/external/llvm/test/CodeGen/ARM/ |
D | vcvt-cost.ll | 9 ; CHECK: vmovl.s8 10 ; CHECK: vmovl.s16 11 ; CHECK: vmovl.s16 25 ; CHECK: vmovl.u8 26 ; CHECK: vmovl.u16 27 ; CHECK: vmovl.u16 55 ; CHECK: vmovl.s16 56 ; CHECK: vmovl.s16 57 ; CHECK: vmovl.s16 58 ; CHECK: vmovl.s16 [all …]
|
D | 2012-08-09-neon-extload.ll | 22 ; CHECK: vmovl.s8 {{q[0-9]+}}, d[[LOAD]] 23 ; CHECK: vmovl.s16 {{q[0-9]+}}, {{d[0-9]+}} 36 ; CHECK: vmovl.s8 {{q[0-9]+}}, d[[LOAD]] 37 ; CHECK: vmovl.s16 {{q[0-9]+}}, {{d[0-9]+}} 38 ; CHECK: vmovl.s32 {{q[0-9]+}}, {{d[0-9]+}} 54 ; CHECK: vmovl.s8 {{q[0-9]+}}, d[[LOAD]] 55 ; CHECK-NOT: vmovl.s16 69 ; CHECK: vmovl.s8 {{q[0-9]+}}, d[[LOAD]] 70 ; CHECK: vmovl.s16 {{q[0-9]+}}, {{d[0-9]+}} 83 ; CHECK: vmovl.s16 {{q[0-9]+}}, d[[LOAD]] [all …]
|
D | big-endian-neon-extend.ll | 7 ; CHECK-NEXT: vmovl.u8 [[QREG:q[0-9]+]], [[REG]] 8 ; CHECK-NEXT: vmovl.u16 [[QREG]], [[REG]] 9 ; CHECK-NEXT: vmovl.u32 [[QREG]], [[REG]] 22 ; CHECK-NEXT: vmovl.u16 [[QREG:q[0-9]+]], [[REG]] 23 ; CHECK-NEXT: vmovl.u32 [[QREG]], [[REG]] 37 ; CHECK-NEXT: vmovl.u8 [[QREG:q[0-9]+]], [[REG]] 38 ; CHECK-NEXT: vmovl.u16 [[QREG]], [[REG]] 52 ; CHECK-NEXT: vmovl.u16 [[QREG:q[0-9]+]], [[REG]] 66 ; CHECK-NEXT: vmovl.u8 [[QREG:q[0-9]+]], [[REG]] 67 ; CHECK-NEXT: vmovl.u16 [[QREG]], [[REG]] [all …]
|
D | vector-extend-narrow.ll | 6 ; CHECK: vmovl.u16 26 ; CHECK: vmovl.u8 27 ; CHECK: vmovl.u16 55 ; CHECK: vmovl.s8 56 ; CHECK: vmovl.s16 68 ; CHECK: vmovl.u8 69 ; CHECK: vmovl.u16
|
/external/renderscript-intrinsics-replacement-toolkit/renderscript-toolkit/src/main/cpp/ |
D | Convolve_neon.S | 55 vmovl.u8 q2, d26 56 vmovl.u8 q3, d27 57 vmovl.u8 q4, d28 58 vmovl.u8 q5, d29 59 vmovl.u8 q6, d30 60 vmovl.u8 q7, d31 153 vmovl.u8 q9, d24 154 vmovl.u8 q10, d25 155 vmovl.u8 q11, d26 156 vmovl.u8 q12, d27 [all …]
|
D | ColorMatrix_neon.S | 116 vmovl.u8 q3, d3 117 vmovl.u8 q2, d2 118 vmovl.u8 q1, d1 119 vmovl.u8 q0, d0 120 vmovl.u16 q3, d6 121 vmovl.u16 q2, d4 122 vmovl.u16 q1, d2 123 vmovl.u16 q0, d0 135 vmovl.u8 q2, d2 136 vmovl.u8 q1, d1 [all …]
|
D | Lut3d_neon.S | 147 vmovl.u8 q0, d0 148 vmovl.u8 q1, d2 149 vmovl.u8 q2, d4 177 vmovl.u16 q8, d26 178 vmovl.u16 q9, d27 179 vmovl.u16 q10, d28 180 vmovl.u16 q11, d29
|
/external/ComputeLibrary/src/core/NEON/wrapper/intrinsics/ |
D | movl.h | 34 inline ptype vmovl(const vtype &a) \ 39 VMOVL_IMPL(uint16x8_t, uint8x8_t, vmovl, u8) 40 VMOVL_IMPL(int16x8_t, int8x8_t, vmovl, s8) 41 VMOVL_IMPL(uint32x4_t, uint16x4_t, vmovl, u16) 42 VMOVL_IMPL(int32x4_t, int16x4_t, vmovl, s16) 43 VMOVL_IMPL(uint64x2_t, uint32x2_t, vmovl, u32) 44 VMOVL_IMPL(int64x2_t, int32x2_t, vmovl, s32)
|
/external/ComputeLibrary/src/cpu/kernels/scale/neon/ |
