/external/libvpx/libvpx/third_party/libyuv/source/ |
compare_msa.cc | 30 v16u8 src0, src1, src2, src3; 34 src0 = (v16u8)__msa_ld_b((v16i8*)src_a, 0); 35 src1 = (v16u8)__msa_ld_b((v16i8*)src_a, 16); 36 src2 = (v16u8)__msa_ld_b((v16i8*)src_b, 0); 37 src3 = (v16u8)__msa_ld_b((v16i8*)src_b, 16); 57 v16u8 src0, src1, src2, src3; 63 src0 = (v16u8)__msa_ld_b((v16i8*)src_a, 0); 64 src1 = (v16u8)__msa_ld_b((v16i8*)src_a, 16); 65 src2 = (v16u8)__msa_ld_b((v16i8*)src_b, 0); 66 src3 = (v16u8)__msa_ld_b((v16i8*)src_b, 16) [all...] |
row_msa.cc | 48 out_y = (v16u8)__msa_insert_d((v2i64)zero_m, 0, (int64_t)y_m); \ 49 out_u = (v16u8)__msa_insert_w(zero_m, 0, (int32_t)u_m); \ 50 out_v = (v16u8)__msa_insert_w(zero_m, 0, (int32_t)v_m); \ 127 v16u8 dst0_m, dst1_m; \ 130 dst0_m = (v16u8)__msa_ilvr_h(vec1_m, vec0_m); \ 131 dst1_m = (v16u8)__msa_ilvl_h(vec1_m, vec0_m); \ 139 v16u8 vec0_m, vec1_m, vec2_m, vec3_m; \ 142 vec0_m = (v16u8)__msa_pckev_h((v8i16)argb1, (v8i16)argb0); \ 143 vec1_m = (v16u8)__msa_pckev_h((v8i16)argb3, (v8i16)argb2); \ 144 vec2_m = (v16u8)__msa_pckod_h((v8i16)argb1, (v8i16)argb0); [all...] |
scale_msa.cc | 37 v16u8 src0, src1, dst0; 41 src0 = (v16u8)__msa_ld_b((v16i8*)src_argb, 0); 42 src1 = (v16u8)__msa_ld_b((v16i8*)src_argb, 16); 43 dst0 = (v16u8)__msa_pckod_w((v4i32)src1, (v4i32)src0); 55 v16u8 src0, src1, vec0, vec1, dst0; 59 src0 = (v16u8)__msa_ld_b((v16i8*)src_argb, 0); 60 src1 = (v16u8)__msa_ld_b((v16i8*)src_argb, 16); 61 vec0 = (v16u8)__msa_pckev_w((v4i32)src1, (v4i32)src0); 62 vec1 = (v16u8)__msa_pckod_w((v4i32)src1, (v4i32)src0); 63 dst0 = (v16u8)__msa_aver_u_b((v16u8)vec0, (v16u8)vec1) [all...] |
rotate_msa.cc | 24 out0 = (v16u8)__msa_ilvr_b((v16i8)in1, (v16i8)in0); \ 25 out1 = (v16u8)__msa_ilvl_b((v16i8)in1, (v16i8)in0); \ 26 out2 = (v16u8)__msa_ilvr_b((v16i8)in3, (v16i8)in2); \ 27 out3 = (v16u8)__msa_ilvl_b((v16i8)in3, (v16i8)in2); \ 32 out0 = (v16u8)__msa_ilvr_h((v8i16)in1, (v8i16)in0); \ 33 out1 = (v16u8)__msa_ilvl_h((v8i16)in1, (v8i16)in0); \ 34 out2 = (v16u8)__msa_ilvr_h((v8i16)in3, (v8i16)in2); \ 35 out3 = (v16u8)__msa_ilvl_h((v8i16)in3, (v8i16)in2); \ 40 out0 = (v16u8)__msa_ilvr_w((v4i32)in1, (v4i32)in0); \ 41 out1 = (v16u8)__msa_ilvl_w((v4i32)in1, (v4i32)in0); [all...] |
/external/libaom/libaom/aom_dsp/mips/ |
add_noise_msa.c | 29 v16u8 temp00, temp01, black_clamp, white_clamp; 30 v16u8 pos0, ref0, pos1, ref1; 37 black_clamp = (v16u8)__msa_fill_b(blackclamp[0]); 38 white_clamp = (v16u8)__msa_fill_b(whiteclamp[0]); 45 temp00 = (v16u8)(temp00_s < pos0); 46 pos0 = (v16u8)__msa_bmnz_v((v16u8)pos0, (v16u8)temp00_s, temp00); 49 pos1 = (v16u8)__msa_bmnz_v((v16u8)pos1, (v16u8)temp01_s, temp01) [all...] |
