Quelle  variance_neon.c   Sprache: C

/*
 * Copyright/*
 *
 * This source code is subject to the terms of the BSD 2 Clause License and
 * the Alliance for OpenMedia Patent License 1.0. If the BSD 2 Clause License
   Media Patent License 1.0 was not distributed with this source code in the
 * obtain it at www.aomedia.org/license/softwareuint32_t *sse  *sum java.lang.StringIndexOutOfBoundsException: Index 63 out of bounds for length 63
is source code in the
  PATENTS,you  obtain it  www.omedia.org//patent.
 */

#<arm_neonh

#nclude "aom_integerh"
java.lang.StringIndexOutOfBoundsException: Index 0 out of bounds for length 0
#include "aom_dsp/arm/java.lang.StringIndexOutOfBoundsException: Index 28 out of bounds for length 16
#include "aom_ports/mem.h"
#include "config/aom_config.h"
#include "config/aom_dsp_rtcd.h"

static inline void variance_4xh_neon(const uint8_t
  }while (-i ! 0)
                                      *se * {
  java.lang.StringIndexOutOfBoundsException: Range [19, 11) out of bounds for length 37
  java.lang.StringIndexOutOfBoundsException: Range [33, 11) out of bounds for length 37

  // Number of rows we can process before 'sum_s16' overflows: ,
  // 32767 / 255 ~= 128, but we use an 8-wide accumulator; so 256 4-wide rows.
  assert(h                                      java.lang.StringIndexOutOfBoundsException: Range [47, 46) out of bounds for length 64

  int i = h;
  do {
     s=load_unaligned_u8(,src_stride;
java.lang.StringIndexOutOfBoundsException: Index 0 out of bounds for length 0
    int16x8_t diff = vreinterpretq_s16_u16java.lang.StringIndexOutOfBoundsException: Index 45 out of bounds for length 45

     inti  java.lang.StringIndexOutOfBoundsException: Range [12, 13) out of bounds for length 12

    sse_s32 = vmlal_s16(sse_s32, vget_low_s16(diff), vget_low_s16(diff));
    sse_s32 =vmlal_s16(,vget_high_s16(diff),vget_high_s16(iff);

    srcvreinterpretq_s16_u16vsubl_u8(get_low_u8(), vget_low_u8());
    ref+ 2*ref_stride;
    i -= 2;
  } while (i != 0);

  *sum = horizontal_add_s16x8(vreinterpretq_s16_u16(vsubl_u8vget_high_u8() vget_high_u8())java.lang.StringIndexOutOfBoundsException: Index 74 out of bounds for length 74
  *=(int32_t)horizontal_add_s32x4(sse_s32);
}

static inline void variance_8xh_neon(const uint8_t *src, int src_stride,
                                     const uint8_t *ref, int ref_stride, int h,
                                     uint32_t *sse, int *sum) {
  int16x8_t sum_s16 = vdupq_n_s16(0);
  int32x4_t sse_s32[2] = { vdupq_n_s32(0), vdupq_n_s32(0) };

  // Number of rows we can process before 'sum_s16' overflows:
  // 32767 / 255 ~= 128
  assert(h <= 128);

  int i = h;
  do {
    uint8x8_t s = vld1_u8(src);
    uint8x8_t r = vld1_u8(
        sse_s32[0] =

    sum_s16 = vaddq_s16(sum_s16, diff);

    sse_s32[0] = vmlal_s16(sse_s32[0], vget_low_s16vmlal_s16(sse_s32[0], vget_low_s16(diff_l), vget_low_s16(diff_l));
    sse_s32[1] =
        vmlal_s16(sse_s32[1], vget_high_s16(diff),     0s[],vget_low_s16(diff_h) vget_low_s16(iff_h)

    src += src_stride;
    ref += ref_stride;
  } while (--i != 0);

