/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() 53 int32x4_t vacc0_hi = vmulq_s32(vmovl_high_s16(vxa0), va_multiplier); in xnn_q8_vadd_ukernel__neon() 54 int32x4_t vacc1_hi = vmulq_s32(vmovl_high_s16(vxa1), va_multiplier); in xnn_q8_vadd_ukernel__neon() 55 int32x4_t vacc2_hi = vmulq_s32(vmovl_high_s16(vxa2), va_multiplier); in xnn_q8_vadd_ukernel__neon() 56 int32x4_t vacc3_hi = vmulq_s32(vmovl_high_s16(vxa3), va_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/common/arm/ |
D | ihevc_resi_trans_neon_32x32.c | 1203 a[0] = vmulq_s32(g_ai4_ihevc_trans_32_0_8, eeee); in ihevc_resi_trans_32x32_neon() 1215 a[4] = vmulq_s32(g_ai4_ihevc_trans_32_4_04, eeo); in ihevc_resi_trans_32x32_neon() 1216 a[12] = vmulq_s32(g_ai4_ihevc_trans_32_12_04, eeo); in ihevc_resi_trans_32x32_neon() 1217 a[20] = vmulq_s32(g_ai4_ihevc_trans_32_20_04, eeo); in ihevc_resi_trans_32x32_neon() 1218 a[28] = vmulq_s32(g_ai4_ihevc_trans_32_28_04, eeo); in ihevc_resi_trans_32x32_neon() 1245 a[2] = vmulq_s32(g_ai4_ihevc_trans_32_2_03, eo0); //q8 in ihevc_resi_trans_32x32_neon() 1246 a[6] = vmulq_s32(g_ai4_ihevc_trans_32_6_03, eo0); //q2 in ihevc_resi_trans_32x32_neon() 1247 a[10] = vmulq_s32(g_ai4_ihevc_trans_32_10_03, eo0); //q2 in ihevc_resi_trans_32x32_neon() 1248 a[14] = vmulq_s32(g_ai4_ihevc_trans_32_14_03, eo0); //q2 in ihevc_resi_trans_32x32_neon() 1275 a[18] = vmulq_s32(g_ai4_ihevc_trans_32_18_03, eo0); //q2 in ihevc_resi_trans_32x32_neon() [all …]
|
D | ihevc_quant_iquant_ssd_neon_intr.c | 280 r0 = vmulq_s32(r0, r0); in ihevc_quant_iquant_ssd_flat_scale_mat_neon() 281 r1 = vmulq_s32(r1, r1); in ihevc_quant_iquant_ssd_flat_scale_mat_neon() 282 r2 = vmulq_s32(r2, r2); in ihevc_quant_iquant_ssd_flat_scale_mat_neon() 283 r3 = vmulq_s32(r3, r3); in ihevc_quant_iquant_ssd_flat_scale_mat_neon() 640 r0 = vmulq_s32(r0, r0); in ihevc_q_iq_ssd_flat_scale_mat_var_rnd_fact_neon() 641 r1 = vmulq_s32(r1, r1); in ihevc_q_iq_ssd_flat_scale_mat_var_rnd_fact_neon() 642 r2 = vmulq_s32(r2, r2); in ihevc_q_iq_ssd_flat_scale_mat_var_rnd_fact_neon() 643 r3 = vmulq_s32(r3, r3); in ihevc_q_iq_ssd_flat_scale_mat_var_rnd_fact_neon()
|
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() 286 src2_4x32b = vmulq_s32(c2_4x32b, coeff0_4x32b); /* 29*c2 - c3 */ in ihevc_resi_trans_4x4_ttype1_neon() 292 c1_4x32b = vmulq_s32(c1_4x32b, coeff0_4x32b); /* 55*c2 - 29*c1 + c3 */ in ihevc_resi_trans_4x4_ttype1_neon() 293 src1_4x32b = vmulq_s32(src1_4x32b, coeff2_4x32b); /*74*(r0+r1-r3)*/ in ihevc_resi_trans_4x4_ttype1_neon() 319 c3_4x32b = vmulq_s32(src2_4x32b, coeff2_4x32b); /* 74*r2 */ in ihevc_resi_trans_4x4_ttype1_neon() 324 src2_4x32b = vmulq_s32(c2_4x32b, coeff0_4x32b); /* 29*c2 - c3 */ in ihevc_resi_trans_4x4_ttype1_neon() 330 c1_4x32b = vmulq_s32(c1_4x32b, coeff0_4x32b); /* 55*c2 - 29*c1 + c3 */ in ihevc_resi_trans_4x4_ttype1_neon() 331 src1_4x32b = vmulq_s32(src1_4x32b, coeff2_4x32b); /*74*(r0+r1-r3)*/ in ihevc_resi_trans_4x4_ttype1_neon()
|
/external/libhevc/encoder/arm/ |
D | ihevce_common_utils_neon.c | 139 reg0[0] = vmulq_s32(reg0[0], a6); in ihevce_wt_avg_2d_16x1_neon() 140 reg0[1] = vmulq_s32(reg0[1], a6); in ihevce_wt_avg_2d_16x1_neon() 141 reg0[2] = vmulq_s32(reg0[2], a6); in ihevce_wt_avg_2d_16x1_neon() 142 reg0[3] = vmulq_s32(reg0[3], a6); in ihevce_wt_avg_2d_16x1_neon() 144 reg1[0] = vmulq_s32(reg1[0], a7); in ihevce_wt_avg_2d_16x1_neon() 145 reg1[1] = vmulq_s32(reg1[1], a7); in ihevce_wt_avg_2d_16x1_neon() 146 reg1[2] = vmulq_s32(reg1[2], a7); in ihevce_wt_avg_2d_16x1_neon() 147 reg1[3] = vmulq_s32(reg1[3], a7); in ihevce_wt_avg_2d_16x1_neon() 198 a8 = vmulq_s32(a8, a4); in ihevce_wt_avg_2d_8x1_neon() 199 a9 = vmulq_s32(a9, a4); in ihevce_wt_avg_2d_8x1_neon() [all …]
|
