| /external/libjpeg-turbo/simd/arm/ |
| D | jidctfst-neon.c | 132 int16x4_t tmp10 = vadd_s16(tmp0, tmp2); /* phase 3 */ in jsimd_idct_ifast_neon()
135 int16x4_t tmp13 = vadd_s16(tmp1, tmp3); /* phases 5-3 */ in jsimd_idct_ifast_neon()
138 tmp12 = vadd_s16(tmp12, tmp1_sub_tmp3); in jsimd_idct_ifast_neon()
141 tmp0 = vadd_s16(tmp10, tmp13); /* phase 2 */ in jsimd_idct_ifast_neon()
143 tmp1 = vadd_s16(tmp11, tmp12); in jsimd_idct_ifast_neon()
152 int16x4_t z13 = vadd_s16(tmp6, tmp5); /* phase 6 */ in jsimd_idct_ifast_neon()
154 int16x4_t z11 = vadd_s16(tmp4, tmp7); in jsimd_idct_ifast_neon()
157 tmp7 = vadd_s16(z11, z13); /* phase 5 */ in jsimd_idct_ifast_neon()
160 tmp11 = vadd_s16(tmp11, z11_sub_z13); in jsimd_idct_ifast_neon()
164 z5 = vadd_s16(z5, z10_add_z12); in jsimd_idct_ifast_neon()
[all …]
|
| D | jidctint-neon.c | 379 int32x4_t tmp0 = vshll_n_s16(vadd_s16(z2_s16, z3_s16), CONST_BITS); in jsimd_idct_islow_pass1_regular()
393 z3_s16 = vadd_s16(tmp0_s16, tmp2_s16); in jsimd_idct_islow_pass1_regular()
394 int16x4_t z4_s16 = vadd_s16(tmp1_s16, tmp3_s16); in jsimd_idct_islow_pass1_regular()
590 int32x4_t tmp0 = vshll_n_s16(vadd_s16(z2_s16, z3_s16), CONST_BITS); in jsimd_idct_islow_pass2_regular()
604 z3_s16 = vadd_s16(tmp0_s16, tmp2_s16); in jsimd_idct_islow_pass2_regular()
605 int16x4_t z4_s16 = vadd_s16(tmp1_s16, tmp3_s16); in jsimd_idct_islow_pass2_regular()
|
| /external/libvpx/libvpx/vpx_dsp/arm/ |
| D | idct16x16_add_neon.c | 504 step1[8] = vadd_s16(step2[8], step2[11]); in vpx_idct16x16_10_add_half1d_pass1()
505 step1[9] = vadd_s16(step2[9], step2[10]); in vpx_idct16x16_10_add_half1d_pass1()
510 step1[14] = vadd_s16(step2[14], step2[13]); in vpx_idct16x16_10_add_half1d_pass1()
511 step1[15] = vadd_s16(step2[15], step2[12]); in vpx_idct16x16_10_add_half1d_pass1()
514 step2[0] = vadd_s16(step1[0], step1[7]); in vpx_idct16x16_10_add_half1d_pass1()
515 step2[1] = vadd_s16(step1[1], step1[6]); in vpx_idct16x16_10_add_half1d_pass1()
516 step2[2] = vadd_s16(step1[2], step1[5]); in vpx_idct16x16_10_add_half1d_pass1()
517 step2[3] = vadd_s16(step1[3], step1[4]); in vpx_idct16x16_10_add_half1d_pass1()
532 out[0] = vadd_s16(step2[0], step2[15]); in vpx_idct16x16_10_add_half1d_pass1()
533 out[1] = vadd_s16(step2[1], step2[14]); in vpx_idct16x16_10_add_half1d_pass1()
[all …]
|
| D | idct_neon.h | 339 step2[4] = vadd_s16(step1[4], step1[5]); in idct8x8_12_pass1_bd8()
342 step2[7] = vadd_s16(step1[7], step1[6]); in idct8x8_12_pass1_bd8()
345 step1[0] = vadd_s16(step2[1], step2[3]); in idct8x8_12_pass1_bd8()
346 step1[1] = vadd_s16(step2[1], step2[2]); in idct8x8_12_pass1_bd8()
357 io[0] = vadd_s16(step1[0], step2[7]); in idct8x8_12_pass1_bd8()
358 io[1] = vadd_s16(step1[1], step1[6]); in idct8x8_12_pass1_bd8()
359 io[2] = vadd_s16(step1[2], step1[5]); in idct8x8_12_pass1_bd8()
360 io[3] = vadd_s16(step1[3], step2[4]); in idct8x8_12_pass1_bd8()
|
| D | deblock_neon.c | 266 *sum = vadd_s16(*sum, x); in accumulate_sum_sumsq()
270 *sum = vadd_s16(*sum, vext_s16(zero, x, 1)); in accumulate_sum_sumsq()
273 *sum = vadd_s16(*sum, vext_s16(zero, x, 2)); in accumulate_sum_sumsq()
276 *sum = vadd_s16(*sum, vext_s16(zero, x, 3)); in accumulate_sum_sumsq()
|
| D | fdct_neon.c | 33 input_0 = vadd_s16(input_0, one); in vpx_fdct4x4_neon()
|
| /external/libvpx/libvpx/vp8/common/arm/neon/ |
| D | iwalsh_neon.c | 26 d4s16 = vadd_s16(vget_low_s16(q0s16), vget_high_s16(q1s16)); in vp8_short_inv_walsh4x4_neon()
27 d6s16 = vadd_s16(vget_high_s16(q0s16), vget_low_s16(q1s16)); in vp8_short_inv_walsh4x4_neon()