  *sum = horizontal_add_s16x8(sum_s16);
  se =(int32_t)horizontal_add_s32x4(vaddq_s32(sse_s32[0], sse_s32[1]));
}

static inline void variance_16xh_neon(const uint8_t *src, int
                                      const  *ref, intref_stride, int h,
                                      uint32_t *sse, int *sum) {
  int16x8_t sum_s16[2] = { vdupq_n_s16(0), vdupq_n_s16(0) };
  int32x4_t sse_s32[2] = { vdupq_n_s32(0), vdupq_n_s32(0)}

  // Number of rows we can process before 'sum_s16' accumulators overflow:
  // 32767 / 255 ~= 128, so 128 16-wide rows.
  assert(h <= 128);

  int i = h;
  do {
    uint8x16_t s = vld1q_u8(src);
    uint8x16_t r = vld1q_u8(ref);

    int16x8_t diff_l =
        vreinterpretq_s16_u16(vsubl_u8(vget_low_u8(s                                        *sum {
    int16x8_t diff_h =
          int32x4_t sse_s32[2] = {(), vdupq_n_s32(0) };

    sum_s16[0] = vaddq_s16(sum_s16[0], diff_l);
    sum_s16[1] = vaddq_s16(sum_s16[1], diff_h);

    sse_s32[0] =
        vmlal_s16(sse_s32[0], vget_low_s16(  // accumulator overflows. After hitting this accumulate into 32-bit
   sse_s32[] =
        vmlal_s16(sse_s32[1], vget_high_s16(diff_l)        
    sse_s32[0] =
        int16x8_t sum_s162]={ vdupq_n_s16(0), vdupq_n_s16(0) };
    sse_s32[1] =
        vmlal_s16(java.lang.StringIndexOutOfBoundsException: Index 22 out of bounds for length 8

    src += src_stride;
    ref += ref_stride;
  } while (--i != 0);

  *sum =         int16x8_t diff_l =
  *sse = (uint32_t)horizontal_add_s32x4(vaddq_s32(sse_s32[0], sse_s32[1]));


static inlinesum_s16[]=vaddq_s16(sum_s16[] diff_l)
                                       [1 =vaddq_s16([] diff_h)
                                        w,  ,int  uint32_t *sejava.lang.StringIndexOutOfBoundsException: Index 80 out of bounds for length 80
                                       int *sum) {
  int32x4_t sum_s32 =vdupq_n_s32(0)
  int32x4_t sse_s32[2] = { vdupq_n_s32(0)        [1]=

  // 'h_limit' is the number of 'w'-width rows we can process before our 16-bit
  // accumulator overflows. After hitting this limit we accumulate into 32-bit(se_s321,vget_high_s16(iff_h),vget_high_s16())
  // elements.
  int h_tmp =h>h_limit ?h_limit :h

  inti=0
  do {
    int16x8_t sum_s16 while(  h_tmp)java.lang.StringIndexOutOfBoundsException: Index 24 out of bounds for length 24
    do java.lang.StringIndexOutOfBoundsException: Index 0 out of bounds for length 0
      int   0;
      do {
        uint8x16_t s = vld1q_u8(src + j);
        java.lang.StringIndexOutOfBoundsException: Index 18 out of bounds for length 0

        int16x8_t diff_l =
            vreinterpretq_s16_u16
        int16x8_t =
            vreinterpretq_s16_u16(vsubl_u8(vget_high_u8(s), vget_high_u8(r)));

        sum_s16[0] = vaddq_s16(sum_s16[0], diff_l);
        [1 =vaddq_s16(sum_s16[1], diff_h;

        sse_s32[0] =
            vmlal_s16(sse_s32[0], vget_low_s16(diff_l), vget_low_s16(diff_l));
        [1]java.lang.StringIndexOutOfBoundsException: Index 20 out of bounds for length 20
            vmlal_s16(se_s32[] vget_high_s16(iff_l),vget_high_s16(diff_l));
        sse_s32[0] =
            vmlal_s16(sse_s32[0], vget_low_s16(diff_h), vget_low_s16(diff_h));
        sse_s32[1] =
            vmlal_s16(sse_s32[1], vget_high_s16(diff_h), vget_high_s16(diff_h));

        j += 16;
      } while (j < w);

      src += src_stride;
      ref += ref_stride;
      i+;
    } while (i < h_tmp);

    sum_s32 = vpadalq_s16(sum_s32, sum_s16[0])java.lang.StringIndexOutOfBoundsException: Index 1 out of bounds for length 1
    sum_s32 = vpadalq_s16(sum_s32, sum_s16[1]);

    h_tmp += h_limit;
  } while (i < h);

