| /external/XNNPACK/src/q8-vadd/ |
| D | neon.c | 49 int32x4_t vacc0_lo = vmulq_s32(vmovl_s16(vget_low_s16(vxa0)), va_multiplier); in xnn_q8_vadd_ukernel__neon()
50 int32x4_t vacc1_lo = vmulq_s32(vmovl_s16(vget_low_s16(vxa1)), va_multiplier); in xnn_q8_vadd_ukernel__neon()
51 int32x4_t vacc2_lo = vmulq_s32(vmovl_s16(vget_low_s16(vxa2)), va_multiplier); in xnn_q8_vadd_ukernel__neon()
52 int32x4_t vacc3_lo = vmulq_s32(vmovl_s16(vget_low_s16(vxa3)), va_multiplier); in xnn_q8_vadd_ukernel__neon()
58 vacc0_lo = vmlaq_s32(vacc0_lo, vmovl_s16(vget_low_s16(vxb0)), vb_multiplier); in xnn_q8_vadd_ukernel__neon()
59 vacc1_lo = vmlaq_s32(vacc1_lo, vmovl_s16(vget_low_s16(vxb1)), vb_multiplier); in xnn_q8_vadd_ukernel__neon()
60 vacc2_lo = vmlaq_s32(vacc2_lo, vmovl_s16(vget_low_s16(vxb2)), vb_multiplier); in xnn_q8_vadd_ukernel__neon()
61 vacc3_lo = vmlaq_s32(vacc3_lo, vmovl_s16(vget_low_s16(vxb3)), vb_multiplier); in xnn_q8_vadd_ukernel__neon()
115 int32x4_t vacc0_lo = vmulq_s32(vmovl_s16(vget_low_s16(vxa0)), va_multiplier); in xnn_q8_vadd_ukernel__neon()
116 int32x4_t vacc1_lo = vmulq_s32(vmovl_s16(vget_low_s16(vxa1)), va_multiplier); in xnn_q8_vadd_ukernel__neon()
[all …]
|
| /external/libhevc/encoder/arm/ |
| D | ihevce_hme_utils_neon.c | 169 dst0_4x32b = vmovl_s16(vget_low_s16(src4_8x16b)); in ihevce_get_wt_inp_4x8_neon()
170 dst1_4x32b = vmovl_s16(vget_low_s16(src5_8x16b)); in ihevce_get_wt_inp_4x8_neon()
171 dst2_4x32b = vmovl_s16(vget_low_s16(src6_8x16b)); in ihevce_get_wt_inp_4x8_neon()
172 dst3_4x32b = vmovl_s16(vget_low_s16(src7_8x16b)); in ihevce_get_wt_inp_4x8_neon()
174 dst4_4x32b = vmovl_s16(vget_high_s16(src4_8x16b)); in ihevce_get_wt_inp_4x8_neon()
175 dst5_4x32b = vmovl_s16(vget_high_s16(src5_8x16b)); in ihevce_get_wt_inp_4x8_neon()
176 dst6_4x32b = vmovl_s16(vget_high_s16(src6_8x16b)); in ihevce_get_wt_inp_4x8_neon()
177 dst7_4x32b = vmovl_s16(vget_high_s16(src7_8x16b)); in ihevce_get_wt_inp_4x8_neon()
592 dst0_4x32b = vmovl_s16(vget_low_s16(src4_8x16b)); in hme_get_wt_inp_ctb_neon()
593 dst1_4x32b = vmovl_s16(vget_low_s16(src5_8x16b)); in hme_get_wt_inp_ctb_neon()
[all …]
|
| D | ihevce_common_utils_neon.c | 129 reg0[0] = vmovl_s16(vget_low_s16(a2)); in ihevce_wt_avg_2d_16x1_neon()
130 reg0[1] = vmovl_s16(vget_high_s16(a2)); in ihevce_wt_avg_2d_16x1_neon()
131 reg0[2] = vmovl_s16(vget_low_s16(a3)); in ihevce_wt_avg_2d_16x1_neon()
132 reg0[3] = vmovl_s16(vget_high_s16(a3)); in ihevce_wt_avg_2d_16x1_neon()
134 reg1[0] = vmovl_s16(vget_low_s16(a4)); in ihevce_wt_avg_2d_16x1_neon()
135 reg1[1] = vmovl_s16(vget_high_s16(a4)); in ihevce_wt_avg_2d_16x1_neon()
136 reg1[2] = vmovl_s16(vget_low_s16(a5)); in ihevce_wt_avg_2d_16x1_neon()
137 reg1[3] = vmovl_s16(vget_high_s16(a5)); in ihevce_wt_avg_2d_16x1_neon()
193 a8 = vmovl_s16(vget_low_s16(a0)); in ihevce_wt_avg_2d_8x1_neon()
194 a9 = vmovl_s16(vget_high_s16(a0)); in ihevce_wt_avg_2d_8x1_neon()