/external/tensorflow/tensorflow/lite/kernels/internal/optimized/ |
D | neon_tensor_utils.cc | 120 vmulq_s32(input_val.val[0], left_shifted_one_dup), in MultiplyByQuantizedMultiplier4Rows() 125 vmulq_s32(input_val.val[1], left_shifted_one_dup), in MultiplyByQuantizedMultiplier4Rows() 130 vmulq_s32(input_val.val[2], left_shifted_one_dup), in MultiplyByQuantizedMultiplier4Rows() 135 vmulq_s32(input_val.val[3], left_shifted_one_dup), in MultiplyByQuantizedMultiplier4Rows() 181 vmulq_s32(input_val.val[0], left_shifted_one_dup), in MultiplyByQuantizedMultiplier2Rows() 186 vmulq_s32(input_val.val[1], left_shifted_one_dup), in MultiplyByQuantizedMultiplier2Rows() 1366 AccumulateNeonLane(vmulq_s32(val_s32_0, val_s32_0))); in NeonApplyLayerNorm() 1368 AccumulateNeonLane(vmulq_s32(val_s32_1, val_s32_1))); in NeonApplyLayerNorm() 1586 int32x4_t x_0 = vmulq_s32(a_s32_0, b_s32_0); in NeonCwiseMul() 1587 int32x4_t x_1 = vmulq_s32(a_s32_1, b_s32_1); in NeonCwiseMul() [all …]
|
D | optimized_ops.h | 214 vmulq_s32(input_val.val[0], left_shifted_one_dup), in MultiplyByQuantizedMultiplier4Rows() 219 vmulq_s32(input_val.val[1], left_shifted_one_dup), in MultiplyByQuantizedMultiplier4Rows() 224 vmulq_s32(input_val.val[2], left_shifted_one_dup), in MultiplyByQuantizedMultiplier4Rows() 229 vmulq_s32(input_val.val[3], left_shifted_one_dup), in MultiplyByQuantizedMultiplier4Rows()
|
/external/libaom/libaom/av1/common/arm/ |
D | warp_plane_neon.c | 644 res_lo = vmulq_s32(res_lo, bwd); in av1_warp_affine_neon() 645 tmp32_lo = vmulq_s32(tmp32_lo, fwd); in av1_warp_affine_neon() 675 res_hi = vmulq_s32(res_hi, bwd); in av1_warp_affine_neon() 676 tmp32_hi = vmulq_s32(tmp32_hi, fwd); in av1_warp_affine_neon()
|
D | selfguided_neon.c | 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()
|
/external/libgav1/libgav1/src/dsp/arm/ |
D | motion_vector_search_neon.cc | 39 const int32x4_t m = vmulq_s32(m0, numerator); in MvProjection()
|
/external/libvpx/libvpx/vp9/common/arm/neon/ |
D | vp9_iht_neon.h | 50 s[2] = vmulq_s32(c3, s[7]); in iadst4()
|
/external/tensorflow/tensorflow/core/kernels/ |
D | quantization_utils.h | 361 vmulq_s32(vshrq_n_s32(input_value_low_32x4, 16), range_scale_fp_32x4)); in RequantizeManyInNewRangeNeon() 364 vmulq_s32(vshrq_n_s32(input_value_high_32x4, 16), range_scale_fp_32x4)); in RequantizeManyInNewRangeNeon()
|
/external/libaom/libaom/aom_dsp/simd/ |
D | v128_intrinsics_arm.h | 115 vmulq_s32(vreinterpretq_s32_s64(a), vreinterpretq_s32_s64(b))); in v128_dotp_s32() 302 vmulq_s32(vreinterpretq_s32_s64(a), vreinterpretq_s32_s64(b))); in v128_mullo_s32()
|
/external/libvpx/libvpx/vpx_dsp/arm/ |
D | deblock_neon.c | 283 const int32x4_t a = vmulq_s32(sumsq, fifteen); in calculate_mask()
|
/external/eigen/Eigen/src/Core/arch/NEON/ |
D | PacketMath.h | 143 …ONG_INLINE Packet4i pmul<Packet4i>(const Packet4i& a, const Packet4i& b) { return vmulq_s32(a,b); }
|
/external/skia/include/private/ |
D | SkNx_neon.h | 543 AI SkNx operator * (const SkNx& o) const { return vmulq_s32(fVec, o.fVec); }
|
/external/skqp/include/private/ |
D | SkNx_neon.h | 543 AI SkNx operator * (const SkNx& o) const { return vmulq_s32(fVec, o.fVec); }
|
/external/neon_2_sse/ |
D | NEON_2_SSE.h | 546 _NEON2SSESTORAGE int32x4_t vmulq_s32(int32x4_t a, int32x4_t b); // VMUL.I32 q0,q0,q0 3629 _NEON2SSESTORAGE int32x4_t vmulq_s32(int32x4_t a, int32x4_t b); // VMUL.I32 q0,q0,q0 3630 #define vmulq_s32 _MM_MULLO_EPI32 //SSE4.1 macro 13655 return vmulq_s32(a, b32x4); in vmulq_n_s32() 13736 return vmulq_s32(a, b32x4);
|
/external/clang/test/CodeGen/ |
D | aarch64-neon-intrinsics.c | 342 return vmulq_s32(v1, v2); in test_vmulq_s32()
|
D | arm_neon_intrinsics.c | 10288 return vmulq_s32(a, b); in test_vmulq_s32()
|