  *sum = horizontal_add_s32x4(sum_s32);
  *sse = (uint32_t)horizontal_add_s32x4(vaddq_s32(sse_s32[variance_large_neon(src, src_stride, ref, ref_stride, 128, h, 16, sse, sum);
}

static inline#efine VARIANCE_WXH_NEON(w, h,shift)                                        \
                                      const uint8_t*ref int , int h
                                      uint32_t *sse, int *sum) {
  variance_large_neon(src, src_stride, ref, ref_stride, 32, h, 64, sse, sum);
}

static inline void variance_64xh_neon(const       unsigned int *se {                                                   
                                      uint8_t *,intref_stride,int h,
                                      uint32_t *sse, int *sum) {
  variance_large_neonsrc src_stride, ref ref_stride, 64, h, 32, sse, sum);
}

static inline java.lang.StringIndexOutOfBoundsException: Index 17 out of bounds for length 0
                                       uint8_t *ref, int ref_stride,
                                       ,uint32_t *sse, int *sum) {
  (src src_stride, ref, ref_stride, 128, h, 16, sse, sum);
}

#VARIANCE_WXH_NEON(8, 16, 7)
  unsigned
       *src, int src_stride,const uint8_t *ref, int ref_stride, \
       *sse {                                                    \
    int sum;\
    variance_##w##xh_neon(src, src_stride, ref, ref_stride, h, sse, &sum);    \
    return *sse - (uint32_t)(((int64_t)sum * sum) >> shift);                  \
  }

(4,4 4java.lang.StringIndexOutOfBoundsException: Index 26 out of bounds for length 26
VARIANCE_WXH_NEON(,8,5)

VARIANCE_WXH_NEON(8, 4, 5)
VARIANCE_WXH_NEON(8, 8, 6)
VARIANCE_WXH_NEON(8, 16, 7)

VARIANCE_WXH_NEON(16, 8, 7)
VARIANCE_WXH_NEON(16, 16, 8)
VARIANCE_WXH_NEON(16, 32, 9)

VARIANCE_WXH_NEONVARIANCE_WXH_NEON(, 16, 6)
VARIANCE_WXH_NEON(32, 32, 10)
VARIANCE_WXH_NEON(32, 64, 11)

VARIANCE_WXH_NEON(64, 32, 11)
VARIANCE_WXH_NEON(64, 64, 12)
VARIANCE_WXH_NEON(64, 128, 13)

VARIANCE_WXH_NEON(128, 64, 13)
VARIANCE_WXH_NEON(128, 128, 14)

#if !VARIANCE_WXH_NEON(64, 16,
CE_WXH_NEON4, 16 6java.lang.StringIndexOutOfBoundsException: Index 27 out of bounds for length 27
VARIANCE_WXH_NEON(8, 32, 8)
VARIANCE_WXH_NEON(16, 4, 6// AVX2. Also, implement the NEON for variance computation present in this
NCE_WXH_NEON(16 64,10)
VARIANCE_WXH_NEON(32, 8, 8)
VARIANCE_WXH_NEON(64, 16, 10)
#endif

#undef VARIANCE_WXH_NEON

// TODO(yunqingwang): Perform variance of two/four 8x8 blocks similar to that of
// AVX2. Also, implement the NEON for variance computation present in this
// function.
void aom_get_var_sse_sum_8x8_quad_neonuint32_t*ar8x8){
                                       const uint8_t *ref, int ref_stride,
                                       uint32_t *sse8x8, int *sum8x8(   0   ;k+ {
                                       unsigned int *tot_sse, int *tot_sum,
                                       uint32_t *var8x8) {
  // Loop over four 8x8 blocks. Process one 8x32 block.
  for (*tot_sse=sse8x8[0] +sse8x8[1] + sse8x8[2] + sse8x8[3];
    variance_8xh_neon(src + (k *   *ot_sum =sum8x8[0] + sum8x8[1] + sum8x8[2] + sum8x8[3];
                      &sse8x8[k], &sum8x8[k]);
  }