D | integer.cpp | 215 const uint16x8_t in00_low = wrapper::vmovl(wrapper::vgetlow(in00)); in u8_neon_scale_bilinear() 216 const uint16x8_t in00_high = wrapper::vmovl(wrapper::vgethigh(in00)); in u8_neon_scale_bilinear() 218 … const auto in00_0 = wrapper::vcvt<float>(wrapper::vmovl(wrapper::vgetlow(in00_low))); in u8_neon_scale_bilinear() 219 … const auto in00_1 = wrapper::vcvt<float>(wrapper::vmovl(wrapper::vgethigh(in00_low))); in u8_neon_scale_bilinear() 220 … const auto in00_2 = wrapper::vcvt<float>(wrapper::vmovl(wrapper::vgetlow(in00_high))); in u8_neon_scale_bilinear() 221 … const auto in00_3 = wrapper::vcvt<float>(wrapper::vmovl(wrapper::vgethigh(in00_high))); in u8_neon_scale_bilinear() 223 const uint16x8_t in01_low = wrapper::vmovl(wrapper::vgetlow(in01)); in u8_neon_scale_bilinear() 224 const uint16x8_t in01_high = wrapper::vmovl(wrapper::vgethigh(in01)); in u8_neon_scale_bilinear() 226 … const auto in01_0 = wrapper::vcvt<float>(wrapper::vmovl(wrapper::vgetlow(in01_low))); in u8_neon_scale_bilinear() 227 … const auto in01_1 = wrapper::vcvt<float>(wrapper::vmovl(wrapper::vgethigh(in01_low))); in u8_neon_scale_bilinear() [all …]
|
D | qasymm8_signed.cpp | 203 const int16x8_t in00_low = wrapper::vmovl(wrapper::vgetlow(in00)); in qasymm8_signed_neon_scale_bilinear() 204 const int16x8_t in00_high = wrapper::vmovl(wrapper::vgethigh(in00)); in qasymm8_signed_neon_scale_bilinear() 206 …t auto in00_0 = wrapper::vmul(wrapper::vcvt<float>(wrapper::vsub(wrapper::vmovl(wrapper::vgetlow(i… in qasymm8_signed_neon_scale_bilinear() 207 …t auto in00_1 = wrapper::vmul(wrapper::vcvt<float>(wrapper::vsub(wrapper::vmovl(wrapper::vgethigh(… in qasymm8_signed_neon_scale_bilinear() 208 …t auto in00_2 = wrapper::vmul(wrapper::vcvt<float>(wrapper::vsub(wrapper::vmovl(wrapper::vgetlow(i… in qasymm8_signed_neon_scale_bilinear() 209 …t auto in00_3 = wrapper::vmul(wrapper::vcvt<float>(wrapper::vsub(wrapper::vmovl(wrapper::vgethigh(… in qasymm8_signed_neon_scale_bilinear() 211 const int16x8_t in01_low = wrapper::vmovl(wrapper::vgetlow(in01)); in qasymm8_signed_neon_scale_bilinear() 212 const int16x8_t in01_high = wrapper::vmovl(wrapper::vgethigh(in01)); in qasymm8_signed_neon_scale_bilinear() 214 …t auto in01_0 = wrapper::vmul(wrapper::vcvt<float>(wrapper::vsub(wrapper::vmovl(wrapper::vgetlow(i… in qasymm8_signed_neon_scale_bilinear() 215 …t auto in01_1 = wrapper::vmul(wrapper::vcvt<float>(wrapper::vsub(wrapper::vmovl(wrapper::vgethigh(… in qasymm8_signed_neon_scale_bilinear() [all …]
|
