_v + (halfheight - 1) * src_stride_v;
    src_stride_y = -src_stride_y;
    src_stride_u = -src_stride_u;
    src_stride_v = -src_stride_v;
  }
  int src_halfwidth = (src_width + 1) >> 1;
  int src_halfheight = (src_height + 1) >> 1;
  int dst_halfwidth = (dst_width + 1) >> 1;
  int dst_halfheight = (dst_height + 1) >> 1;

#ifdef UNDER_ALLOCATED_HACK
  // If caller passed width / 2 for stride, adjust halfwidth to match.
  if ((src_width & 1) && src_stride_u && src_halfwidth > abs(src_stride_u)) {
    src_halfwidth = src_width >> 1;
  }
  if ((dst_width & 1) && dst_stride_u && dst_halfwidth > abs(dst_stride_u)) {
    dst_halfwidth = dst_width >> 1;
  }
  // If caller used height / 2 when computing src_v, it will point into what
  // should be the src_u plane. Detect this and reduce halfheight to match.
  int uv_src_plane_size = src_halfwidth * src_halfheight;
  if ((src_height & 1) &&
      (src_v > src_u) && (src_v < (src_u + uv_src_plane_size))) {
    src_halfheight = src_height >> 1;
  }
  int uv_dst_plane_size = dst_halfwidth * dst_halfheight;
  if ((dst_height & 1) &&
      (dst_v > dst_u) && (dst_v < (dst_u + uv_dst_plane_size))) {
    dst_halfheight = dst_height >> 1;
  }
#endif

  ScalePlane(src_y, src_stride_y, src_width, src_height,
             dst_y, dst_stride_y, dst_width, dst_height,
             filtering);
  ScalePlane(src_u, src_stride_u, src_halfwidth, src_halfheight,
             dst_u, dst_stride_u, dst_halfwidth, dst_halfheight,
             filtering);
  ScalePlane(src_v, src_stride_v, src_halfwidth, src_halfheight,
             dst_v, dst_stride_v, dst_halfwidth, dst_halfheight,
             filtering);
  return 0;
}

// Deprecated api
LIBYUV_API
int Scale(const uint8* src_y, const uint8* src_u, const uint8* src_v,
          int src_stride_y, int src_stride_u, int src_stride_v,
          int src_width, int src_height,
          uint8* dst_y, uint8* dst_u, uint8* dst_v,
          int dst_stride_y, int dst_stride_u, int dst_stride_v,
          int dst_width, int dst_height,
          bool interpolate) {
  if (!src_y || !src_u || !src_v || src_width <= 0 || src_height == 0 ||
      !dst_y || !dst_u || !dst_v || dst_width <= 0 || dst_height <= 0) {
    return -1;
  }
  // Negative height means invert the image.
  if (src_height < 0) {
    src_height = -src_height;
    int halfheight = (src_height + 1) >> 1;
    src_y = src_y + (src_height - 1) * src_stride_y;
    src_u = src_u + (halfheight - 1) * src_stride_u;
    src_v = src_v + (halfheight - 1) * src_stride_v;
    src_stride_y = -src_stride_y;
    src_stride_u = -src_stride_u;
    src_stride_v = -src_stride_v;
  }
  int src_halfwidth = (src_width + 1) >> 1;
  int src_halfheight = (src_height + 1) >> 1;
  int dst_halfwidth = (dst_width + 1) >> 1;
  int dst_halfheight = (dst_height + 1) >> 1;
  FilterMode filtering = interpolate ? kFilterBox : kFilterNone;

#ifdef UNDER_ALLOCATED_HACK
  // If caller passed width / 2 for stride, adjust halfwidth to match.
  if ((src_width & 1) && src_stride_u && src_halfwidth > abs(src_stride_u)) {
    src_halfwidth = src_width >> 1;
  }
  if ((dst_width & 1) && dst_stride_u && dst_halfwidth > abs(dst_stride_u)) {
    dst_halfwidth = dst_width >> 1;
  }
  // If caller used height / 2 when computing src_v, it will point into what
  // should be the src_u plane. Detect this and reduce halfheight to match.
  int uv_src_plane_size = src_halfwidth * src_halfheight;
  if ((src_height & 1) &&
      (src_v > src_u) && (src_v < (src_u + uv_src_plane_size))) {
    src_halfheight = src_height >> 1;
  }
  int uv_dst_plane_size = dst_halfwidth * dst_halfheight;
  if ((dst_height & 1) &&
      (dst_v > dst_u) && (dst_v < (dst_u + uv_dst_plane_size))) {
    dst_halfheight = dst_height >> 1;
  }
#endif

  ScalePlane(src_y, src_stride_y, src_width, src_height,
             dst_y, dst_stride_y, dst_width, dst_height,
             filtering);
  ScalePlane(src_u, src_stride_u, src_halfwidth, src_halfheight,
             dst_u, dst_stride_u, dst_halfwidth, dst_halfheight,
             filtering);
  ScalePlane(src_v, src_stride_v, src_halfwidth, src_halfheight,
             dst_v, dst_stride_v, dst_halfwidth, dst_halfheight,
             filtering);
  return 0;
}

// Deprecated api
LIBYUV_API
int ScaleOffset(const uint8* src, int src_width, int src_height,
                uint8* dst, int dst_width, int dst_height, int dst_yoffset,
                bool interpolate) {
  if (!src || src_width <= 0 || src_height <= 0 ||
      !dst || dst_width <= 0 || dst_height <= 0 || dst_yoffset < 0 ||
      dst_yoffset >= dst_height) {
    return -1;
  }
  dst_yoffset = dst_yoffset & ~1;  // chroma requires offset to multiple of 2.
  int src_halfwidth = (src_width + 1) >> 1;
  int src_halfheight = (src_height + 1) >> 1;
  int dst_halfwidth = (dst_width + 1) >> 1;
  int dst_halfheight = (dst_height + 1) >> 1;
  int aheight = dst_height - dst_yoffset * 2;  // actual output height
  const uint8* src_y = src;
  const uint8* src_u = src + src_width * src_height;
  const uint8* src_v = src + src_width * src_height +
                             src_halfwidth * src_halfheight;
  uint8* dst_y = dst + dst_yoffset * dst_width;
  uint8* dst_u = dst + dst_width * dst_height +
                 (dst_yoffset >> 1) * dst_halfwidth;
  uint8* dst_v = dst + dst_width * dst_height + dst_halfwidth * dst_halfheight +
                 (dst_yoffset >> 1) * dst_halfwidth;
  return Scale(src_y, src_u, src_v, src_width, src_halfwidth, src_halfwidth,
               src_width, src_height, dst_y, dst_u, dst_v, dst_width,
               dst_halfwidth, dst_halfwidth, dst_width, aheight, interpolate);
}

#ifdef __cplusplus
}  // extern "C"
}  // namespace libyuv
#endif