loopfilter_4_msa.c | 19 v16u8 mask, hev, flat, thresh, b_limit, limit; 20 v16u8 p3, p2, p1, p0, q3, q2, q1, q0, p1_out, p0_out, q0_out, q1_out; 25 thresh = (v16u8)__msa_fill_b(*thresh_ptr); 26 b_limit = (v16u8)__msa_fill_b(*b_limit_ptr); 27 limit = (v16u8)__msa_fill_b(*limit_ptr); 47 v16u8 mask, hev, flat, thresh0, b_limit0, limit0, thresh1, b_limit1, limit1; 48 v16u8 p3, p2, p1, p0, q3, q2, q1, q0; 53 thresh0 = (v16u8)__msa_fill_b(*thresh0_ptr); 54 thresh1 = (v16u8)__msa_fill_b(*thresh1_ptr); 55 thresh0 = (v16u8)__msa_ilvr_d((v2i64)thresh1, (v2i64)thresh0) [all...] |
loopfilter_8_msa.c | 19 v16u8 mask, hev, flat, thresh, b_limit, limit; 20 v16u8 p3, p2, p1, p0, q3, q2, q1, q0; 21 v16u8 p2_out, p1_out, p0_out, q0_out, q1_out, q2_out; 29 thresh = (v16u8)__msa_fill_b(*thresh_ptr); 30 b_limit = (v16u8)__msa_fill_b(*b_limit_ptr); 31 limit = (v16u8)__msa_fill_b(*limit_ptr); 38 flat = (v16u8)__msa_ilvr_d((v2i64)zero, (v2i64)flat); 58 p2_out = __msa_bmnz_v(p2, (v16u8)p2_filter8, flat); 59 p1_out = __msa_bmnz_v(p1_out, (v16u8)p1_filter8, flat); 60 p0_out = __msa_bmnz_v(p0_out, (v16u8)p0_filter8, flat) [all...] |
loopfilter_16_msa.c | 19 v16u8 p3, p2, p1, p0, q3, q2, q1, q0; 20 v16u8 p2_out, p1_out, p0_out, q0_out, q1_out, q2_out; 21 v16u8 flat, mask, hev, thresh, b_limit, limit; 26 v16u8 zero = { 0 }; 31 thresh = (v16u8)__msa_fill_b(*thresh_ptr); 32 b_limit = (v16u8)__msa_fill_b(*b_limit_ptr); 33 limit = (v16u8)__msa_fill_b(*limit_ptr); 64 p2_out = __msa_bmnz_v(p2, (v16u8)p2_filt8_r, flat); 65 p1_out = __msa_bmnz_v(p1_out, (v16u8)p1_filt8_r, flat); 66 p0_out = __msa_bmnz_v(p0_out, (v16u8)p0_filt8_r, flat) [all...] |
loopfilter_msa.h | 54 q0_out = __msa_xori_b((v16u8)q0_m, 0x80); \ 56 p0_out = __msa_xori_b((v16u8)p0_m, 0x80); \ 59 hev_in = __msa_xori_b((v16u8)hev_in, 0xff); \ 63 q1_out = __msa_xori_b((v16u8)q1_m, 0x80); \ 65 p1_out = __msa_xori_b((v16u8)p1_m, 0x80); \ 112 q0_out = __msa_xori_b((v16u8)q0_m, 0x80); \ 114 p0_out = __msa_xori_b((v16u8)p0_m, 0x80); \ 117 hev_in = __msa_xori_b((v16u8)hev_in, 0xff); \ 121 q1_out = __msa_xori_b((v16u8)q1_m, 0x80); \ 123 p1_out = __msa_xori_b((v16u8)p1_m, 0x80); [all...] |
/external/libyuv/files/source/ |
scale_msa.cc | 29 v16u8 src0, src1, dst0; 33 src0 = (v16u8)__msa_ld_b((v16i8*)src_argb, 0); 34 src1 = (v16u8)__msa_ld_b((v16i8*)src_argb, 16); 35 dst0 = (v16u8)__msa_pckod_w((v4i32)src1, (v4i32)src0); 47 v16u8 src0, src1, vec0, vec1, dst0; 51 src0 = (v16u8)__msa_ld_b((v16i8*)src_argb, 0); 52 src1 = (v16u8)__msa_ld_b((v16i8*)src_argb, 16); 53 vec0 = (v16u8)__msa_pckev_w((v4i32)src1, (v4i32)src0); 54 vec1 = (v16u8)__msa_pckod_w((v4i32)src1, (v4i32)src0); 55 dst0 = (v16u8)__msa_aver_u_b((v16u8)vec0, (v16u8)vec1) [all...] |
