C++ (Cpp) SRARI_H4_SHの例

コード例 #1

ファイル: idct4x4_msa.c プロジェクト: Wafflespeanut/gecko-dev

void aom_idct4x4_16_add_msa(const int16_t *input, uint8_t *dst,
                            int32_t dst_stride) {
  v8i16 in0, in1, in2, in3;

  /* load vector elements of 4x4 block */
  LD4x4_SH(input, in0, in1, in2, in3);
  /* rows */
  TRANSPOSE4x4_SH_SH(in0, in1, in2, in3, in0, in1, in2, in3);
  AOM_IDCT4x4(in0, in1, in2, in3, in0, in1, in2, in3);
  /* columns */
  TRANSPOSE4x4_SH_SH(in0, in1, in2, in3, in0, in1, in2, in3);
  AOM_IDCT4x4(in0, in1, in2, in3, in0, in1, in2, in3);
  /* rounding (add 2^3, divide by 2^4) */
  SRARI_H4_SH(in0, in1, in2, in3, 4);
  ADDBLK_ST4x4_UB(in0, in1, in2, in3, dst, dst_stride);
}

コード例 #2

ファイル: vp9_idct4x4_msa.c プロジェクト: ALEJANDROJ19/VTW-server

void vp9_iht4x4_16_add_msa(const int16_t *input, uint8_t *dst,
                           int32_t dst_stride, int32_t tx_type) {
  v8i16 in0, in1, in2, in3;

  /* load vector elements of 4x4 block */
  LD4x4_SH(input, in0, in1, in2, in3);
  TRANSPOSE4x4_SH_SH(in0, in1, in2, in3, in0, in1, in2, in3);

  switch (tx_type) {
    case DCT_DCT:
      /* DCT in horizontal */
      VP9_IDCT4x4(in0, in1, in2, in3, in0, in1, in2, in3);
      /* DCT in vertical */
      TRANSPOSE4x4_SH_SH(in0, in1, in2, in3, in0, in1, in2, in3);
      VP9_IDCT4x4(in0, in1, in2, in3, in0, in1, in2, in3);
      break;
    case ADST_DCT:
      /* DCT in horizontal */
      VP9_IDCT4x4(in0, in1, in2, in3, in0, in1, in2, in3);
      /* ADST in vertical */
      TRANSPOSE4x4_SH_SH(in0, in1, in2, in3, in0, in1, in2, in3);
      VP9_IADST4x4(in0, in1, in2, in3, in0, in1, in2, in3);
      break;
    case DCT_ADST:
      /* ADST in horizontal */
      VP9_IADST4x4(in0, in1, in2, in3, in0, in1, in2, in3);
      /* DCT in vertical */
      TRANSPOSE4x4_SH_SH(in0, in1, in2, in3, in0, in1, in2, in3);
      VP9_IDCT4x4(in0, in1, in2, in3, in0, in1, in2, in3);
      break;
    case ADST_ADST:
      /* ADST in horizontal */
      VP9_IADST4x4(in0, in1, in2, in3, in0, in1, in2, in3);
      /* ADST in vertical */
      TRANSPOSE4x4_SH_SH(in0, in1, in2, in3, in0, in1, in2, in3);
      VP9_IADST4x4(in0, in1, in2, in3, in0, in1, in2, in3);
      break;
    default:
      assert(0);
      break;
  }

  /* final rounding (add 2^3, divide by 2^4) and shift */
  SRARI_H4_SH(in0, in1, in2, in3, 4);
  /* add block and store 4x4 */
  ADDBLK_ST4x4_UB(in0, in1, in2, in3, dst, dst_stride);
}

コード例 #3

ファイル: dct-c.c プロジェクト: 0x0B501E7E/x264

static void avc_idct4x4_addblk_msa( uint8_t *p_dst, int16_t *p_src,
                                    int32_t i_dst_stride )
{
    v8i16 src0, src1, src2, src3;
    v8i16 hres0, hres1, hres2, hres3;
    v8i16 vres0, vres1, vres2, vres3;
    v8i16 zeros = { 0 };

