/external/XNNPACK/src/u8-maxpool/ |
D | 9p8x-minmax-neon-c16.c | 31 const uint8x16_t voutput_max = vld1q_dup_u8(¶ms->neon.max); in xnn_u8_maxpool_minmax_ukernel_9p8x__neon_c16() 32 const uint8x16_t voutput_min = vld1q_dup_u8(¶ms->neon.min); in xnn_u8_maxpool_minmax_ukernel_9p8x__neon_c16() 81 const uint8x16_t vi0 = vld1q_u8(i0); i0 += 16; in xnn_u8_maxpool_minmax_ukernel_9p8x__neon_c16() 82 const uint8x16_t vi1 = vld1q_u8(i1); i1 += 16; in xnn_u8_maxpool_minmax_ukernel_9p8x__neon_c16() 83 const uint8x16_t vi2 = vld1q_u8(i2); i2 += 16; in xnn_u8_maxpool_minmax_ukernel_9p8x__neon_c16() 84 const uint8x16_t vi3 = vld1q_u8(i3); i3 += 16; in xnn_u8_maxpool_minmax_ukernel_9p8x__neon_c16() 85 const uint8x16_t vi4 = vld1q_u8(i4); i4 += 16; in xnn_u8_maxpool_minmax_ukernel_9p8x__neon_c16() 86 const uint8x16_t vi5 = vld1q_u8(i5); i5 += 16; in xnn_u8_maxpool_minmax_ukernel_9p8x__neon_c16() 87 const uint8x16_t vi6 = vld1q_u8(i6); i6 += 16; in xnn_u8_maxpool_minmax_ukernel_9p8x__neon_c16() 88 const uint8x16_t vi7 = vld1q_u8(i7); i7 += 16; in xnn_u8_maxpool_minmax_ukernel_9p8x__neon_c16() [all …]
|
/external/libvpx/libvpx/vp8/encoder/arm/neon/ |
D | denoising_neon.c | 62 const uint8x16_t v_level1_adjustment = vmovq_n_u8( in vp8_denoiser_filter_neon() 64 const uint8x16_t v_delta_level_1_and_2 = vdupq_n_u8(1); in vp8_denoiser_filter_neon() 65 const uint8x16_t v_delta_level_2_and_3 = vdupq_n_u8(2); in vp8_denoiser_filter_neon() 66 const uint8x16_t v_level1_threshold = vmovq_n_u8(4 + shift_inc); in vp8_denoiser_filter_neon() 67 const uint8x16_t v_level2_threshold = vdupq_n_u8(8); in vp8_denoiser_filter_neon() 68 const uint8x16_t v_level3_threshold = vdupq_n_u8(16); in vp8_denoiser_filter_neon() 75 const uint8x16_t v_sig = vld1q_u8(sig); in vp8_denoiser_filter_neon() 76 const uint8x16_t v_mc_running_avg_y = vld1q_u8(mc_running_avg_y); in vp8_denoiser_filter_neon() 79 const uint8x16_t v_abs_diff = vabdq_u8(v_sig, v_mc_running_avg_y); in vp8_denoiser_filter_neon() 80 const uint8x16_t v_diff_pos_mask = vcltq_u8(v_sig, v_mc_running_avg_y); in vp8_denoiser_filter_neon() [all …]
|
/external/libvpx/libvpx/vpx_dsp/ppc/ |
D | intrapred_vsx.c | 16 const uint8x16_t d = vec_vsx_ld(0, above); in vpx_v_predictor_16x16_vsx() 27 const uint8x16_t d0 = vec_vsx_ld(0, above); in vpx_v_predictor_32x32_vsx() 28 const uint8x16_t d1 = vec_vsx_ld(16, above); in vpx_v_predictor_32x32_vsx() 44 const uint8x16_t d = vec_vsx_ld(0, left); 45 const uint8x16_t v0 = vec_splat(d, 0); 46 const uint8x16_t v1 = vec_splat(d, 1); 47 const uint8x16_t v2 = vec_splat(d, 2); 48 const uint8x16_t v3 = vec_splat(d, 3); 52 vec_vsx_st(vec_sel(v0, vec_vsx_ld(0, dst), (uint8x16_t)mask4), 0, dst); 54 vec_vsx_st(vec_sel(v1, vec_vsx_ld(0, dst), (uint8x16_t)mask4), 0, dst); [all …]