row_msa.cc | 48 out_y = (v16u8)__msa_insert_d((v2i64)zero_m, 0, (int64)y_m); \ 49 out_u = (v16u8)__msa_insert_w(zero_m, 0, (int32)u_m); \ 50 out_v = (v16u8)__msa_insert_w(zero_m, 0, (int32)v_m); \ 127 v16u8 dst0_m, dst1_m; \ 130 dst0_m = (v16u8)__msa_ilvr_h(vec1_m, vec0_m); \ 131 dst1_m = (v16u8)__msa_ilvl_h(vec1_m, vec0_m); \ 139 v16u8 vec0_m, vec1_m, vec2_m, vec3_m; \ 142 vec0_m = (v16u8)__msa_pckev_h((v8i16)argb1, (v8i16)argb0); \ 143 vec1_m = (v16u8)__msa_pckev_h((v8i16)argb3, (v8i16)argb2); \ 144 vec2_m = (v16u8)__msa_pckod_h((v8i16)argb1, (v8i16)argb0); [all...] |
rotate_msa.cc | 24 out0 = (v16u8)__msa_ilvr_b((v16i8)in1, (v16i8)in0); \ 25 out1 = (v16u8)__msa_ilvl_b((v16i8)in1, (v16i8)in0); \ 26 out2 = (v16u8)__msa_ilvr_b((v16i8)in3, (v16i8)in2); \ 27 out3 = (v16u8)__msa_ilvl_b((v16i8)in3, (v16i8)in2); \ 32 out0 = (v16u8)__msa_ilvr_h((v8i16)in1, (v8i16)in0); \ 33 out1 = (v16u8)__msa_ilvl_h((v8i16)in1, (v8i16)in0); \ 34 out2 = (v16u8)__msa_ilvr_h((v8i16)in3, (v8i16)in2); \ 35 out3 = (v16u8)__msa_ilvl_h((v8i16)in3, (v8i16)in2); \ 40 out0 = (v16u8)__msa_ilvr_w((v4i32)in1, (v4i32)in0); \ 41 out1 = (v16u8)__msa_ilvl_w((v4i32)in1, (v4i32)in0); [all...] |
/external/libvpx/libvpx/vp8/encoder/mips/msa/ |
denoising_msa.c | 29 v16u8 src0, src1, src2, src3, src4, src5, src6, src7; 30 v16u8 src8, src9, src10, src11, src12, src13, src14, src15; 31 v16u8 mc_running_avg_y0, running_avg_y, sig0; 32 v16u8 mc_running_avg_y1, running_avg_y1, sig1; 33 v16u8 coeff0, coeff1; 109 adjust0 = (v8i16)__msa_bmnz_v((v16u8)adjust0, (v16u8)diff0, (v16u8)temp2_h); 110 adjust1 = (v8i16)__msa_bmnz_v((v16u8)adjust1, (v16u8)diff1, (v16u8)temp3_h) [all...] |
/external/libvpx/libvpx/vpx_dsp/mips/ |
loopfilter_4_msa.c | 19 v16u8 mask, hev, flat, thresh, b_limit, limit; 20 v16u8 p3, p2, p1, p0, q3, q2, q1, q0, p1_out, p0_out, q0_out, q1_out; 25 thresh = (v16u8)__msa_fill_b(*thresh_ptr); 26 b_limit = (v16u8)__msa_fill_b(*b_limit_ptr); 27 limit = (v16u8)__msa_fill_b(*limit_ptr); 47 v16u8 mask, hev, flat, thresh0, b_limit0, limit0, thresh1, b_limit1, limit1; 48 v16u8 p3, p2, p1, p0, q3, q2, q1, q0; 53 thresh0 = (v16u8)__msa_fill_b(*thresh0_ptr); 54 thresh1 = (v16u8)__msa_fill_b(*thresh1_ptr); 55 thresh0 = (v16u8)__msa_ilvr_d((v2i64)thresh1, (v2i64)thresh0) [all...] |
