| /external/libgav1/src/dsp/x86/ |
| D | intrapred_smooth_sse4.cc | 103 const __m128i& weights_y, in WriteSmoothPredSum4() argument
109 const __m128i weight_y = _mm_shuffle_epi32(weights_y, y_mask); in WriteSmoothPredSum4()
172 inline void WriteSmoothPred4x8(const __m128i* pixel, const __m128i* weights_y, in WriteSmoothPred4x8() argument
186 const __m128i weight_y = _mm_shuffle_epi8(weights_y[0], y_select); in WriteSmoothPred4x8()
187 const __m128i inverted_weight_y = _mm_shuffle_epi8(weights_y[1], y_select); in WriteSmoothPred4x8()
249 __m128i weights_y[2]; in Smooth4x8_SSE4_1() local
250 LoadSmoothWeights4(kSmoothWeights, 8, weights_y, weights_x); in Smooth4x8_SSE4_1()
254 WriteSmoothPred4x8(pixels, weights_y, weights_x, dst, stride, false); in Smooth4x8_SSE4_1()
264 __m128i weights_y[4]; in Smooth4x16_SSE4_1() local
265 LoadSmoothWeights4(kSmoothWeights, 16, weights_y, weights_x); in Smooth4x16_SSE4_1()
[all …]
|
| /external/libaom/aom_dsp/x86/ |
| D | intrapred_ssse3.c | 955 const __m128i weights_y = _mm_cvtsi32_si128(sm_weights_h[y]); in smooth_predictor_wxh() local
957 const __m128i scale_m_weights_y = _mm_sub_epi16(scale_value, weights_y); in smooth_predictor_wxh()
961 _mm_shuffle_epi32(_mm_unpacklo_epi16(weights_y, left_y), 0); in smooth_predictor_wxh()
1229 __m128i weights_y = _mm_shuffle_epi8(weights, y_select); in aom_smooth_v_predictor_8x4_ssse3() local
1231 write_smooth_directional_sum8(dst, &top, &weights_y, &scaled_bottom_left_y, in aom_smooth_v_predictor_8x4_ssse3()
1235 weights_y = _mm_shuffle_epi8(weights, y_select); in aom_smooth_v_predictor_8x4_ssse3()
1237 write_smooth_directional_sum8(dst, &top, &weights_y, &scaled_bottom_left_y, in aom_smooth_v_predictor_8x4_ssse3()
1241 weights_y = _mm_shuffle_epi8(weights, y_select); in aom_smooth_v_predictor_8x4_ssse3()
1243 write_smooth_directional_sum8(dst, &top, &weights_y, &scaled_bottom_left_y, in aom_smooth_v_predictor_8x4_ssse3()
1247 weights_y = _mm_shuffle_epi8(weights, y_select); in aom_smooth_v_predictor_8x4_ssse3()
[all …]
|
| /external/libgav1/src/dsp/arm/ |
| D | intrapred_smooth_neon.cc | 58 const uint8_t* const weights_y = kSmoothWeights + height - 4; in Smooth4xN_NEON() local
70 const uint8x8_t weights_y_v = vdup_n_u8(weights_y[y]); in Smooth4xN_NEON()
96 const uint8x8_t scaled_weights_y, const uint8x8_t weights_y) { in CalculateWeightsAndPred() argument
97 const uint16x8_t weighted_top = vmull_u8(weights_y, top); in CalculateWeightsAndPred()
113 const uint8_t* const weights_y = kSmoothWeights + height - 4; in Smooth8xN_NEON() local
125 const uint8x8_t weights_y_v = vdup_n_u8(weights_y[y]); in Smooth8xN_NEON()
138 const uint8x8_t weights_y, const uint8x16_t weights_x, in CalculateWeightsAndPred() argument
141 vmlal_u8(weighted_bl, weights_y, vget_low_u8(top)); in CalculateWeightsAndPred()
149 vmlal_u8(weighted_bl, weights_y, vget_high_u8(top)); in CalculateWeightsAndPred()
172 const uint8_t* const weights_y = kSmoothWeights + height - 4; in Smooth16PlusxN_NEON() local
[all …]
|
| /external/libaom/aom_dsp/arm/ |
| D | highbd_intrapred_neon.c | 393 const uint16_t *const weights_y = smooth_weights_u16 + height - 4; in highbd_smooth_4xh_neon() local