|
D | types_vsx.h | 17 typedef vector unsigned char uint8x16_t; typedef 27 static const uint8x16_t xxpermdi0_perm = { 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 30 static const uint8x16_t xxpermdi1_perm = { 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 33 static const uint8x16_t xxpermdi2_perm = { 0x08, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 36 static const uint8x16_t xxpermdi3_perm = { 0x08, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 47 (uint16x8_t) vec_mergeh(vec_splat_u8(0), (uint8x16_t)v) 49 (uint16x8_t) vec_mergel(vec_splat_u8(0), (uint8x16_t)v) 51 (int16x8_t) vec_mergeh(vec_splat_u8(0), (uint8x16_t)v) 53 (int16x8_t) vec_mergel(vec_splat_u8(0), (uint8x16_t)v) 59 (uint16x8_t) vec_mergeh((uint8x16_t)v, vec_splat_u8(0)) [all …]
|
D | deblock_vsx.c | 18 static const uint8x16_t load_merge = { 0x00, 0x02, 0x04, 0x06, 0x08, 0x0A, 22 static const uint8x16_t st8_perm = { 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 26 static INLINE uint8x16_t apply_filter(uint8x16_t ctx[4], uint8x16_t v, in apply_filter() 27 uint8x16_t filter) { in apply_filter() 28 const uint8x16_t k1 = vec_avg(ctx[0], ctx[1]); in apply_filter() 29 const uint8x16_t k2 = vec_avg(ctx[3], ctx[2]); in apply_filter() 30 const uint8x16_t k3 = vec_avg(k1, k2); in apply_filter() 31 const uint8x16_t f_a = vec_max(vec_absd(v, ctx[0]), vec_absd(v, ctx[1])); in apply_filter() 32 const uint8x16_t f_b = vec_max(vec_absd(v, ctx[2]), vec_absd(v, ctx[3])); in apply_filter() 37 static INLINE void vert_ctx(uint8x16_t ctx[4], int col, uint8_t *src, in vert_ctx() [all …]
|
/external/libvpx/libvpx/vp9/encoder/arm/neon/ |
D | vp9_denoiser_neon.c | 36 const uint8x16_t v_level1_threshold, const uint8x16_t v_level2_threshold, in denoiser_16x1_neon() 37 const uint8x16_t v_level3_threshold, const uint8x16_t v_level1_adjustment, in denoiser_16x1_neon() 38 const uint8x16_t v_delta_level_1_and_2, in denoiser_16x1_neon() 39 const uint8x16_t v_delta_level_2_and_3, int8x16_t v_sum_diff_total) { in denoiser_16x1_neon() 40 const uint8x16_t v_sig = vld1q_u8(sig); in denoiser_16x1_neon() 41 const uint8x16_t v_mc_running_avg_y = vld1q_u8(mc_running_avg_y); in denoiser_16x1_neon() 44 const uint8x16_t v_abs_diff = vabdq_u8(v_sig, v_mc_running_avg_y); in denoiser_16x1_neon() 45 const uint8x16_t v_diff_pos_mask = vcltq_u8(v_sig, v_mc_running_avg_y); in denoiser_16x1_neon() 46 const uint8x16_t v_diff_neg_mask = vcgtq_u8(v_sig, v_mc_running_avg_y); in denoiser_16x1_neon() 49 const uint8x16_t v_level1_mask = vcleq_u8(v_level1_threshold, v_abs_diff); in denoiser_16x1_neon() [all …]
|
/external/libvpx/libvpx/vpx_dsp/arm/ |