[all …]
|
| /external/tensorflow/tensorflow/lite/kernels/internal/optimized/integer_ops/ |
| D | add.h | 56 int32x4_t x11 = vmovl_s16(input1_val_low); in AddElementwise()
57 int32x4_t x12 = vmovl_s16(input1_val_high); in AddElementwise()
58 int32x4_t x21 = vmovl_s16(input2_val_low); in AddElementwise()
59 int32x4_t x22 = vmovl_s16(input2_val_high); in AddElementwise()
146 int32x4_t x11 = vmovl_s16(input1_val_low); in AddScalarBroadcast()
147 int32x4_t x12 = vmovl_s16(input1_val_high); in AddScalarBroadcast()
163 int32x4_t x21 = vmovl_s16(input2_val_low); in AddScalarBroadcast()
164 int32x4_t x22 = vmovl_s16(input2_val_high); in AddScalarBroadcast()
|
| D | softmax.h | 111 int32x4_t input_diff_s32_0 = vmovl_s16(vget_low_s16(input_diff_s16)); in Softmax()
112 int32x4_t input_diff_s32_1 = vmovl_s16(vget_high_s16(input_diff_s16)); in Softmax()
178 int32x4_t input_diff_s32_0 = vmovl_s16(vget_low_s16(input_diff_s16)); in Softmax()
179 int32x4_t input_diff_s32_1 = vmovl_s16(vget_high_s16(input_diff_s16)); in Softmax()
|
| D | mean.h | 77 vmovl_s16(vget_low_s16(input_data_low_shift)); in MeanImpl()
79 vmovl_s16(vget_high_s16(input_data_low_shift)); in MeanImpl()
81 vmovl_s16(vget_low_s16(input_data_high_shift)); in MeanImpl()
83 vmovl_s16(vget_high_s16(input_data_high_shift)); in MeanImpl()
|
| /external/tensorflow/tensorflow/lite/kernels/internal/optimized/ |
| D | neon_tensor_utils.cc | 847 const int32x4_t first_half = vmovl_s16(vget_low_s16(output_val)); in NeonMatrixBatchVectorAccumulateImpl()
848 const int32x4_t second_half = vmovl_s16(vget_high_s16(output_val)); in NeonMatrixBatchVectorAccumulateImpl()
905 const int32x4_t output_val_1 = vmovl_s16(vget_low_s16(first_half)); in NeonMatrixBatchVectorAccumulateImpl()
906 const int32x4_t output_val_2 = vmovl_s16(vget_high_s16(first_half)); in NeonMatrixBatchVectorAccumulateImpl()
907 const int32x4_t output_val_3 = vmovl_s16(vget_low_s16(second_half)); in NeonMatrixBatchVectorAccumulateImpl()
908 const int32x4_t output_val_4 = vmovl_s16(vget_high_s16(second_half)); in NeonMatrixBatchVectorAccumulateImpl()
1359 const int32x4_t val_s32_0 = vmovl_s16(vget_low_s16(val_s16)); in NeonApplyLayerNorm()
1360 const int32x4_t val_s32_1 = vmovl_s16(vget_high_s16(val_s16)); in NeonApplyLayerNorm()
1401 vshlq_n_s32(vmovl_s16(vget_low_s16(val_s16_0)), 10), mean_dup); in NeonApplyLayerNorm()
1403 vshlq_n_s32(vmovl_s16(vget_high_s16(val_s16_0)), 10), mean_dup); in NeonApplyLayerNorm()
[all …]
|
| D | optimized_ops.h | 973 vmovl_s16(vget_low_s16(input_data_low_shift)); in MeanImpl()
975 vmovl_s16(vget_high_s16(input_data_low_shift)); in MeanImpl()
977 vmovl_s16(vget_low_s16(input_data_high_shift)); in MeanImpl()
979 vmovl_s16(vget_high_s16(input_data_high_shift)); in MeanImpl()
1756 int32x4_t x11 = vmovl_s16(input1_val_low); in AddElementwise()
1757 int32x4_t x12 = vmovl_s16(input1_val_high); in AddElementwise()