405 vmlal_n_u16(weighted_tr, top_v, weights_y[y]); in highbd_smooth_4xh_neon()
409 vmlal_n_u16(weighted_left, bottom_left_v, 256 - weights_y[y]); in highbd_smooth_4xh_neon()
451 const uint16_t *const weights_y = smooth_weights_u16 + height - 4; in highbd_smooth_8xh_neon() local
465 vmull_n_u16(bottom_left_v, 256 - weights_y[y]); in highbd_smooth_8xh_neon()
471 top_vals, weights_x, left_column[y], weights_y[y]); in highbd_smooth_8xh_neon()
501 const uint16_t *const weights_y = smooth_weights_u16 + height - 4; \
519 vmull_n_u16(bottom_left_v, 256 - weights_y[y]); \
536 left_column[y], weights_y[y]); \
577 const uint16_t *const weights_y = smooth_weights_u16 + height - 4; in highbd_smooth_v_4xh_neon() local
[all …]
|
| D | intrapred_neon.c | 2664 const uint8_t *const weights_y = smooth_weights + height - 4; in smooth_4xh_neon() local
2679 const uint8x8_t weights_y_v = vdup_n_u8(weights_y[y]); in smooth_4xh_neon()
2705 const uint8x8_t scaled_weights_y, const uint8x8_t weights_y) { in calculate_weights_and_pred() argument
2706 const uint16x8_t weighted_top = vmull_u8(weights_y, top); in calculate_weights_and_pred()
2719 const uint8_t *const weights_y = smooth_weights + height - 4; in smooth_8xh_neon() local
2732 const uint8x8_t weights_y_v = vdup_n_u8(weights_y[y]); in smooth_8xh_neon()
2762 const uint8x8_t weights_y, const uint8x16_t weights_x, in calculate_weights_and_predq() argument
2765 vmlal_u8(weighted_bl, weights_y, vget_low_u8(top)); in calculate_weights_and_predq()
2773 vmlal_u8(weighted_bl, weights_y, vget_high_u8(top)); in calculate_weights_and_predq()
2795 const uint8_t *const weights_y = smooth_weights + height - 4; \
[all …]
|
| /external/libgav1/src/dsp/ |
| D | intrapred_smooth.cc | 62 const uint8_t* const weights_y = kSmoothWeights + block_height - 4; in Smooth() local
69 assert(scale_value >= weights_y[y] && scale_value >= weights_x[x]); in Smooth()
70 uint32_t pred = weights_y[y] * top[x]; in Smooth()
72 pred += static_cast<uint8_t>(scale_value - weights_y[y]) * bottom_left; in Smooth()
94 const uint8_t* const weights_y = kSmoothWeights + block_height - 4; in SmoothVertical() local
101 assert(scale_value >= weights_y[y]); in SmoothVertical()
102 uint32_t pred = weights_y[y] * top[x]; in SmoothVertical()
103 pred += static_cast<uint8_t>(scale_value - weights_y[y]) * bottom_left; in SmoothVertical()
|
| /external/ComputeLibrary/src/core/CL/cl_kernels/nhwc/ |
| D | transposed_convolution.cl | 166 const int weights_y = cout * _IY_MULTIPLIER + yk * _IWEI_WIDTH + xk;
203 …T_LOAD(WEI_DATA_TYPE, N0, K0, WEI_TENSOR_TYPE, wei, ck, weights_y, _IY_MULTIPLIER, wei_stride_y, b…
235 … T_LOAD(WEI_DATA_TYPE, N0, 1, BUFFER, wei, ck, weights_y, _IY_MULTIPLIER, wei_stride_y, b);
|
| /external/ComputeLibrary/cl_kernels/nhwc/ |
| D | transposed_convolution.clembed | 2179 const int weights_y = cout * _IY_MULTIPLIER + yk * _IWEI_WIDTH + xk;
2216 …T_LOAD(WEI_DATA_TYPE, N0, K0, WEI_TENSOR_TYPE, wei, ck, weights_y, _IY_MULTIPLIER, wei_stride_y, b…
2248 … T_LOAD(WEI_DATA_TYPE, N0, 1, BUFFER, wei, ck, weights_y, _IY_MULTIPLIER, wei_stride_y, b);
|