D | intrapred_neon.c | 95 const uint8x16_t above_and_left = vcombine_u8(above_u8, left_u8); in vpx_dc_predictor_8x8_neon() 133 const uint8x16_t ref_u8 = vld1q_u8(ref); in dc_sum_16() 142 const uint8x16_t dc_dup = vdupq_lane_u8(dc, 0); in dc_store_16x16() 151 const uint8x16_t ref0 = vld1q_u8(above); in vpx_dc_predictor_16x16_neon() 152 const uint8x16_t ref1 = vld1q_u8(left); in vpx_dc_predictor_16x16_neon() 314 const uint8x16_t above_right, uint8x16_t *row) { in d45_store_16() 322 const uint8x16_t A0 = vld1q_u8(above); in vpx_d45_predictor_16x16_neon() 323 const uint8x16_t above_right = vdupq_lane_u8(vget_high_u8(A0), 7); in vpx_d45_predictor_16x16_neon() 324 const uint8x16_t A1 = vextq_u8(A0, above_right, 1); in vpx_d45_predictor_16x16_neon() 325 const uint8x16_t A2 = vextq_u8(A0, above_right, 2); in vpx_d45_predictor_16x16_neon() [all …]
|
D | sad_neon.c | 22 const uint8x16_t src_u8 = load_unaligned_u8q(src_ptr, src_stride); in vpx_sad4x4_neon() 23 const uint8x16_t ref_u8 = load_unaligned_u8q(ref_ptr, ref_stride); in vpx_sad4x4_neon() 32 const uint8x16_t src_u8 = load_unaligned_u8q(src_ptr, src_stride); in vpx_sad4x4_avg_neon() 33 const uint8x16_t ref_u8 = load_unaligned_u8q(ref_ptr, ref_stride); in vpx_sad4x4_avg_neon() 34 const uint8x16_t second_pred_u8 = vld1q_u8(second_pred); in vpx_sad4x4_avg_neon() 35 const uint8x16_t avg = vrhaddq_u8(ref_u8, second_pred_u8); in vpx_sad4x4_avg_neon() 46 const uint8x16_t src_u8 = load_unaligned_u8q(src_ptr, src_stride); in vpx_sad4x8_neon() 47 const uint8x16_t ref_u8 = load_unaligned_u8q(ref_ptr, ref_stride); in vpx_sad4x8_neon() 63 const uint8x16_t src_u8 = load_unaligned_u8q(src_ptr, src_stride); in vpx_sad4x8_avg_neon() 64 const uint8x16_t ref_u8 = load_unaligned_u8q(ref_ptr, ref_stride); in vpx_sad4x8_avg_neon() [all …]
|
D | idct32x32_1_add_neon.c | 18 const uint8x16_t res) { in idct32x32_1_add_pos_kernel() 19 const uint8x16_t a0 = vld1q_u8(*dest); in idct32x32_1_add_pos_kernel() 20 const uint8x16_t a1 = vld1q_u8(*dest + 16); in idct32x32_1_add_pos_kernel() 21 const uint8x16_t b0 = vqaddq_u8(a0, res); in idct32x32_1_add_pos_kernel() 22 const uint8x16_t b1 = vqaddq_u8(a1, res); in idct32x32_1_add_pos_kernel() 29 const uint8x16_t res) { in idct32x32_1_add_neg_kernel() 30 const uint8x16_t a0 = vld1q_u8(*dest); in idct32x32_1_add_neg_kernel() 31 const uint8x16_t a1 = vld1q_u8(*dest + 16); in idct32x32_1_add_neg_kernel() 32 const uint8x16_t b0 = vqsubq_u8(a0, res); in idct32x32_1_add_neg_kernel() 33 const uint8x16_t b1 = vqsubq_u8(a1, res); in idct32x32_1_add_neg_kernel() [all …]
|