loopfilter_8_msa.c | 19 v16u8 mask, hev, flat, thresh, b_limit, limit; 20 v16u8 p3, p2, p1, p0, q3, q2, q1, q0; 21 v16u8 p2_out, p1_out, p0_out, q0_out, q1_out, q2_out; 29 thresh = (v16u8)__msa_fill_b(*thresh_ptr); 30 b_limit = (v16u8)__msa_fill_b(*b_limit_ptr); 31 limit = (v16u8)__msa_fill_b(*limit_ptr); 38 flat = (v16u8)__msa_ilvr_d((v2i64)zero, (v2i64)flat); 58 p2_out = __msa_bmnz_v(p2, (v16u8)p2_filter8, flat); 59 p1_out = __msa_bmnz_v(p1_out, (v16u8)p1_filter8, flat); 60 p0_out = __msa_bmnz_v(p0_out, (v16u8)p0_filter8, flat) [all...] |
loopfilter_msa.h | 40 q0_out = __msa_xori_b((v16u8)q0_m, 0x80); \ 42 p0_out = __msa_xori_b((v16u8)p0_m, 0x80); \ 47 q1_out = __msa_xori_b((v16u8)q1_m, 0x80); \ 49 p1_out = __msa_xori_b((v16u8)p1_m, 0x80); \ 54 v16u8 tmp_flat4, p2_a_sub_p0, q2_a_sub_q0, p3_a_sub_p0, q3_a_sub_q0; \ 55 v16u8 zero_in = { 0 }; \ 68 flat_out = (tmp_flat4 < (v16u8)flat_out); \ 76 v16u8 tmp_flat5, zero_in = { 0 }; \ 77 v16u8 p4_a_sub_p0, q4_a_sub_q0, p5_a_sub_p0, q5_a_sub_q0; \ 78 v16u8 p6_a_sub_p0, q6_a_sub_q0, p7_a_sub_p0, q7_a_sub_q0; [all...] |
loopfilter_16_msa.c | 20 v16u8 p3, p2, p1, p0, q3, q2, q1, q0; 21 v16u8 p2_out, p1_out, p0_out, q0_out, q1_out, q2_out; 22 v16u8 flat, mask, hev, thresh, b_limit, limit; 27 v16u8 zero = { 0 }; 32 thresh = (v16u8)__msa_fill_b(*thresh_ptr); 33 b_limit = (v16u8)__msa_fill_b(*b_limit_ptr); 34 limit = (v16u8)__msa_fill_b(*limit_ptr); 65 p2_out = __msa_bmnz_v(p2, (v16u8)p2_filt8_r, flat); 66 p1_out = __msa_bmnz_v(p1_out, (v16u8)p1_filt8_r, flat); 67 p0_out = __msa_bmnz_v(p0_out, (v16u8)p0_filt8_r, flat) [all...] |
deblock_msa.c | 36 out0 = (v16u8)temp6; \ 37 out2 = (v16u8)temp7; \ 38 out4 = (v16u8)temp8; \ 39 out6 = (v16u8)temp9; \ 40 out9 = (v16u8)__msa_ilvl_d((v2i64)out8, (v2i64)out8); \ 41 out11 = (v16u8)__msa_ilvl_d((v2i64)out10, (v2i64)out10); \ 42 out13 = (v16u8)__msa_ilvl_d((v2i64)out12, (v2i64)out12); \ 43 out15 = (v16u8)__msa_ilvl_d((v2i64)out14, (v2i64)out14); \ 44 out1 = (v16u8)__msa_ilvl_d((v2i64)out0, (v2i64)out0); \ 45 out3 = (v16u8)__msa_ilvl_d((v2i64)out2, (v2i64)out2); [all...] |
/external/gemmlowp/fixedpoint/ |
fixedpoint_msa.h | 39 return reinterpret_cast<v4i32>(__builtin_msa_and_v(reinterpret_cast<v16u8>(a), 40 reinterpret_cast<v16u8>(b))); 45 return reinterpret_cast<v8i16>(__builtin_msa_and_v(reinterpret_cast<v16u8>(a), 46 reinterpret_cast<v16u8>(b))); 51 return reinterpret_cast<v4i32>(__builtin_msa_or_v(reinterpret_cast<v16u8>(a), 52 reinterpret_cast<v16u8>(b))); 57 return reinterpret_cast<v8i16>(__builtin_msa_or_v(reinterpret_cast<v16u8>(a), 58 reinterpret_cast<v16u8>(b))); 63 return reinterpret_cast<v4i32>(__builtin_msa_xor_v(reinterpret_cast<v16u8>(a), 64 reinterpret_cast<v16u8>(b))) [all...] |
/external/libvpx/libvpx/vp8/common/mips/msa/ |