    LD4x4_SH( p_src, src0, src1, src2, src3 );
    AVC_ITRANS_H( src0, src1, src2, src3, hres0, hres1, hres2, hres3 );
    TRANSPOSE4x4_SH_SH( hres0, hres1, hres2, hres3,
                        hres0, hres1, hres2, hres3 );
    AVC_ITRANS_H( hres0, hres1, hres2, hres3, vres0, vres1, vres2, vres3 );
    SRARI_H4_SH( vres0, vres1, vres2, vres3, 6 );
    ADDBLK_ST4x4_UB( vres0, vres1, vres2, vres3, p_dst, i_dst_stride );
    ST_SH2( zeros, zeros, p_src, 8 );
}

コード例 #4

ファイル: idct8x8_msa.c プロジェクト: Crawping/chromium_extract

void vp10_iht8x8_64_add_msa(const int16_t *input, uint8_t *dst,
                           int32_t dst_stride, int32_t tx_type) {
  v8i16 in0, in1, in2, in3, in4, in5, in6, in7;

  /* load vector elements of 8x8 block */
  LD_SH8(input, 8, in0, in1, in2, in3, in4, in5, in6, in7);

  TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7,
                     in0, in1, in2, in3, in4, in5, in6, in7);

  switch (tx_type) {
    case DCT_DCT:
      /* DCT in horizontal */
      VP9_IDCT8x8_1D(in0, in1, in2, in3, in4, in5, in6, in7,
                     in0, in1, in2, in3, in4, in5, in6, in7);
      /* DCT in vertical */
      TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7,
                         in0, in1, in2, in3, in4, in5, in6, in7);
      VP9_IDCT8x8_1D(in0, in1, in2, in3, in4, in5, in6, in7,
                     in0, in1, in2, in3, in4, in5, in6, in7);
      break;
    case ADST_DCT:
      /* DCT in horizontal */
      VP9_IDCT8x8_1D(in0, in1, in2, in3, in4, in5, in6, in7,
                     in0, in1, in2, in3, in4, in5, in6, in7);
      /* ADST in vertical */
      TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7,
                         in0, in1, in2, in3, in4, in5, in6, in7);
      VP9_ADST8(in0, in1, in2, in3, in4, in5, in6, in7,
                in0, in1, in2, in3, in4, in5, in6, in7);
      break;
    case DCT_ADST:
      /* ADST in horizontal */
      VP9_ADST8(in0, in1, in2, in3, in4, in5, in6, in7,
                in0, in1, in2, in3, in4, in5, in6, in7);
      /* DCT in vertical */
      TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7,
                         in0, in1, in2, in3, in4, in5, in6, in7);
      VP9_IDCT8x8_1D(in0, in1, in2, in3, in4, in5, in6, in7,
                     in0, in1, in2, in3, in4, in5, in6, in7);
      break;
    case ADST_ADST:
      /* ADST in horizontal */
      VP9_ADST8(in0, in1, in2, in3, in4, in5, in6, in7,
                in0, in1, in2, in3, in4, in5, in6, in7);
      /* ADST in vertical */
      TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7,
                         in0, in1, in2, in3, in4, in5, in6, in7);
      VP9_ADST8(in0, in1, in2, in3, in4, in5, in6, in7,
                in0, in1, in2, in3, in4, in5, in6, in7);
      break;
    default:
      assert(0);
      break;
  }

  /* final rounding (add 2^4, divide by 2^5) and shift */
  SRARI_H4_SH(in0, in1, in2, in3, 5);
  SRARI_H4_SH(in4, in5, in6, in7, 5);

  /* add block and store 8x8 */
  VP9_ADDBLK_ST8x4_UB(dst, dst_stride, in0, in1, in2, in3);
  dst += (4 * dst_stride);
  VP9_ADDBLK_ST8x4_UB(dst, dst_stride, in4, in5, in6, in7);
}