D | avg_neon.c | 23 const uint8x16_t b = load_unaligned_u8q(a, a_stride); in vpx_avg_4x4_neon() 87 const uint8x16_t vec_row1 = vld1q_u8(ref); in vpx_int_pro_row_neon() 88 const uint8x16_t vec_row2 = vld1q_u8(ref + ref_stride); in vpx_int_pro_row_neon() 89 const uint8x16_t vec_row3 = vld1q_u8(ref + ref_stride * 2); in vpx_int_pro_row_neon() 90 const uint8x16_t vec_row4 = vld1q_u8(ref + ref_stride * 3); in vpx_int_pro_row_neon() 91 const uint8x16_t vec_row5 = vld1q_u8(ref + ref_stride * 4); in vpx_int_pro_row_neon() 92 const uint8x16_t vec_row6 = vld1q_u8(ref + ref_stride * 5); in vpx_int_pro_row_neon() 93 const uint8x16_t vec_row7 = vld1q_u8(ref + ref_stride * 6); in vpx_int_pro_row_neon() 94 const uint8x16_t vec_row8 = vld1q_u8(ref + ref_stride * 7); in vpx_int_pro_row_neon() 136 const uint8x16_t vec_row = vld1q_u8(ref); in vpx_int_pro_col_neon() [all …]
|
/external/webp/src/dsp/ |
D | dec_neon.c | 47 uint8x16_t* const p1, in Load4x16_NEON() 48 uint8x16_t* const p0, in Load4x16_NEON() 49 uint8x16_t* const q0, in Load4x16_NEON() 50 uint8x16_t* const q1) { in Load4x16_NEON() 69 uint8x16_t* const p1, in Load4x16_NEON() 70 uint8x16_t* const p0, in Load4x16_NEON() 71 uint8x16_t* const q0, in Load4x16_NEON() 72 uint8x16_t* const q1) { in Load4x16_NEON() 115 uint8x16_t* const p3, uint8x16_t* const p2, uint8x16_t* const p1, in Load8x16_NEON() 116 uint8x16_t* const p0, uint8x16_t* const q0, uint8x16_t* const q1, in Load8x16_NEON() [all …]
|
D | lossless_neon.c | 35 const uint8x16_t tmp = pixel.val[0]; in ConvertBGRAToRGBA_NEON() 210 const uint8x16_t black = vreinterpretq_u8_u32(vdupq_n_u32(ARGB_BLACK)); in PredictorAdd0_NEON() 212 const uint8x16_t src = LOADQ_U32P_AS_U8(&in[i]); in PredictorAdd0_NEON() 213 const uint8x16_t res = vaddq_u8(src, black); in PredictorAdd0_NEON() 223 const uint8x16_t zero = LOADQ_U32_AS_U8(0); in PredictorAdd1_NEON() 226 const uint8x16_t src = LOADQ_U32P_AS_U8(&in[i]); in PredictorAdd1_NEON() 228 const uint8x16_t shift0 = vextq_u8(zero, src, 12); in PredictorAdd1_NEON() 230 const uint8x16_t sum0 = vaddq_u8(src, shift0); in PredictorAdd1_NEON() 232 const uint8x16_t shift1 = vextq_u8(zero, sum0, 8); in PredictorAdd1_NEON() 234 const uint8x16_t sum1 = vaddq_u8(sum0, shift1); in PredictorAdd1_NEON() [all …]
|
/external/webrtc/modules/video_processing/util/ |
D | denoiser_filter_neon.cc | 71 uint8x16_t qtmp; in CopyMem16x16() 105 const uint8x16_t v_level1_adjustment = vmovq_n_u8( in MbDenoise() 107 const uint8x16_t v_delta_level_1_and_2 = vdupq_n_u8(1); in MbDenoise() 108 const uint8x16_t v_delta_level_2_and_3 = vdupq_n_u8(2); in MbDenoise() 109 const uint8x16_t v_level1_threshold = vmovq_n_u8(4 + shift_inc); in MbDenoise() 110 const uint8x16_t v_level2_threshold = vdupq_n_u8(8); in MbDenoise() 111 const uint8x16_t v_level3_threshold = vdupq_n_u8(16); in MbDenoise() 117 const uint8x16_t v_sig = vld1q_u8(sig); in MbDenoise() 118 const uint8x16_t v_mc_running_avg_y = vld1q_u8(mc_running_avg_y); in MbDenoise() 121 const uint8x16_t v_abs_diff = vabdq_u8(v_sig, v_mc_running_avg_y); in MbDenoise() [all …]