D | qasymm8.cpp | 204 const uint16x8_t in00_low = wrapper::vmovl(wrapper::vgetlow(in00)); in qasymm8_neon_scale_bilinear() 205 const uint16x8_t in00_high = wrapper::vmovl(wrapper::vgethigh(in00)); in qasymm8_neon_scale_bilinear() 207 …r::vmul(wrapper::vcvt<float>(wrapper::vsub(wrapper::vreinterpret(wrapper::vmovl(wrapper::vgetlow(i… in qasymm8_neon_scale_bilinear() 208 …r::vmul(wrapper::vcvt<float>(wrapper::vsub(wrapper::vreinterpret(wrapper::vmovl(wrapper::vgethigh(… in qasymm8_neon_scale_bilinear() 209 …r::vmul(wrapper::vcvt<float>(wrapper::vsub(wrapper::vreinterpret(wrapper::vmovl(wrapper::vgetlow(i… in qasymm8_neon_scale_bilinear() 210 …r::vmul(wrapper::vcvt<float>(wrapper::vsub(wrapper::vreinterpret(wrapper::vmovl(wrapper::vgethigh(… in qasymm8_neon_scale_bilinear() 212 const uint16x8_t in01_low = wrapper::vmovl(wrapper::vgetlow(in01)); in qasymm8_neon_scale_bilinear() 213 const uint16x8_t in01_high = wrapper::vmovl(wrapper::vgethigh(in01)); in qasymm8_neon_scale_bilinear() 215 …r::vmul(wrapper::vcvt<float>(wrapper::vsub(wrapper::vreinterpret(wrapper::vmovl(wrapper::vgetlow(i… in qasymm8_neon_scale_bilinear() 216 …r::vmul(wrapper::vcvt<float>(wrapper::vsub(wrapper::vreinterpret(wrapper::vmovl(wrapper::vgethigh(… in qasymm8_neon_scale_bilinear() [all …]
|
/external/libhevc/common/arm/ |
D | ihevc_inter_pred_chroma_copy_w16out.s | 141 vmovl.u8 q0,d0 @vmovl_u8(vld1_u8(pu1_src_tmp) 149 vmovl.u8 q11,d22 @vmovl_u8(vld1_u8(pu1_src_tmp) 152 vmovl.u8 q12,d24 @vmovl_u8(vld1_u8(pu1_src_tmp) 157 vmovl.u8 q13,d26 @vmovl_u8(vld1_u8(pu1_src_tmp) 183 vmovl.u8 q0,d0 @vmovl_u8(vld1_u8(pu1_src_tmp) 191 vmovl.u8 q11,d22 @vmovl_u8(vld1_u8(pu1_src_tmp) 194 vmovl.u8 q12,d24 @vmovl_u8(vld1_u8(pu1_src_tmp) 220 vmovl.u8 q8,d8 @vmovl_u8(vld1_u8(pu1_src_tmp)) 221 vmovl.u8 q9,d10 @vmovl_u8(vld1_u8(pu1_src_tmp) 222 vmovl.u8 q10,d12 @vmovl_u8(vld1_u8(pu1_src_tmp) [all …]
|
D | ihevc_inter_pred_luma_copy_w16out.s | 109 vmovl.u8 q0,d0 @vmovl_u8(vld1_u8(pu1_src_tmp) 117 vmovl.u8 q11,d22 @vmovl_u8(vld1_u8(pu1_src_tmp) 120 vmovl.u8 q12,d24 @vmovl_u8(vld1_u8(pu1_src_tmp) 125 vmovl.u8 q13,d26 @vmovl_u8(vld1_u8(pu1_src_tmp) 158 vmovl.u8 q8,d8 @vmovl_u8(vld1_u8(pu1_src_tmp)) 159 vmovl.u8 q9,d10 @vmovl_u8(vld1_u8(pu1_src_tmp) 160 vmovl.u8 q10,d12 @vmovl_u8(vld1_u8(pu1_src_tmp) 161 vmovl.u8 q11,d14 @vmovl_u8(vld1_u8(pu1_src_tmp) 188 vmovl.u8 q8,d8 @vmovl_u8(vld1_u8(pu1_src_tmp)) 191 vmovl.u8 q9,d10 @vmovl_u8(vld1_u8(pu1_src_tmp) [all …]
|
D | ihevc_deblk_chroma_horz.s | 70 vmovl.u8 q0,d0 80 vmovl.u8 q1,d2 91 vmovl.u8 q2,d4 103 vmovl.u8 q8,d16
|
/external/llvm/test/MC/ARM/ |
D | neont2-mov-encoding.s | 70 vmovl.s8 q8, d16 71 vmovl.s16 q8, d16 72 vmovl.s32 q8, d16 73 vmovl.u8 q8, d16 74 vmovl.u16 q8, d16 75 vmovl.u32 q8, d16 89 @ CHECK: vmovl.s8 q8, d16 @ encoding: [0xc8,0xef,0x30,0x0a] 90 @ CHECK: vmovl.s16 q8, d16 @ encoding: [0xd0,0xef,0x30,0x0a] 91 @ CHECK: vmovl.s32 q8, d16 @ encoding: [0xe0,0xef,0x30,0x0a] 92 @ CHECK: vmovl.u8 q8, d16 @ encoding: [0xc8,0xff,0x30,0x0a] [all …]