1758 int32x4_t x21 = vmovl_s16(input2_val_low); in AddElementwise()
1759 int32x4_t x22 = vmovl_s16(input2_val_high); in AddElementwise()
1847 int32x4_t x11 = vmovl_s16(input1_val_low); in AddScalarBroadcast()
1848 int32x4_t x12 = vmovl_s16(input1_val_high); in AddScalarBroadcast()
[all …]
|
| D | legacy_optimized_ops.h | 4062 int32x4_t input_diff_s32_0 = vmovl_s16(vget_low_s16(input_diff_s16)); in Softmax()
4063 int32x4_t input_diff_s32_1 = vmovl_s16(vget_high_s16(input_diff_s16)); in Softmax()
4130 int32x4_t input_diff_s32_0 = vmovl_s16(vget_low_s16(input_diff_s16)); in Softmax()
4131 int32x4_t input_diff_s32_1 = vmovl_s16(vget_high_s16(input_diff_s16)); in Softmax()
4301 vshlq_s32(vmovl_s16(vget_low_s16(input_val_centered_0)), in Logistic()
4304 vshlq_s32(vmovl_s16(vget_high_s16(input_val_centered_0)), in Logistic()
4307 vshlq_s32(vmovl_s16(vget_low_s16(input_val_centered_1)), in Logistic()
4310 vshlq_s32(vmovl_s16(vget_high_s16(input_val_centered_1)), in Logistic()
4496 vshlq_s32(vmovl_s16(vget_low_s16(input_val_centered_0)), in Tanh()
4499 vshlq_s32(vmovl_s16(vget_high_s16(input_val_centered_0)), in Tanh()
[all …]
|
| /external/libopus/silk/arm/ |
| D | NSQ_neon.c | 86 int32x4_t coef0 = vmovl_s16(vget_low_s16(coef16)); in silk_NSQ_noise_shape_feedback_loop_neon()
87 int32x4_t coef1 = vmovl_s16(vget_high_s16(coef16)); in silk_NSQ_noise_shape_feedback_loop_neon()
|
| D | NSQ_del_dec_neon_intr.c | 759 tmp1_s32x4 = vbslq_s32( t_u32x4, vmovl_s16( q1_Q10_s16x4 ), vmovl_s16( q2_Q10_s16x4 ) ); in silk_noise_shape_quantizer_del_dec_neon()
760 tmp2_s32x4 = vbslq_s32( t_u32x4, vmovl_s16( q2_Q10_s16x4 ), vmovl_s16( q1_Q10_s16x4 ) ); in silk_noise_shape_quantizer_del_dec_neon()
|
| /external/libaom/libaom/av1/common/arm/ |
| D | convolve_neon.h | 89 sum_0 = vmovl_s16(vget_low_s16(sum)); in wiener_convolve8_horiz_8x8()
90 sum_1 = vmovl_s16(vget_high_s16(sum)); in wiener_convolve8_horiz_8x8()
140 sum_0 = vmovl_s16(sum); in wiener_convolve8_horiz_4x8()
|
| D | selfguided_neon.c | 436 q12345 = vaddq_s32(vmovl_s16(s1), q2345); in boxsum2()
437 q23456 = vaddq_s32(q2345, vmovl_s16(s6)); in boxsum2()
438 q34567 = vaddq_s32(q4567, vmovl_s16(s3)); in boxsum2()
439 q45678 = vaddq_s32(q4567, vmovl_s16(s8)); in boxsum2()
1073 a_res0 = vmulq_s32(vmovl_s16(vget_low_s16(s0)), a_res0); in final_filter_fast_internal()
1074 a_res1 = vmulq_s32(vmovl_s16(vget_high_s16(s0)), a_res1); in final_filter_fast_internal()
1099 a_res0 = vmulq_s32(vmovl_s16(vget_low_s16(s0)), a_res0); in final_filter_fast_internal()
1100 a_res1 = vmulq_s32(vmovl_s16(vget_high_s16(s0)), a_res1); in final_filter_fast_internal()
1151 a_res0 = vmulq_s32(vmovl_s16(vget_low_s16(s0)), a_res0); in final_filter_internal()
1152 a_res1 = vmulq_s32(vmovl_s16(vget_high_s16(s0)), a_res1); in final_filter_internal()
|