47 d4s16 = vadd_s16(v2tmp0.val[0], v2tmp1.val[1]); in vp8_short_inv_walsh4x4_neon()
48 d6s16 = vadd_s16(v2tmp0.val[1], v2tmp1.val[0]); in vp8_short_inv_walsh4x4_neon()
|
| /external/libvpx/libvpx/vp8/encoder/arm/neon/ |
| D | shortfdct_neon.c | 48 d4s16 = vadd_s16(v2tmp0.val[0], v2tmp1.val[1]); in vp8_short_fdct4x4_neon()
49 d5s16 = vadd_s16(v2tmp0.val[1], v2tmp1.val[0]); in vp8_short_fdct4x4_neon()
58 d0s16 = vadd_s16(d4s16, d5s16); in vp8_short_fdct4x4_neon()
77 d4s16 = vadd_s16(v2tmp0.val[0], v2tmp1.val[1]); in vp8_short_fdct4x4_neon()
78 d5s16 = vadd_s16(v2tmp0.val[1], v2tmp1.val[0]); in vp8_short_fdct4x4_neon()
83 d4s16 = vadd_s16(d4s16, d26s16); in vp8_short_fdct4x4_neon()
85 d0s16 = vadd_s16(d4s16, d5s16); in vp8_short_fdct4x4_neon()
|
| D | vp8_shortwalsh4x4_neon.c | 51 d4s16 = vadd_s16(v2tmp0.val[0], v2tmp1.val[0]); in vp8_short_walsh4x4_neon()
52 d5s16 = vadd_s16(v2tmp0.val[1], v2tmp1.val[1]); in vp8_short_walsh4x4_neon()
64 d0s16 = vadd_s16(d4s16, d5s16); in vp8_short_walsh4x4_neon()
66 d1s16 = vadd_s16(d7s16, d6s16); in vp8_short_walsh4x4_neon()
|
| /external/tensorflow/tensorflow/lite/kernels/internal/optimized/ |
| D | depthwiseconv_uint8.h | 206 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
251 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
281 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
312 const int16x4_t filter = vadd_s16(filter_s16, vdup_n_s16(filter_offset));
352 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
378 const int16x4_t filter = vadd_s16(filter_s16, vdup_n_s16(filter_offset));
448 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
467 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
492 const int16x4_t filter = vadd_s16(filter_s16, vdup_n_s16(filter_offset));
552 const int16x4_t filter = vadd_s16(filter_s16, vdup_n_s16(filter_offset));
[all …]
|
| /external/libaom/libaom/av1/common/arm/ |
| D | convolve_neon.h | 133 temp0 = vadd_s16(s0, s6); in wiener_convolve8_horiz_4x8()
134 temp1 = vadd_s16(s1, s5); in wiener_convolve8_horiz_4x8()
135 temp2 = vadd_s16(s2, s4); in wiener_convolve8_horiz_4x8()
|
| D | selfguided_neon.c | 853 q23 = vadd_s16(d2, d3); in boxsum1()
854 q234 = vadd_s16(q23, d4); in boxsum1()
855 q34 = vadd_s16(d3, d4); in boxsum1()
870 q345 = vadd_s16(d5, q34); in boxsum1()
871 q56 = vadd_s16(d5, d6); in boxsum1()
872 q456 = vadd_s16(d4, q56); in boxsum1()
873 q567 = vadd_s16(d7, q56); in boxsum1()
874 q78 = vadd_s16(d7, d8); in boxsum1()
875 q678 = vadd_s16(d6, q78); in boxsum1()
|
| D | jnt_convolve_neon.c | 997 d0 = vadd_s16(d0, round_offset_vec); in av1_dist_wtd_convolve_x_neon()
1001 d1 = vadd_s16(d1, round_offset_vec); in av1_dist_wtd_convolve_x_neon()
1005 d2 = vadd_s16(d2, round_offset_vec); in av1_dist_wtd_convolve_x_neon()
1009 d3 = vadd_s16(d3, round_offset_vec); in av1_dist_wtd_convolve_x_neon()
1090 d0 = vadd_s16(d0, round_offset_vec); in av1_dist_wtd_convolve_x_neon()
1461 d0 = vadd_s16(d0, round_offset64); in av1_dist_wtd_convolve_y_neon()
1464 d1 = vadd_s16(d1, round_offset64); in av1_dist_wtd_convolve_y_neon()
1467 d2 = vadd_s16(d2, round_offset64); in av1_dist_wtd_convolve_y_neon()
1470 d3 = vadd_s16(d3, round_offset64); in av1_dist_wtd_convolve_y_neon()
1525 d0 = vadd_s16(d0, round_offset64); in av1_dist_wtd_convolve_y_neon()
|
| D | cfl_neon.c | 392 return veor_s16(vadd_s16(a, mask), mask); in CFL_SUB_AVG_FN()
415 return vadd_s16(vsign_s16(scaled_luma, ac_sign), dc); in predict_w4()
|