|
D | neon-mov-encoding.s | 67 vmovl.s8 q8, d16 68 vmovl.s16 q8, d16 69 vmovl.s32 q8, d16 70 vmovl.u8 q8, d16 71 vmovl.u16 q8, d16 72 vmovl.u32 q8, d16 74 @ CHECK: vmovl.s8 q8, d16 @ encoding: [0x30,0x0a,0xc8,0xf2] 75 @ CHECK: vmovl.s16 q8, d16 @ encoding: [0x30,0x0a,0xd0,0xf2] 76 @ CHECK: vmovl.s32 q8, d16 @ encoding: [0x30,0x0a,0xe0,0xf2] 77 @ CHECK: vmovl.u8 q8, d16 @ encoding: [0x30,0x0a,0xc8,0xf3] [all …]
|
/external/libavc/common/arm/ |
D | ih264_weighted_bi_pred_a9q.s | 174 vmovl.u8 q2, d4 @converting rows 1,2 in source 1 to 16-bit 177 vmovl.u8 q3, d6 @converting rows 1,2 in source 2 to 16-bit 181 vmovl.u8 q4, d8 @converting rows 3,4 in source 1 to 16-bit 182 vmovl.u8 q5, d10 @converting rows 3,4 in source 2 to 16-bit 214 vmovl.u8 q2, d4 @converting row 1 in source 1 to 16-bit 217 vmovl.u8 q3, d6 @converting row 1 in source 2 to 16-bit 221 vmovl.u8 q4, d8 @converting row 2 in source 1 to 16-bit 222 vmovl.u8 q5, d10 @converting row 2 in source 2 to 16-bit 226 vmovl.u8 q6, d12 @converting row 3 in source 1 to 16-bit 227 vmovl.u8 q7, d14 @converting row 3 in source 2 to 16-bit [all …]
|
D | ih264_weighted_pred_a9q.s | 141 vmovl.u8 q2, d4 @converting rows 1,2 to 16-bit 142 vmovl.u8 q3, d6 @converting rows 3,4 to 16-bit 171 vmovl.u8 q2, d4 @converting row 1 to 16-bit 173 vmovl.u8 q3, d6 @converting row 2 to 16-bit 175 vmovl.u8 q4, d8 @converting row 3 to 16-bit 177 vmovl.u8 q5, d10 @converting row 4 to 16-bit 210 vmovl.u8 q6, d4 @converting row 1L to 16-bit 212 vmovl.u8 q7, d5 @converting row 1H to 16-bit 215 vmovl.u8 q8, d6 @converting row 2L to 16-bit 217 vmovl.u8 q9, d7 @converting row 2H to 16-bit [all …]
|
/external/capstone/suite/MC/ARM/ |
D | neont2-mov-encoding.s.cs | 30 0xc8,0xef,0x30,0x0a = vmovl.s8 q8, d16 31 0xd0,0xef,0x30,0x0a = vmovl.s16 q8, d16 32 0xe0,0xef,0x30,0x0a = vmovl.s32 q8, d16 33 0xc8,0xff,0x30,0x0a = vmovl.u8 q8, d16 34 0xd0,0xff,0x30,0x0a = vmovl.u16 q8, d16 35 0xe0,0xff,0x30,0x0a = vmovl.u32 q8, d16
|
D | neon-mov-encoding.s.cs | 30 0x30,0x0a,0xc8,0xf2 = vmovl.s8 q8, d16 31 0x30,0x0a,0xd0,0xf2 = vmovl.s16 q8, d16 32 0x30,0x0a,0xe0,0xf2 = vmovl.s32 q8, d16 33 0x30,0x0a,0xc8,0xf3 = vmovl.u8 q8, d16 34 0x30,0x0a,0xd0,0xf3 = vmovl.u16 q8, d16 35 0x30,0x0a,0xe0,0xf3 = vmovl.u32 q8, d16
|
/external/ComputeLibrary/src/core/NEON/kernels/ |
