Home
last modified time | relevance | path

Searched refs:vdupq_n_s32 (Results 1 – 25 of 73) sorted by relevance

123

/external/libopus/silk/fixed/arm/
Dwarped_autocorrelation_FIX_neon_intr.c93 vst1q_s32( input_QS, vdupq_n_s32( 0 ) ); in silk_warped_autocorrelation_FIX_neon()
95 vst1q_s32( input_QS, vdupq_n_s32( 0 ) ); in silk_warped_autocorrelation_FIX_neon()
97 vst1q_s32( input_QS, vdupq_n_s32( 0 ) ); in silk_warped_autocorrelation_FIX_neon()
99 vst1q_s32( input_QS, vdupq_n_s32( 0 ) ); in silk_warped_autocorrelation_FIX_neon()
101 vst1q_s32( input_QS, vdupq_n_s32( 0 ) ); in silk_warped_autocorrelation_FIX_neon()
103 vst1q_s32( input_QS, vdupq_n_s32( 0 ) ); in silk_warped_autocorrelation_FIX_neon()
115 vst1q_s32( input_QS, vdupq_n_s32( 0 ) ); in silk_warped_autocorrelation_FIX_neon()
117 vst1q_s32( input_QS, vdupq_n_s32( 0 ) ); in silk_warped_autocorrelation_FIX_neon()
119 vst1q_s32( input_QS, vdupq_n_s32( 0 ) ); in silk_warped_autocorrelation_FIX_neon()
121 vst1q_s32( input_QS, vdupq_n_s32( 0 ) ); in silk_warped_autocorrelation_FIX_neon()
[all …]
/external/libaom/libaom/av1/common/arm/
Dconvolve_neon.h30 const int32x4_t round_bits = vdupq_n_s32(-round1_bits); in wiener_convolve8_vert_4x8()
31 const int32x4_t zero = vdupq_n_s32(0); in wiener_convolve8_vert_4x8()
32 const int32x4_t round_vec = vdupq_n_s32(round_const); in wiener_convolve8_vert_4x8()
79 const int32x4_t round_bits = vdupq_n_s32(-round0_bits); in wiener_convolve8_horiz_8x8()
81 const int32x4_t round_vec_0 = vdupq_n_s32(round_const_0); in wiener_convolve8_horiz_8x8()
82 const int32x4_t round_vec_1 = vdupq_n_s32(round_const_1); in wiener_convolve8_horiz_8x8()
128 const int32x4_t round_bits = vdupq_n_s32(-round0_bits); in wiener_convolve8_horiz_4x8()
129 const int32x4_t zero = vdupq_n_s32(0); in wiener_convolve8_horiz_4x8()
130 const int32x4_t round_vec_0 = vdupq_n_s32(round_const_0); in wiener_convolve8_horiz_4x8()
131 const int32x4_t round_vec_1 = vdupq_n_s32(round_const_1); in wiener_convolve8_horiz_4x8()
[all …]
Dav1_txfm_neon.c23 const int32x4_t dup_bits_n_32x4 = vdupq_n_s32((int32_t)(-bit)); in av1_round_shift_array_neon()
/external/skqp/src/core/
DSkBitmapProcState_matrixProcs.cpp223 int32x4_t vdx8 = vdupq_n_s32(dx8); in decal_nofilter_scale_neon()
227 lbase = vdupq_n_s32(fx); in decal_nofilter_scale_neon()
231 hbase = lbase + vdupq_n_s32(4 * dx); in decal_nofilter_scale_neon()
257 int32x4_t vdx8 = vdupq_n_s32(dx8); in decal_filter_scale_neon()
260 wide_fx = vdupq_n_s32(fx); in decal_filter_scale_neon()
265 wide_fx2 = vaddq_s32(wide_fx, vdupq_n_s32(4 * dx)); in decal_filter_scale_neon()
272 wide_out = wide_out | (vshrq_n_s32(wide_fx,16) + vdupq_n_s32(1)); in decal_filter_scale_neon()