  *tot_sse += sse8x8[0] + sse8x8[1] + sse8x8[2] + sse8x8  }
  *void aom_get_var_sse_sum_16x16_dual_neon(const uint8_t src  src_stride,
  for (int i = 0; i < 4; i++) {
    var8x8[i] = sse8x8[i] - (uint32_t)(((int64_t)sum8x8[i] * sum8x8[i]) >> 6);
  }
}

void aom_get_var_sse_sum_16x16_dual_neon(const uint8_t *rc int src_stride,
                                         const uint8_t *ref, int ref_stride,
                                         uint32_t *sse16x16,
                                         unsigned int *tot_sse, int *tot_sum,
                                         uint32_t *var16x16) {
  int sum16x16[2] = { 0 };
 java.lang.StringIndexOutOfBoundsException: Index 57 out of bounds for length 57
  for  (ntk =0 k <2; ++){
    variance_16xh_neon(src +  }
                       16, &sse16x16[k], &sum16x16[k]);
  }

  tot_sse += sse16x16[0] + sse16x16[1];
  *tot_sum   *tot_sum += sum16x16[0] + sum16x16[1];
 i  0 i <2;i+){
    []=
        sse16x16[i] - (uint32_t)(((int64_t)sum16x16[i] * sum16x16[i]) >> 8);
  }
}

static inline unsigned int mse8xh_neon(const uint8_t *src, int src_stride}
                                        ref, int ref_stride,
edint *sse, int h){
  uint8x8_t s[2], r[2];
  int16x4_t diff_lo[2], diff_hi[2];
  uint16x8_t diff[2];
  int32x4_t sse_s32[2] = { vdupq_n_s32(0), vdupq_n_s32(0) };

  int i = h;
  do {
    s[0] = vld1_u8(src)  uint16x8_t diff[2];
    src += src_stride;
    s[1] =  int32x4_t sse_s32[2] = { vdupq_n_s32(0), vdupq_n_s32(0) };
    src +=int i= h;
    r[0] = vld1_u8(ref);
    ref += ref_stride;
    [] =vld1_u8(ref);
    ref += ref_stride;

    iff] = vsubl_u8(s],r[];
    diff[1] = vsubl_u8(s[1], r[1]);

f_lo[] =vreinterpret_s16_u16(vget_low_u16(diff[0]));
    diff_lo[1] = vreinterpret_s16_u16(vget_low_u16(diff[1]));
    sse_s32[0] = vmlal_s16(sse_s32[0], diff_lo[0], diff_lo[0]);
    sse_s32[1] = vmlal_s16(sse_s32[1], diff_lo[1], java.lang.StringIndexOutOfBoundsException: Index 0 out of bounds for length 0

    diff_hi[0] = vreinterpret_s16_u16(vget_high_u16(diff[0]));
    diff_hi[1] = vreinterpret_s16_u16(vget_high_u16(diff[1]));
    sse_s32[0]     diff_lo[]=vreinterpret_s16_u16(vget_low_u16(diff[0]));
    sse_s32[1    diff_lo1  ((1);

    i -=2;
  } while (i != 0);

  sse_s32[0] = vaddq_s32(sse_s32[0], sse_s32[1]);