D | NEQLSTMLayerNormalizationKernel.cpp | 59 const int64x2_t a_low = vmovl(vgetlow(a)); in mul_add() 60 const int64x2_t a_high = vmovl(vgethigh(a)); in mul_add() 61 const int64x2_t b_low = vmovl(vgetlow(b)); in mul_add() 62 const int64x2_t b_high = vmovl(vgethigh(b)); in mul_add() 77 result.val[0] = vadd(vmovl(vgetlow(bias)), result_0); in mul_add() 78 result.val[1] = vadd(vmovl(vgethigh(bias)), result_1); in mul_add() 193 const int32x4_t val_low = vmovl(vgetlow(val)); in sum_qsymm16() 194 const int32x4_t val_high = vmovl(vgethigh(val)); in sum_qsymm16() 244 shifted.val[0] = vsub(vshlq_n_s32(vmovl(vgetlow(val)), 10), mean_vec); in normalize_qasymm16() 245 shifted.val[1] = vsub(vshlq_n_s32(vmovl(vgethigh(val)), 10), mean_vec); in normalize_qasymm16() [all …]
|
D | NEReductionOperationKernel.cpp | 99 …auto wide_u16_1 = wrapper::vorr(vshll_n_u8(wrapper::vgetlow(mask_u8), 8), wrapper::vmovl(wrapper::… in calculate_index_quantized() 100 …auto wide_u16_2 = wrapper::vorr(vshll_n_u8(wrapper::vgethigh(mask_u8), 8), wrapper::vmovl(wrapper:… in calculate_index_quantized() 101 …mask.val[0] = wrapper::vorr(vshll_n_u16(wrapper::vgetlow(wide_u16_1), 16), wrapper::vmovl(wrap… in calculate_index_quantized() 102 …mask.val[1] = wrapper::vorr(vshll_n_u16(wrapper::vgethigh(wide_u16_1), 16), wrapper::vmovl(wra… in calculate_index_quantized() 103 …mask.val[2] = wrapper::vorr(vshll_n_u16(wrapper::vgetlow(wide_u16_2), 16), wrapper::vmovl(wrap… in calculate_index_quantized() 104 …mask.val[3] = wrapper::vorr(vshll_n_u16(wrapper::vgethigh(wide_u16_2), 16), wrapper::vmovl(wra… in calculate_index_quantized() 221 …auto wide_u16_1 = wrapper::vorr(vshll_n_u8(wrapper::vgetlow(mask_u8), 8), wrapper::vmovl(wrapp… in calculate_vector_index_quantized() 222 …auto wide_u16_2 = wrapper::vorr(vshll_n_u8(wrapper::vgethigh(mask_u8), 8), wrapper::vmovl(wrap… in calculate_vector_index_quantized() 223 … = wrapper::vorr(vshll_n_u16(wrapper::vgetlow(wide_u16_1), 16), wrapper::vmovl(wrapper::vgetlow(w… in calculate_vector_index_quantized() 224 … = wrapper::vorr(vshll_n_u16(wrapper::vgethigh(wide_u16_1), 16), wrapper::vmovl(wrapper::vgethigh(… in calculate_vector_index_quantized() [all …]
|
/external/ComputeLibrary/src/cpu/kernels/conv3d/neon/ |
D | quantized.h | 188 const auto src_q16_0 = wrapper::vmovl(wrapper::vgetlow(src_vec)); in directconv3d_quantized_neon_ndhwc() 189 const auto src_q16_1 = wrapper::vmovl(wrapper::vgethigh(src_vec)); in directconv3d_quantized_neon_ndhwc() 190 const auto wei_q16_0 = wrapper::vmovl(wrapper::vgetlow(w_vec)); in directconv3d_quantized_neon_ndhwc() 191 const auto wei_q16_1 = wrapper::vmovl(wrapper::vgethigh(w_vec)); in directconv3d_quantized_neon_ndhwc() 193 … src_q32_0 = wrapper::vadd(src_q32_0, wrapper::vmovl(wrapper::vgetlow(src_q16_0))); in directconv3d_quantized_neon_ndhwc() 194 … src_q32_1 = wrapper::vadd(src_q32_1, wrapper::vmovl(wrapper::vgethigh(src_q16_0))); in directconv3d_quantized_neon_ndhwc() 195 … src_q32_2 = wrapper::vadd(src_q32_2, wrapper::vmovl(wrapper::vgetlow(src_q16_1))); in directconv3d_quantized_neon_ndhwc() 196 … src_q32_3 = wrapper::vadd(src_q32_3, wrapper::vmovl(wrapper::vgethigh(src_q16_1))); in directconv3d_quantized_neon_ndhwc() 198 … wei_q32_0 = wrapper::vadd(wei_q32_0, wrapper::vmovl(wrapper::vgetlow(wei_q16_0))); in directconv3d_quantized_neon_ndhwc() 199 … wei_q32_1 = wrapper::vadd(wei_q32_1, wrapper::vmovl(wrapper::vgethigh(wei_q16_0))); in directconv3d_quantized_neon_ndhwc() [all …]
|
/external/llvm/test/CodeGen/AArch64/ |
D | arm64-neon-3vdiff.ll | 57 %vmovl.i.i = sext <8 x i8> %a to <8 x i16> 58 %vmovl.i2.i = sext <8 x i8> %b to <8 x i16> 59 %add.i = add <8 x i16> %vmovl.i.i, %vmovl.i2.i 67 %vmovl.i.i = sext <4 x i16> %a to <4 x i32> 68 %vmovl.i2.i = sext <4 x i16> %b to <4 x i32> 69 %add.i = add <4 x i32> %vmovl.i.i, %vmovl.i2.i 77 %vmovl.i.i = sext <2 x i32> %a to <2 x i64> 78 %vmovl.i2.i = sext <2 x i32> %b to <2 x i64> 79 %add.i = add <2 x i64> %vmovl.i.i, %vmovl.i2.i 87 %vmovl.i.i = zext <8 x i8> %a to <8 x i16> [all …]
|
/external/ComputeLibrary/src/core/NEON/kernels/detail/ |
D | NEDirectConvolutionDetail.h | 503 …wrapper::vaddw(v_input_offset, wrapper::vreinterpret(wrapper::vgetlow(wrapper::vmovl(vtop.val[0]))… in single_convolve_3x3_dilation() 504 …wrapper::vaddw(v_input_offset, wrapper::vreinterpret(wrapper::vgetlow(wrapper::vmovl(vtop.val[1]))… in single_convolve_3x3_dilation() 505 …wrapper::vaddw(v_input_offset, wrapper::vreinterpret(wrapper::vgetlow(wrapper::vmovl(vtop.val[2]))… in single_convolve_3x3_dilation() 511 …wrapper::vaddw(v_input_offset, wrapper::vreinterpret(wrapper::vgetlow(wrapper::vmovl(vmid.val[0]))… in single_convolve_3x3_dilation() 512 …wrapper::vaddw(v_input_offset, wrapper::vreinterpret(wrapper::vgetlow(wrapper::vmovl(vmid.val[1]))… in single_convolve_3x3_dilation() 513 …wrapper::vaddw(v_input_offset, wrapper::vreinterpret(wrapper::vgetlow(wrapper::vmovl(vmid.val[2]))… in single_convolve_3x3_dilation() 519 …wrapper::vaddw(v_input_offset, wrapper::vreinterpret(wrapper::vgetlow(wrapper::vmovl(vlow.val[0]))… in single_convolve_3x3_dilation() 520 …wrapper::vaddw(v_input_offset, wrapper::vreinterpret(wrapper::vgetlow(wrapper::vmovl(vlow.val[1]))… in single_convolve_3x3_dilation() 521 …wrapper::vaddw(v_input_offset, wrapper::vreinterpret(wrapper::vgetlow(wrapper::vmovl(vlow.val[2]))… in single_convolve_3x3_dilation() 628 …wrapper::vaddw(v_input_offset, wrapper::vreinterpret(wrapper::vgetlow(wrapper::vmovl(vtop.val[0]))… in convolve_3x3() [all …]
|