|
/external/llvm-project/clang/test/CodeGen/ |
D | aarch64-neon-tbl.c | 430 int8x16_t test_vqtbx1q_s8(int8x16_t a, int8x16_t b, uint8x16_t c) { in test_vqtbx1q_s8() 517 uint8x8_t test_vqtbl1_u8(uint8x16_t a, uint8x8_t b) { in test_vqtbl1_u8() 673 uint8x16_t test_vqtbl1q_u8(uint8x16_t a, uint8x16_t b) { in test_vqtbl1q_u8() 694 uint8x16_t test_vqtbl2q_u8(uint8x16x2_t a, uint8x16_t b) { in test_vqtbl2q_u8() 718 uint8x16_t test_vqtbl3q_u8(uint8x16x3_t a, uint8x16_t b) { in test_vqtbl3q_u8() 745 uint8x16_t test_vqtbl4q_u8(uint8x16x4_t a, uint8x16_t b) { in test_vqtbl4q_u8() 849 uint8x8_t test_vqtbx1_u8(uint8x8_t a, uint8x16_t b, uint8x8_t c) { in test_vqtbx1_u8() 928 uint8x16_t test_vqtbx1q_u8(uint8x16_t a, uint8x16_t b, uint8x16_t c) { in test_vqtbx1q_u8() 949 uint8x16_t test_vqtbx2q_u8(uint8x16_t a, uint8x16x2_t b, uint8x16_t c) { in test_vqtbx2q_u8() 973 uint8x16_t test_vqtbx3q_u8(uint8x16_t a, uint8x16x3_t b, uint8x16_t c) { in test_vqtbx3q_u8() [all …]
|
D | arm64_crypto.c | 7 uint8x16_t test_aese(uint8x16_t data, uint8x16_t key) { in test_aese() 13 uint8x16_t test_aesd(uint8x16_t data, uint8x16_t key) { in test_aesd() 19 uint8x16_t test_aesmc(uint8x16_t data, uint8x16_t key) { in test_aesmc() 25 uint8x16_t test_aesimc(uint8x16_t data, uint8x16_t key) { in test_aesimc()
|
D | arm-cde-reinterpret.c | 16 int8x16_t test_s8(uint8x16_t x) { in test_s8() 26 uint16x8_t test_u16(uint8x16_t x) { in test_u16() 36 int32x4_t test_s32(uint8x16_t x) { in test_s32() 46 uint32x4_t test_u32(uint8x16_t x) { in test_u32() 56 int64x2_t test_s64(uint8x16_t x) { in test_s64() 66 uint64x2_t test_u64(uint8x16_t x) { in test_u64() 76 float16x8_t test_f16(uint8x16_t x) { in test_f16() 86 float32x4_t test_f32(uint8x16_t x) { in test_f32()
|
D | neon-crypto.c | 13 uint8x16_t test_vaeseq_u8(uint8x16_t data, uint8x16_t key) { in test_vaeseq_u8() 20 uint8x16_t test_vaesdq_u8(uint8x16_t data, uint8x16_t key) { in test_vaesdq_u8() 26 uint8x16_t test_vaesmcq_u8(uint8x16_t data) { in test_vaesmcq_u8() 32 uint8x16_t test_vaesimcq_u8(uint8x16_t data) { in test_vaesimcq_u8()
|
/external/clang/test/CodeGen/ |
D | aarch64-neon-tbl.c | 517 uint8x8_t test_vqtbl1_u8(uint8x16_t a, uint8x8_t b) { in test_vqtbl1_u8() 673 uint8x16_t test_vqtbl1q_u8(uint8x16_t a, uint8x16_t b) { in test_vqtbl1q_u8() 694 uint8x16_t test_vqtbl2q_u8(uint8x16x2_t a, uint8x16_t b) { in test_vqtbl2q_u8() 718 uint8x16_t test_vqtbl3q_u8(uint8x16x3_t a, uint8x16_t b) { in test_vqtbl3q_u8() 745 uint8x16_t test_vqtbl4q_u8(uint8x16x4_t a, uint8x16_t b) { in test_vqtbl4q_u8() 849 uint8x8_t test_vqtbx1_u8(uint8x8_t a, uint8x16_t b, uint8x8_t