275 wide_out2 = wide_out2 | (vshrq_n_s32(wide_fx2,16) + vdupq_n_s32(1)); in decal_filter_scale_neon()
325 res = vmaxq_s32(res, vdupq_n_s32(0)); in clamp4()
326 res = vminq_s32(res, vdupq_n_s32(max)); in clamp4()
[all …]
/external/libaom/libaom/aom_dsp/arm/
Dvariance_neon.c26 int32x4_t v_sse_lo = vdupq_n_s32(0); in variance_neon_w8()
27 int32x4_t v_sse_hi = vdupq_n_s32(0); in variance_neon_w8()
164 q8s32 = vdupq_n_s32(0); in aom_variance16x8_neon()
165 q9s32 = vdupq_n_s32(0); in aom_variance16x8_neon()
166 q10s32 = vdupq_n_s32(0); in aom_variance16x8_neon()
240 q8s32 = vdupq_n_s32(0); in aom_variance8x16_neon()
241 q9s32 = vdupq_n_s32(0); in aom_variance8x16_neon()
242 q10s32 = vdupq_n_s32(0); in aom_variance8x16_neon()
300 q7s32 = vdupq_n_s32(0); in aom_mse16x16_neon()
301 q8s32 = vdupq_n_s32(0); in aom_mse16x16_neon()
[all …]
/external/libvpx/libvpx/vpx_dsp/arm/
Dvariance_neon.c35 int32x4_t sse_lo_s32 = vdupq_n_s32(0); in variance_neon_w4x4()
36 int32x4_t sse_hi_s32 = vdupq_n_s32(0); in variance_neon_w4x4()
81 int32x4_t sse_lo_s32 = vdupq_n_s32(0); in variance_neon_w16()
82 int32x4_t sse_hi_s32 = vdupq_n_s32(0); in variance_neon_w16()
130 int32x4_t sse_lo_s32 = vdupq_n_s32(0); in variance_neon_w8x2()
131 int32x4_t sse_hi_s32 = vdupq_n_s32(0); in variance_neon_w8x2()
278 q7s32 = vdupq_n_s32(0); in vpx_mse16x16_neon()
279 q8s32 = vdupq_n_s32(0); in vpx_mse16x16_neon()
280 q9s32 = vdupq_n_s32(0); in vpx_mse16x16_neon()
281 q10s32 = vdupq_n_s32(0); in vpx_mse16x16_neon()
Ddeblock_neon.c263 const int32x4_t zeroq = vdupq_n_s32(0); in accumulate_sum_sumsq()
292 const int32x4_t fifteen = vdupq_n_s32(15); in combine_mask()
310 const int32x4_t f = vdupq_n_s32(flimit); in vpx_mbpost_proc_across_ip_neon()
333 sumsq_high = vdupq_n_s32(sumsq); in vpx_mbpost_proc_across_ip_neon()
398 const int32x4_t f = vdupq_n_s32(flimit); in vpx_mbpost_proc_down_neon()
/external/tensorflow/tensorflow/lite/kernels/internal/optimized/integer_ops/
Dmean.h51 const int32x4_t bias_dup = vdupq_n_s32(bias); in MeanImpl()
52 const int32x4_t min_dup = vdupq_n_s32(kMinValue); in MeanImpl()
53 const int32x4_t max_dup = vdupq_n_s32(kMaxValue); in MeanImpl()
61 temp_sum.val[0] = vdupq_n_s32(0); in MeanImpl()
62 temp_sum.val[1] = vdupq_n_s32(0); in MeanImpl()
63 temp_sum.val[2] = vdupq_n_s32(0); in MeanImpl()
64 temp_sum.val[3] = vdupq_n_s32(0); in MeanImpl()
Dadd.h60 const int32x4_t left_shift_dup = vdupq_n_s32(params.left_shift); in AddElementwise()
69 const int32x4_t input1_shift_dup = vdupq_n_s32(params.input1_shift); in AddElementwise()
70 const int32x4_t input2_shift_dup = vdupq_n_s32(params.input2_shift); in AddElementwise()
133 const int32x4_t left_shift_dup = vdupq_n_s32(params.left_shift); in AddScalarBroadcast()
