Esempi in C++ (Cpp) per vreinterpret_u32_u64

Esempio n. 1

File: vp9_avg_neon.c Progetto: d4rkuzs/ffmpeg-codecs-libs

static INLINE unsigned int horizontal_add_u16x8(const uint16x8_t v_16x8) {
  const uint32x4_t a = vpaddlq_u16(v_16x8);
  const uint64x2_t b = vpaddlq_u32(a);
  const uint32x2_t c = vadd_u32(vreinterpret_u32_u64(vget_low_u64(b)),
                                vreinterpret_u32_u64(vget_high_u64(b)));
  return vget_lane_u32(c, 0);
}

Esempio n. 2

File: intrapred_neon.c Progetto: jmvalin/aom

void vpx_d135_predictor_4x4_neon(uint8_t *dst, ptrdiff_t stride,
                                 const uint8_t *above, const uint8_t *left) {
  const uint8x8_t XABCD_u8 = vld1_u8(above - 1);
  const uint64x1_t XABCD = vreinterpret_u64_u8(XABCD_u8);
  const uint64x1_t ____XABC = vshl_n_u64(XABCD, 32);
  const uint32x2_t zero = vdup_n_u32(0);
  const uint32x2_t IJKL = vld1_lane_u32((const uint32_t *)left, zero, 0);
  const uint8x8_t IJKL_u8 = vreinterpret_u8_u32(IJKL);
  const uint64x1_t LKJI____ = vreinterpret_u64_u8(vrev32_u8(IJKL_u8));
  const uint64x1_t LKJIXABC = vorr_u64(LKJI____, ____XABC);
  const uint8x8_t KJIXABC_ = vreinterpret_u8_u64(vshr_n_u64(LKJIXABC, 8));
  const uint8x8_t JIXABC__ = vreinterpret_u8_u64(vshr_n_u64(LKJIXABC, 16));
  const uint8_t D = vget_lane_u8(XABCD_u8, 4);
  const uint8x8_t JIXABCD_ = vset_lane_u8(D, JIXABC__, 6);
  const uint8x8_t LKJIXABC_u8 = vreinterpret_u8_u64(LKJIXABC);
  const uint8x8_t avg1 = vhadd_u8(JIXABCD_, LKJIXABC_u8);
  const uint8x8_t avg2 = vrhadd_u8(avg1, KJIXABC_);
  const uint64x1_t avg2_u64 = vreinterpret_u64_u8(avg2);
  const uint32x2_t r3 = vreinterpret_u32_u8(avg2);
  const uint32x2_t r2 = vreinterpret_u32_u64(vshr_n_u64(avg2_u64, 8));
  const uint32x2_t r1 = vreinterpret_u32_u64(vshr_n_u64(avg2_u64, 16));
  const uint32x2_t r0 = vreinterpret_u32_u64(vshr_n_u64(avg2_u64, 24));
  vst1_lane_u32((uint32_t *)(dst + 0 * stride), r0, 0);
  vst1_lane_u32((uint32_t *)(dst + 1 * stride), r1, 0);
  vst1_lane_u32((uint32_t *)(dst + 2 * stride), r2, 0);
  vst1_lane_u32((uint32_t *)(dst + 3 * stride), r3, 0);
}

Esempio n. 3

File: vreinterpretu32_u64.c Progetto: Akheon23/chromecast-mirrored-source.toolchain

void test_vreinterpretu32_u64 (void)
{
  uint32x2_t out_uint32x2_t;
  uint64x1_t arg0_uint64x1_t;

  out_uint32x2_t = vreinterpret_u32_u64 (arg0_uint64x1_t);
}

Esempio n. 4

File: intrapred_neon.c Progetto: jmvalin/aom

void vpx_d45_predictor_4x4_neon(uint8_t *dst, ptrdiff_t stride,
                                const uint8_t *above, const uint8_t *left) {
  const uint64x1_t A0 = vreinterpret_u64_u8(vld1_u8(above));  // top row
  const uint64x1_t A1 = vshr_n_u64(A0, 8);
  const uint64x1_t A2 = vshr_n_u64(A0, 16);
  const uint8x8_t ABCDEFGH = vreinterpret_u8_u64(A0);
  const uint8x8_t BCDEFGH0 = vreinterpret_u8_u64(A1);
  const uint8x8_t CDEFGH00 = vreinterpret_u8_u64(A2);
  const uint8x8_t avg1 = vhadd_u8(ABCDEFGH, CDEFGH00);
  const uint8x8_t avg2 = vrhadd_u8(avg1, BCDEFGH0);
  const uint64x1_t avg2_u64 = vreinterpret_u64_u8(avg2);
  const uint32x2_t r0 = vreinterpret_u32_u8(avg2);
  const uint32x2_t r1 = vreinterpret_u32_u64(vshr_n_u64(avg2_u64, 8));
  const uint32x2_t r2 = vreinterpret_u32_u64(vshr_n_u64(avg2_u64, 16));
  const uint32x2_t r3 = vreinterpret_u32_u64(vshr_n_u64(avg2_u64, 24));
  (void)left;
  vst1_lane_u32((uint32_t *)(dst + 0 * stride), r0, 0);
  vst1_lane_u32((uint32_t *)(dst + 1 * stride), r1, 0);
  vst1_lane_u32((uint32_t *)(dst + 2 * stride), r2, 0);
  vst1_lane_u32((uint32_t *)(dst + 3 * stride), r3, 0);
  dst[3 * stride + 3] = above[7];
}