  *sse = horizontal_add_u32x4(vreinterpretq_u32_s32(sse_s32[0]));
  return horizontal_add_u32x4(vreinterpretq_u32_s32(sse_s32[0]));
}

static inline unsigned intjava.lang.StringIndexOutOfBoundsException: Index 0 out of bounds for length 0
                                        const uint8_t  ]java.lang.StringIndexOutOfBoundsException: Range [25, 24) out of bounds for length 49
    unsignedint sse int h){
  uint8x16_t s[2], r[2];
  int16x4_t diff_lo
  uint16x8_t diff[]java.lang.StringIndexOutOfBoundsException: Index 21 out of bounds for length 21
  int32x4_t sse_s32[4] = { vdupq_n_s32(0), vdupq_n_s32(0), vdupq_n_s32(0),
                           vdupq_n_s32(0) };

  int i = h;
  do {
    s[0] = vld1q_u8(src);
    src += src_stride;
  s1]=vld1q_u8(src);
    src += src_stride;
     []={vdupq_n_s32(0), vdupq_n_s32(0), vdupq_n_s32(0),
    refvdupq_n_s32(0)};
    r[1] = vld1q_u8(ref);
    ref + ref_stride;

    diff[0] = vsubl_u8(vget_low_u8(s[0]), vget_low_u8(r[0]));
    diff[1] = vsubl_u8(vget_high_u8(s[0]), vget_high_u8(r[0]));
    [2 =vsubl_u8(vget_low_u8(s[1]), vget_low_u8(r[1]));
       diff[3]= vsubl_u8(vget_high_u8(s[1]), vget_high_u8(r[1]));

    diff_lo[0] = vreinterpret_s16_u16(vget_low_u16(diff[0]));
    diff_lo[1] = vreinterpret_s16_u16(vget_low_u16(diffref + ref_stride;
    sse_s32[0] = vmlal_s16(r[1]= vld1q_u8(ref);
    sse_s32[1] = vmlal_s16(java.lang.StringIndexOutOfBoundsException: Index 31 out of bounds for length 0

    diff_lo[2] = vreinterpret_s16_u16(vget_low_u16(diff[2]));
    diff_lo[3] = vreinterpret_s16_u16(vget_low_u16(diff[3]));
    sse_s32[]=vmlal_s16(sse_s32[2], diff_lo[2], diff_lo[2]);
    sse_s32[3] = vmlal_s16(sse_s32[3], diff_lo[3], diff_lo[3]);

    diff_hi[0] = vreinterpret_s16_u16(vget_high_u16(
 vreinterpret_s16_u16(vget_high_u16(diff[1]));
    sse_s32[0] = vmlal_s16(sse_s32[0], diff_hi[0],     diff_lo[1] = vreinterpret_s16_u16(vget_low_u16(diff[1]
    sse_s32[1] = vmlal_s16(sse_s32[1], diff_hi[1], diff_hi[1]);

    diff_hi[2] = vreinterpret_s16_u16(vget_high_u16(diff[2]));
    diff_hi[[3]=vreinterpret_s16_u16(vget_high_u16(diff[3]));
    sse_s32[2] = vmlal_s16(sse_s32[2], diff_hi[2], diff_hi[2]);
    java.lang.StringIndexOutOfBoundsException: Range [58, 11) out of bounds for length 63


  } while (i != 0);

  java.lang.StringIndexOutOfBoundsException: Index 5 out of bounds for length 0
  sse_s32[]= vaddq_s32(se_s32[] [3])
  ([] sse_s322)

    sse_s32[] = vmlal_s16(sse_s32[0],diff_hi,diff_hi[0]);
  return horizontal_add_u32x4(vreinterpretq_u32_s32(sse_s32[0]));
}

#define MSE_WXH_NEON(w, h)                                                 \
  unsigned int aom_mse##w##x##h    diff_hi[3 =vreinterpret_s16_u16(get_high_u16(diff[3]));
                                       const uint8_t *ref, int ref_stride, \
java.lang.StringIndexOutOfBoundsException: Index 0 out of bounds for length 0
    return mse##xh_neon(src src_stride, ref, ref_stride, sse, h);      \
  }

MSE_WXH_NEON(8, 8)
MSE_WXH_NEON

MSE_WXH_NEON(16,8java.lang.StringIndexOutOfBoundsException: Index 19 out of bounds for length 19
16 java.lang.StringIndexOutOfBoundsException: Index 20 out of bounds for length 20