c) { in test_vqtbx1_u8() 928 uint8x16_t test_vqtbx1q_u8(uint8x16_t a, uint8x16_t b, uint8x16_t c) { in test_vqtbx1q_u8() 949 uint8x16_t test_vqtbx2q_u8(uint8x16_t a, uint8x16x2_t b, uint8x16_t c) { in test_vqtbx2q_u8() 973 uint8x16_t test_vqtbx3q_u8(uint8x16_t a, uint8x16x3_t b, uint8x16_t c) { in test_vqtbx3q_u8() 1000 uint8x16_t test_vqtbx4q_u8(uint8x16_t a, uint8x16x4_t b, uint8x16_t c) { in test_vqtbx4q_u8() [all …]
|
D | arm64_crypto.c | 6 uint8x16_t test_aese(uint8x16_t data, uint8x16_t key) { in test_aese() 12 uint8x16_t test_aesd(uint8x16_t data, uint8x16_t key) { in test_aesd() 18 uint8x16_t test_aesmc(uint8x16_t data, uint8x16_t key) { in test_aesmc() 24 uint8x16_t test_aesimc(uint8x16_t data, uint8x16_t key) { in test_aesimc()
|
D | neon-crypto.c | 13 uint8x16_t test_vaeseq_u8(uint8x16_t data, uint8x16_t key) { in test_vaeseq_u8() 20 uint8x16_t test_vaesdq_u8(uint8x16_t data, uint8x16_t key) { in test_vaesdq_u8() 26 uint8x16_t test_vaesmcq_u8(uint8x16_t data) { in test_vaesmcq_u8() 32 uint8x16_t test_vaesimcq_u8(uint8x16_t data) { in test_vaesimcq_u8()
|
/external/llvm-project/clang/test/CodeGen/arm-mve-intrinsics/ |
D | vector-shift-imm-dyadic.c | 40 uint8x16_t test_vshrnbq_n_u16(uint8x16_t a, uint16x8_t b) in test_vshrnbq_n_u16() 96 uint8x16_t test_vshrntq_n_u16(uint8x16_t a, uint16x8_t b) in test_vshrntq_n_u16() 158 uint8x16_t test_vshrnbq_m_n_u16(uint8x16_t a, uint16x8_t b, mve_pred16_t p) in test_vshrnbq_m_n_u16() 222 uint8x16_t test_vshrntq_m_n_u16(uint8x16_t a, uint16x8_t b, mve_pred16_t p) in test_vshrntq_m_n_u16() 280 uint8x16_t test_vrshrnbq_n_u16(uint8x16_t a, uint16x8_t b) in test_vrshrnbq_n_u16() 336 uint8x16_t test_vrshrntq_n_u16(uint8x16_t a, uint16x8_t b) in test_vrshrntq_n_u16() 398 uint8x16_t test_vrshrnbq_m_n_u16(uint8x16_t a, uint16x8_t b, mve_pred16_t p) in test_vrshrnbq_m_n_u16() 462 uint8x16_t test_vrshrntq_m_n_u16(uint8x16_t a, uint16x8_t b, mve_pred16_t p) in test_vrshrntq_m_n_u16() 520 uint8x16_t test_vqshrnbq_n_u16(uint8x16_t a, uint16x8_t b) in test_vqshrnbq_n_u16() 576 uint8x16_t test_vqshrntq_n_u16(uint8x16_t a, uint16x8_t b) in test_vqshrntq_n_u16() [all …]
|
D | vqmovn.c | 40 uint8x16_t test_vqmovnbq_u16(uint8x16_t a, uint16x8_t b) in test_vqmovnbq_u16() 96 uint8x16_t test_vqmovntq_u16(uint8x16_t a, uint16x8_t b) in test_vqmovntq_u16() 124 uint8x16_t test_vqmovunbq_s16(uint8x16_t a, int16x8_t b) in test_vqmovunbq_s16() 152 uint8x16_t test_vqmovuntq_s16(uint8x16_t a, int16x8_t b) in test_vqmovuntq_s16() 214 uint8x16_t test_vqmovnbq_m_u16(uint8x16_t a, uint16x8_t b, mve_pred16_t p) in test_vqmovnbq_m_u16() 278 uint8x16_t test_vqmovntq_m_u16(uint8x16_t a, uint16x8_t b, mve_pred16_t p) in test_vqmovntq_m_u16() 310 uint8x16_t test_vqmovunbq_m_s16(uint8x16_t a, int16x8_t b, mve_pred16_t p) in test_vqmovunbq_m_s16() 342 uint8x16_t test_vqmovuntq_m_s16(uint8x16_t a, int16x8_t b, mve_pred16_t p) in test_vqmovuntq_m_s16()