| D | mem_neon.h | 533 const int32x4_t v0 = vmovl_s16(vget_low_s16(a)); in store_s16q_to_tran_low()
534 const int32x4_t v1 = vmovl_s16(vget_high_s16(a)); in store_s16q_to_tran_low()
|
| D | jnt_convolve_neon.c | 44 dst0 = vsubq_s32(vreinterpretq_s32_u32(sum0), vmovl_s16(sub_const_vec)); in compute_avg_4x1()
78 const int32x4_t sub_const_vec = vmovl_s16(sub_const); in compute_avg_8x1()
133 const int32x4_t const_vec = vmovl_s16(sub_const_vec); in compute_avg_4x4()
214 const int32x4_t sub_const_vec = vmovl_s16(sub_const); in compute_avg_8x4()
|
| /external/libvpx/libvpx/vpx_dsp/arm/ |
| D | mem_neon.h | 78 const int32x4_t v0 = vmovl_s16(vget_low_s16(a)); in store_s16q_to_tran_low()
79 const int32x4_t v1 = vmovl_s16(vget_high_s16(a)); in store_s16q_to_tran_low()
|
| D | fdct32x32_neon.c | 547 temp##_lo[temp_index] = vmovl_s16(vget_low_s16(a[a_index])); \
548 temp##_hi[temp_index] = vmovl_s16(vget_high_s16(a[a_index])); \
|
| /external/webrtc/webrtc/modules/audio_coding/codecs/isac/fix/source/ |
| D | transform_neon.c | 396 int32x4_t tmpr_0 = vmovl_s16(vget_low_s16(tmpr)); in PostShiftAndDivideAndDemodulateNeon()
397 int32x4_t tmpi_0 = vmovl_s16(vget_low_s16(tmpi)); in PostShiftAndDivideAndDemodulateNeon()
402 int32x4_t tmpr_1 = vmovl_s16(vget_high_s16(tmpr)); in PostShiftAndDivideAndDemodulateNeon()
403 int32x4_t tmpi_1 = vmovl_s16(vget_high_s16(tmpi)); in PostShiftAndDivideAndDemodulateNeon()
|
| /external/libaom/libaom/aom_dsp/simd/ |
| D | v128_intrinsics_arm.h | 616 return vreinterpretq_s64_s32(vmovl_s16(vreinterpret_s16_s64(a))); in v128_unpack_s16_s32()
626 vmovl_s16(vreinterpret_s16_s64(vget_low_s64(a)))); in v128_unpacklo_s16_s32()
636 vmovl_s16(vreinterpret_s16_s64(vget_high_s64(a)))); in v128_unpackhi_s16_s32()
|
| D | v64_intrinsics_arm.h | 506 return vreinterpret_s64_s32(vget_low_s32(vmovl_s16(vreinterpret_s16_s64(x)))); in v64_unpacklo_s16_s32()
515 vget_high_s32(vmovl_s16(vreinterpret_s16_s64(x)))); in v64_unpackhi_s16_s32()
|
| /external/tensorflow/tensorflow/core/kernels/ |
| D | quantization_utils.h | 554 const int32x4_t input_value_low_32x4 = vmovl_s16(input_value_low_16x4); in Requantize8x8To32Neon()
555 const int32x4_t input_value_high_32x4 = vmovl_s16(input_value_high_16x4); in Requantize8x8To32Neon()
|
| /external/libhevc/common/arm/ |
| D | ihevc_resi_trans_neon.c | 281 int32x4_t c3_4x32b = vmulq_s32(vmovl_s16(src2_4x16b), coeff2_4x32b); /* 74*r2 */ in ihevc_resi_trans_4x4_ttype1_neon()
|
| /external/clang/test/CodeGen/ |
| D | aarch64-neon-intrinsics.c | 7035 return vmovl_s16(a); in test_vmovl_s16()
|
| D | arm_neon_intrinsics.c | 9833 return vmovl_s16(a); in test_vmovl_s16()
|
| /external/neon_2_sse/ |
| D | NEON_2_SSE.h | 1801 _NEON2SSESTORAGE int32x4_t vmovl_s16(int16x4_t a); // VMOVL.S16 q0,d0
12933 _NEON2SSESTORAGE int32x4_t vmovl_s16(int16x4_t a); // VMOVL.S16 q0,d0
12934 _NEON2SSE_INLINE int32x4_t vmovl_s16(int16x4_t a) in vmovl_s16() function
|