152 const int32x4_t input1_shift_dup = vdupq_n_s32(params.input1_shift); in AddScalarBroadcast()
169 const int32x4_t input2_shift_dup = vdupq_n_s32(params.input2_shift); in AddScalarBroadcast()
/external/webrtc/webrtc/common_audio/signal_processing/
Ddownsample_fast_neon.c44 int32x4_t out32x4_0 = vdupq_n_s32(2048); in WebRtcSpl_DownsampleFastNeon()
45 int32x4_t out32x4_1 = vdupq_n_s32(2048); in WebRtcSpl_DownsampleFastNeon()
113 int32x4_t out32x4_0 = vdupq_n_s32(2048); in WebRtcSpl_DownsampleFastNeon()
114 int32x4_t out32x4_1 = vdupq_n_s32(2048); in WebRtcSpl_DownsampleFastNeon()
174 int32x4_t out32x4_0 = vdupq_n_s32(2048); in WebRtcSpl_DownsampleFastNeon()
175 int32x4_t out32x4_1 = vdupq_n_s32(2048); in WebRtcSpl_DownsampleFastNeon()
Dmin_max_operations_neon.c172 int32x4_t max32x4_0 = vdupq_n_s32(WEBRTC_SPL_WORD32_MIN); in WebRtcSpl_MaxValueW32Neon()
173 int32x4_t max32x4_1 = vdupq_n_s32(WEBRTC_SPL_WORD32_MIN); in WebRtcSpl_MaxValueW32Neon()
252 int32x4_t min32x4_0 = vdupq_n_s32(WEBRTC_SPL_WORD32_MAX); in WebRtcSpl_MinValueW32Neon()
253 int32x4_t min32x4_1 = vdupq_n_s32(WEBRTC_SPL_WORD32_MAX); in WebRtcSpl_MinValueW32Neon()
/external/libvpx/libvpx/vp8/encoder/arm/neon/
Dshortfdct_neon.c26 q9s32 = vdupq_n_s32(14500); in vp8_short_fdct4x4_neon()
27 q10s32 = vdupq_n_s32(7500); in vp8_short_fdct4x4_neon()
28 q11s32 = vdupq_n_s32(12000); in vp8_short_fdct4x4_neon()
29 q12s32 = vdupq_n_s32(51000); in vp8_short_fdct4x4_neon()
126 q9s32 = vdupq_n_s32(14500); in vp8_short_fdct8x4_neon()
127 q10s32 = vdupq_n_s32(7500); in vp8_short_fdct8x4_neon()
187 q9s32 = vdupq_n_s32(12000); in vp8_short_fdct8x4_neon()
188 q10s32 = vdupq_n_s32(51000); in vp8_short_fdct8x4_neon()
Dvp8_shortwalsh4x4_neon.c33 qEmptys32 = vdupq_n_s32(0); in vp8_short_walsh4x4_neon()
34 q15s32 = vdupq_n_s32(3); in vp8_short_walsh4x4_neon()
/external/libhevc/common/arm/
Dihevc_quant_iquant_ssd_neon_intr.c101 int32x4_t ssd0 = vdupq_n_s32(0); in ihevc_quant_iquant_ssd_flat_scale_mat_neon()
129 int32x4_t add_iq = vdupq_n_s32(1); in ihevc_quant_iquant_ssd_flat_scale_mat_neon()
145 q_v_bits = vdupq_n_s32(-q_bits); in ihevc_quant_iquant_ssd_flat_scale_mat_neon()
146 add_q = vdupq_n_s32(temp); in ihevc_quant_iquant_ssd_flat_scale_mat_neon()
151 sh_iq_1 = vdupq_n_s32(sh_tmp); in ihevc_quant_iquant_ssd_flat_scale_mat_neon()
155 sh_iq = vdupq_n_s32(s_iq); in ihevc_quant_iquant_ssd_flat_scale_mat_neon()
375 int32x4_t ssd0 = vdupq_n_s32(0); in ihevc_q_iq_ssd_flat_scale_mat_var_rnd_fact_neon()
413 int32x4_t add_iq = vdupq_n_s32(1); in ihevc_q_iq_ssd_flat_scale_mat_var_rnd_fact_neon()
427 stmp = vdupq_n_s32(q_bits - QUANT_ROUND_FACTOR_Q); in ihevc_q_iq_ssd_flat_scale_mat_var_rnd_fact_neon()
429 add_q = vdupq_n_s32((1 << QUANT_ROUND_FACTOR_Q) / 2); in ihevc_q_iq_ssd_flat_scale_mat_var_rnd_fact_neon()