|
D | vrev.c | 26 uint8x16_t test_vrev16q_u8(uint8x16_t a) in test_vrev16q_u8() 54 uint8x16_t test_vrev32q_u8(uint8x16_t a) in test_vrev32q_u8() 124 uint8x16_t test_vrev64q_u8(uint8x16_t a) in test_vrev64q_u8() 240 uint8x16_t test_vrev16q_m_u8(uint8x16_t inactive, uint8x16_t a, mve_pred16_t p) in test_vrev16q_m_u8() 304 uint8x16_t test_vrev32q_m_u8(uint8x16_t inactive, uint8x16_t a, mve_pred16_t p) in test_vrev32q_m_u8() 416 uint8x16_t test_vrev64q_m_u8(uint8x16_t inactive, uint8x16_t a, mve_pred16_t p) in test_vrev64q_m_u8() 480 uint8x16_t test_vrev16q_x_u8(uint8x16_t a, mve_pred16_t p) in test_vrev16q_x_u8() 544 uint8x16_t test_vrev32q_x_u8(uint8x16_t a, mve_pred16_t p) in test_vrev32q_x_u8() 656 uint8x16_t test_vrev64q_x_u8(uint8x16_t a, mve_pred16_t p) in test_vrev64q_x_u8()
|
/external/libaom/libaom/aom_dsp/arm/ |
D | sad4d_neon.c | 35 static void sad_neon_64(const uint8x16_t vec_src_00, in sad_neon_64() 36 const uint8x16_t vec_src_16, in sad_neon_64() 37 const uint8x16_t vec_src_32, in sad_neon_64() 38 const uint8x16_t vec_src_48, const uint8_t *ref, in sad_neon_64() 41 const uint8x16_t vec_ref_00 = vld1q_u8(ref); in sad_neon_64() 42 const uint8x16_t vec_ref_16 = vld1q_u8(ref + 16); in sad_neon_64() 43 const uint8x16_t vec_ref_32 = vld1q_u8(ref + 32); in sad_neon_64() 44 const uint8x16_t vec_ref_48 = vld1q_u8(ref + 48); in sad_neon_64() 66 static void sad_neon_32(const uint8x16_t vec_src_00, in sad_neon_32() 67 const uint8x16_t vec_src_16, const uint8_t *ref, in sad_neon_32() [all …]
|
/external/XNNPACK/src/u8-clamp/ |
D | neon-x64.c | 24 const uint8x16_t voutput_max = vld1q_dup_u8(¶ms->neon.max); in xnn_u8_clamp_ukernel__neon_x64() 25 const uint8x16_t voutput_min = vld1q_dup_u8(¶ms->neon.min); in xnn_u8_clamp_ukernel__neon_x64() 28 const uint8x16_t vx0 = vld1q_u8(x); x += 16; in xnn_u8_clamp_ukernel__neon_x64() 29 const uint8x16_t vx1 = vld1q_u8(x); x += 16; in xnn_u8_clamp_ukernel__neon_x64() 30 const uint8x16_t vx2 = vld1q_u8(x); x += 16; in xnn_u8_clamp_ukernel__neon_x64() 31 const uint8x16_t vx3 = vld1q_u8(x); x += 16; in xnn_u8_clamp_ukernel__neon_x64() 33 const uint8x16_t vy0 = vminq_u8(vmaxq_u8(vx0, voutput_min), voutput_max); in xnn_u8_clamp_ukernel__neon_x64() 34 const uint8x16_t vy1 = vminq_u8(vmaxq_u8(vx1, voutput_min), voutput_max); in xnn_u8_clamp_ukernel__neon_x64() 35 const uint8x16_t vy2 = vminq_u8(vmaxq_u8(vx2, voutput_min), voutput_max); in xnn_u8_clamp_ukernel__neon_x64() 36 const uint8x16_t vy3 = vminq_u8(vmaxq_u8(vx3, voutput_min), voutput_max); in xnn_u8_clamp_ukernel__neon_x64()
|