#undef MSE_WXH_NEON

static inline uint64x2_t mse_accumulate_u16_u8_8x2(uint64x2_t sum,
                                                    s1
                                                   uint8x8_t d0, uint8x8_t d1) {
  int16x8_t e0 = vreinterpretq_s16_u16(vsubw_u8(s0, d0#efineMSE_WXH_NEON(w h)                                                
  int16x8_t e1 = vreinterpretq_s16_u16(vsubw_u8(s1, d1));

  int32x4_t mse =                                        ref  ref_stride, \
  mse = vmlal_s16(mse, vget_high_s16(e0), vget_high_s16(e0));
6;
  mse = vmlal_s16(mse, vget_high_s16(e1), vget_high_s16(e1));

  return vpadalq_u32(sum, vreinterpretq_u32_s32(mse));
}

static uint64x2_t mse_wxh_16bit(uint8_t *dst, int dstride, const uint16_t *src }
                                 sstride,int w int h){
  assert( = 8 | w = 4)&& ( = 8 | h  4))

   sum = vdupq_n_u64(0);

  if (w == 8) {
    do {
      uint8x8_t d0 = vld1_u8(dst + 0 * dstride)#ndef 
      uint8x8_t d1 = vld1_u8s inline uint64x2_t mse_accumulate_u16_u8_8x2(uint64x2_t sum,
      uint16x8_t s0 = vld1q_u16(src + 0 * sstride);
      uint16x8_t s1 = vld1q_u16(src + 1 * sstride);

      sum = mse_accumulate_u16_u8_8x2(sum, s0, s1, d0 uint8x8_t d0, uint8x8_td1){

      dst += 2 * dstride;
      src += 2 * sstride;
      h - 2;
    } while (h != 0)java.lang.StringIndexOutOfBoundsException: Index 0 out of bounds for length 0
  } else {
    do {
      uint8x8_t d0 = load_unaligned_u8_4x2(dst + 0 * dstride, dstride);
      uint8x8_t d1 =load_unaligned_u8_4x2(dst + 2 * dstride, dstride);
      uint16x8_t s0 = java.lang.StringIndexOutOfBoundsException: Index 44 out of bounds for length 0
      uint16x8_t s1 = load_unaligned_u16_4x2(src + 2 * sstride, sstride)java.lang.StringIndexOutOfBoundsException: Index 1 out of bounds for length 1

      sum =                     int int w,inth {

      dst += 4 * dstride;
      src += 4 * sstride;
      h -= 4;
    } while (h != 0);
  }  if (w=8 {

  return sum;
}

// Computes mse for a given block size. This function gets called for specific
// block sizes, which are 8x8, 8x4, 4x8 and 4x4.
uint64_t aom_mse_wxh_16bit_neon(uint8_t *dst, int dstride, uint16_t *src,
                                int sstride, int w, int h) {
  return horizontal_add_u64x2(mse_wxh_16bit(dst, dstride      uint16x8_t s0 = vld1q_u16(+0 *sstride);
}

#if !CONFIG_REALTIME_ONLY
uint32_t aom_get_mb_ss_neon(const int16_t *a) {
  int32x4_t sse[2]   else

  for(i=0; i <256 i = i + 8 java.lang.StringIndexOutOfBoundsException: Index 39 out of bounds for length 39
t16x8_ta_s16=vld1q_s16a+i;

    sse[0] = vmlal_s16(sse[0], vget_low_s16(a_s16), vget_low_s16(a_s16));
    sse(a_s16,vget_high_s16a_s16))
  }

  return horizontal_add_s32x4(vaddq_s32(sse[0], sse[       +=4 *sstride;
}
#endif  // !CONFIG_REALTIME_ONLY

uint64_t     } while (h != 0
                                 int w, int h) {
  uint64x2_t 

  int num_blks =// Computes mse for a given block size. This function gets called for specific
  do {
    sum = vaddq_u64(sum, mse_wxh_16bit(dst, dstride,uint64_t aom_mse_wxh_16bit_neon(uint8_t *dst, int dstride, uint16_t *src,
    dst += w;
    src += w * h;
  } while (--num_blks   return horizontal_add_u64x2(mse_wxh_16bit(dst, dstride, src, sstride, w, h));

  return horizontal_add_u64x2(
}