| D | warp_plane_neon.c | 652 int16x4_t res_low = vadd_s16(tmp16_low, res_sub_const); in av1_warp_affine_neon()
683 int16x4_t res_high = vadd_s16(tmp16_high, res_sub_const); in av1_warp_affine_neon()
|
| /external/tensorflow/tensorflow/lite/kernels/internal/optimized/integer_ops/ |
| D | depthwise_conv.h | 212 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
255 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
284 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
352 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
445 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
463 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
593 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
681 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
758 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
1284 const int16x4_t input = vadd_s16(input_s16, vdup_n_s16(input_offset));
[all …]
|
| /external/libhevc/encoder/arm/ |
| D | ihevce_had_compute_neon.c | 201 int16x4_t q0 = vadd_s16(vget_low_s16(p0), vget_high_s16(p0)); in hadamard_sad8x8_using4x4()
203 int16x4_t q2 = vadd_s16(vget_low_s16(p1), vget_high_s16(p1)); in hadamard_sad8x8_using4x4()
404 h[0] = vadd_s16(vget_low_s16(res_01), vget_high_s16(res_23)); in ihevce_HAD_4x4_8bit_plane_neon()
405 h[1] = vadd_s16(vget_high_s16(res_01), vget_low_s16(res_23)); in ihevce_HAD_4x4_8bit_plane_neon()
409 v[0] = vadd_s16(h[0], h[1]); in ihevce_HAD_4x4_8bit_plane_neon()
410 v[1] = vadd_s16(h[3], h[2]); in ihevce_HAD_4x4_8bit_plane_neon()
|
| /external/webp/src/dsp/ |
| D | enc_neon.c | 316 const int16x4_t tmp0 = vadd_s16(vget_low_s16(a0a1_2), in FTransform_NEON()
334 const int16x4_t a0_k7 = vadd_s16(vget_low_s16(a0a1), vdup_n_s16(7)); in FTransform_NEON()
335 const int16x4_t out0 = vshr_n_s16(vadd_s16(a0_k7, vget_high_s16(a0a1)), 4); in FTransform_NEON()
345 const int16x4_t out1 = vadd_s16(tmp1, a3_eq_0); in FTransform_NEON()
|
| /external/libopus/silk/arm/ |
| D | NSQ_del_dec_neon_intr.c | 719 …q1_Q0_s16x4 = vbsl_s16( lessThanMinusRdo, vadd_s16( q1_Q10_s16x4, vdup_n_s16( rdo_offset ) ), q1_Q… in silk_noise_shape_quantizer_del_dec_neon()
729 … tmp1_s16x4 = vadd_s16( q1_Q10_s16x4, vdup_n_s16( offset_Q10 - QUANT_LEVEL_ADJUST_Q10 ) ); in silk_noise_shape_quantizer_del_dec_neon()
730 … q1_Q10_s16x4 = vadd_s16( q1_Q10_s16x4, vdup_n_s16( offset_Q10 + QUANT_LEVEL_ADJUST_Q10 ) ); in silk_noise_shape_quantizer_del_dec_neon()
734 q2_Q10_s16x4 = vadd_s16( q1_Q10_s16x4, vdup_n_s16( 1024 ) ); in silk_noise_shape_quantizer_del_dec_neon()
|
| /external/libaom/libaom/aom_dsp/arm/ |
| D | fwd_txfm_neon.c | 30 *input_0 = vadd_s16(*input_0, one); in aom_fdct4x4_helper()
|
| /external/libgav1/libgav1/src/dsp/arm/ |
| D | intra_edge_neon.cc | 246 vadd_s16(vget_low_s16(half_sum), vget_high_s16(half_sum)); in IntraEdgeUpsampler_NEON()
|
| D | loop_filter_neon.cc | 102 const int16x4_t plus_four = vadd_s16(a, vdup_n_s16(4)); in Filter4()
103 const int16x4_t plus_three = vadd_s16(a, vdup_n_s16(3)); in Filter4()
|
| D | inverse_transform_neon.cc | 2148 s[0] = vadd_s16(s[0], s[2]); in Wht4_NEON()
2154 s[3] = vadd_s16(s[3], s[2]); in Wht4_NEON()
2162 s[0] = vadd_s16(s[0], s[2]); in Wht4_NEON()
2168 s[3] = vadd_s16(s[3], s[2]); in Wht4_NEON()
|
| D | intrapred_directional_neon.cc | 987 const int16x4_t base_x = vadd_s16(vdup_n_s16(top_base_x), index_offset); in DirectionalZone1_4xH()
|
| /external/libaom/libaom/aom_dsp/simd/ |
| D | v64_intrinsics_arm.h | 228 vadd_s16(vreinterpret_s16_s64(x), vreinterpret_s16_s64(y))); in v64_add_16()
|