[all …]
/external/gemmlowp/fixedpoint/
Dfixedpoint_neon.h69 return veorq_s32(a, vdupq_n_s32(-1));
109 return vshlq_s32(a, vdupq_n_s32(offset));
129 return vshlq_s32(a, vdupq_n_s32(-offset));
171 return MaskIfEqual(a, vdupq_n_s32(0));
281 const int32x4_t shift_vec = vdupq_n_s32(-exponent);
341 return vdupq_n_s32(x);
/external/webrtc/webrtc/modules/audio_coding/codecs/isac/fix/source/
Dentropy_coding_neon.c44 int32x4_t shift32x4 = vdupq_n_s32(shift); in WebRtcIsacfix_MatrixProduct1Neon()
46 int32x4_t sum_32x4 = vdupq_n_s32(0); in WebRtcIsacfix_MatrixProduct1Neon()
61 vdupq_n_s32((int32_t)(matrix0[matrix0_index]) << 15); in WebRtcIsacfix_MatrixProduct1Neon()
102 int32x4_t matrix1_32x4 = vdupq_n_s32(matrix1[matrix1_index] << shift); in WebRtcIsacfix_MatrixProduct1Neon()
Dlattice_neon.c36 int32x4_t input0_v = vdupq_n_s32((int32_t)input0 << 16); in WebRtcIsacfix_FilterMaLoopNeon()
37 int32x4_t input1_v = vdupq_n_s32((int32_t)input1 << 16); in WebRtcIsacfix_FilterMaLoopNeon()
38 int32x4_t input2_v = vdupq_n_s32(input2); in WebRtcIsacfix_FilterMaLoopNeon()
Dtransform_neon.c35 int32x4_t factq = vdupq_n_s32(fact); in ComplexMulAndFindMaxNeon()
110 int32x4_t sh32x4 = vdupq_n_s32(sh); in PreShiftW32toW16Neon()
159 int32x4_t shift = vdupq_n_s32(-sh - 23); in PostShiftAndSeparateNeon()
357 int32x4_t shift = vdupq_n_s32(-sh - 16); in PostShiftAndDivideAndDemodulateNeon()
365 int32x4_t fact = vdupq_n_s32(factQ19); in PostShiftAndDivideAndDemodulateNeon()
/external/libopus/silk/arm/
DNSQ_del_dec_neon_intr.c189 const int32x4_t shift_s32x4 = vdupq_n_s32( -shift ); in copy_winner_state()
192 t0_s32x4 = t1_s32x4 = vdupq_n_s32( 0 ); /* initialization */ in copy_winner_state()
284 vst1q_s32( psDelDec->RD_Q10, vdupq_n_s32( 0 ) ); in silk_NSQ_del_dec_neon()
363 … RD_Q10_s32x4 = vaddq_s32( RD_Q10_s32x4, vdupq_n_s32( silk_int32_MAX >> 4 ) ); in silk_NSQ_del_dec_neon()
412 t_s32x4 = vdupq_n_s32( 0 ); /* initialization */ in silk_NSQ_del_dec_neon()
479 …t0_s32x4 = vdupq_n_s32( 0 ); /* zero zero zero zer… in silk_short_prediction_create_arch_coef_neon_local()
521 LPC_pred_Q14_s32x4 = vdupq_n_s32( silk_RSHIFT( order, 1 ) ); in silk_noise_shape_quantizer_short_prediction_neon_local()
660 …n_AR_Q14_s32x4 = vaddq_s32( vdupq_n_s32( silk_RSHIFT( shapingLPCOrder, 1 ) ), vqdmulhq_lane_s32( t… in silk_noise_shape_quantizer_del_dec_neon()
688 tmp2_s32x4 = vaddq_s32( vdupq_n_s32( n_LTP_Q14 ), LPC_pred_Q14_s32x4 ); /* Q13 */ in silk_noise_shape_quantizer_del_dec_neon()
691 … tmp1_s32x4 = vsubq_s32( vdupq_n_s32( x_Q10[ i ] ), tmp1_s32x4 ); /* residual error Q10 */ in silk_noise_shape_quantizer_del_dec_neon()
[all …]
/external/libhevc/encoder/arm/
Dihevce_hme_utils_neon.c122 int32x4_t log_wdc = vdupq_n_s32(ps_wt_inp_prms->wpred_log_wdc); in ihevce_get_wt_inp_4x8_neon()
143 add_4x32b = vdupq_n_s32(0x4000); in ihevce_get_wt_inp_4x8_neon()
157 inv_wt_4x32b = vdupq_n_s32(inv_wt); in ihevce_get_wt_inp_4x8_neon()
519 int32x4_t add_4x32b = vdupq_n_s32(0x4000); in hme_get_wt_inp_ctb_neon()
520 int32x4_t log_wdc = vdupq_n_s32(ps_wt_inp_prms->wpred_log_wdc); in hme_get_wt_inp_ctb_neon()
584 inv_wt_4x32b = vdupq_n_s32(ps_wt_inp_prms->a_inv_wpred_wt[u1_ref_idx]); in hme_get_wt_inp_ctb_neon()
691 off_4x32b = vdupq_n_s32(ps_wt_inp_prms->a_wpred_off[u1_ref_idx]); in hme_get_wt_inp_ctb_neon()
692 inv_wt_4x32b = vdupq_n_s32(ps_wt_inp_prms->a_inv_wpred_wt[u1_ref_idx]); in hme_get_wt_inp_ctb_neon()
/external/XNNPACK/src/requantization/
Dq31-neon.c45 const int32x4_t vmultiplier = vdupq_n_s32(multiplier); in xnn_requantize_q31__neon()
47 const int32x4_t vshift = vdupq_n_s32(-shift); in xnn_requantize_q31__neon()
Dgemmlowp-neon.c57 const int32x4_t vmultiplier = vdupq_n_s32(multiplier); in xnn_requantize_gemmlowp__neon()
59 const int32x4_t vshift = vdupq_n_s32(-shift); in xnn_requantize_gemmlowp__neon()
/external/tensorflow/tensorflow/lite/kernels/internal/optimized/
Doptimized_ops.h211 const int32x4_t left_shifted_one_dup = vdupq_n_s32(1 << left_shift); in MultiplyByQuantizedMultiplier4Rows()
445 int32x4_t row_accum0 = vdupq_n_s32(0); in ShuffledFullyConnectedWorkerImpl()
446 int32x4_t row_accum1 = vdupq_n_s32(0); in ShuffledFullyConnectedWorkerImpl()
447 int32x4_t row_accum2 = vdupq_n_s32(0); in ShuffledFullyConnectedWorkerImpl()
448 int32x4_t row_accum3 = vdupq_n_s32(0); in ShuffledFullyConnectedWorkerImpl()
497 reduced = vshlq_s32(reduced, vdupq_n_s32(left_shift)); in ShuffledFullyConnectedWorkerImpl()
514 int32x4_t row_accum00 = vdupq_n_s32(0); in ShuffledFullyConnectedWorkerImpl()
515 int32x4_t row_accum10 = vdupq_n_s32(0); in ShuffledFullyConnectedWorkerImpl()
516 int32x4_t row_accum20 = vdupq_n_s32(0); in ShuffledFullyConnectedWorkerImpl()
517 int32x4_t row_accum30 = vdupq_n_s32(0); in ShuffledFullyConnectedWorkerImpl()
[all …]
/external/tensorflow/tensorflow/lite/kernels/
Dcpu_backend_gemm_custom_gemv.h357 int32x4_t acc0 = vdupq_n_s32(0);
560 multiplier_exponent = vdupq_n_s32(params.multiplier_exponent);
561 multiplier_fixedpoint = vdupq_n_s32(params.multiplier_fixedpoint);
566 vmaxq_s32(multiplier_exponent, vdupq_n_s32(0));
572 vminq_s32(multiplier_exponent, vdupq_n_s32(0));
576 const int32x4_t output_offset_vec = vdupq_n_s32(dst_params.zero_point);
/external/libopus/celt/arm/
Dpitch_neon_intr.c42 int32x4_t xy_s32x4 = vdupq_n_s32(0); in celt_inner_prod_neon()
81 int32x4_t xy01_s32x4 = vdupq_n_s32(0); in dual_inner_prod_neon()
82 int32x4_t xy02_s32x4 = vdupq_n_s32(0); in dual_inner_prod_neon()

123