loopfilter_filters_msa.c | 17 v16u8 p1_a_sub_q1, p0_a_sub_q0; \ 21 p1_a_sub_q1 = (v16u8)__msa_srli_b((v16i8)p1_a_sub_q1, 1); \ 24 mask = ((v16u8)mask <= b_limit); \ 50 q0 = __msa_xori_b((v16u8)q0_m, 0x80); \ 52 p0 = __msa_xori_b((v16u8)p0_m, 0x80); \ 57 q1 = __msa_xori_b((v16u8)q1_m, 0x80); \ 59 p1 = __msa_xori_b((v16u8)p1_m, 0x80); \ 86 q0_in = __msa_xori_b((v16u8)q0_m, 0x80); \ 87 p0_in = __msa_xori_b((v16u8)p0_m, 0x80); \ 138 q2 = __msa_xori_b((v16u8)q2_m, 0x80); [all...] |
/external/libpng/mips/ |
filter_msa_intrinsics.c | 251 #define LD_UB(...) LD_B(v16u8, __VA_ARGS__) 257 #define LD_UB2(...) LD_B2(v16u8, __VA_ARGS__) 263 #define LD_UB4(...) LD_B4(v16u8, __VA_ARGS__) 266 #define ST_UB(...) ST_B(v16u8, __VA_ARGS__) 272 #define ST_UB2(...) ST_B2(v16u8, __VA_ARGS__) 278 #define ST_UB4(...) ST_B4(v16u8, __VA_ARGS__) 307 out0 = (RTYPE) __msa_hsub_u_h((v16u8) in0, (v16u8) in0); \ 308 out1 = (RTYPE) __msa_hsub_u_h((v16u8) in1, (v16u8) in1); [all...] |
/external/webp/src/dsp/ |
enc_msa.c | 86 v16u8 srcl0, srcl1, src0 = { 0 }, src1 = { 0 }; 122 tmp4 = (v4i32)__msa_nor_v((v16u8)tmp5, (v16u8)tmp5); 124 tmp5 = (v4i32)__msa_and_v((v16u8)tmp5, (v16u8)tmp4); 263 const v16u8 A1 = { 0 }; 265 const v16u8 A = (v16u8)__msa_insert_d((v2i64)A1, 0, val_m); 266 const v16u8 B = SLDI_UB(A, A, 1); 267 const v16u8 C = SLDI_UB(A, A, 2) [all...] |
dec_msa.c | 138 v16u8 dest0, dest1, dest2, dest3; 186 q = __msa_xori_b((v16u8)q_m, 0x80); \ 187 p = __msa_xori_b((v16u8)p_m, 0x80); \ 205 q0 = __msa_xori_b((v16u8)q0_m, 0x80); \ 207 p0 = __msa_xori_b((v16u8)p0_m, 0x80); \ 212 q1 = __msa_xori_b((v16u8)q1_m, 0x80); \ 214 p1 = __msa_xori_b((v16u8)p1_m, 0x80); \ 263 v16u8 p3_asub_p2_m, p2_asub_p1_m, p1_asub_p0_m, q1_asub_q0_m; \ 264 v16u8 p1_asub_q1_m, p0_asub_q0_m, q3_asub_q2_m, q2_asub_q1_m; \ 265 v16u8 flat_out; [all...] |
lossless_msa.c | 25 v16u8 src0, src1, src2, src3, dst0, dst1, dst2; \ 35 v16u8 src0, src1, src2, dst0, dst1, dst2; \ 46 v16u8 src0, src1, src2 = { 0 }, dst0, dst1; \ 55 const v16u8 src0 = LD_UB(psrc); \ 56 const v16u8 dst0 = VSHF_UB(src0, src0, m); \ 117 v16u8 src0, dst0; 118 const v16u8 mask = { 2, 1, 0, 3, 6, 5, 4, 7, 10, 9, 8, 11, 14, 13, 12, 15 }; 121 v16u8 src1, dst1; 157 const v16u8 mask0 = { 0, 1, 2, 4, 5, 6, 8, 9, 10, 12, 13, 14, 159 const v16u8 mask1 = { 5, 6, 8, 9, 10, 12, 13, 14, 16, 17, 18, 20 [all...] |
upsampling_msa.c | 31 v16u8 t0, t1; \ 60 dst = (v16u8)__msa_pckev_b((v16i8)b1, (v16i8)b0); \ 69 dst = (v16u8)__msa_pckev_b((v16i8)b0, (v16i8)b0); \ 82 dst = (v16u8)__msa_pckev_b((v16i8)a1, (v16i8)a0); \ 92 dst = (v16u8)__msa_pckev_b((v16i8)a0, (v16i8)a0); \ 103 dst = (v16u8)__msa_pckev_b((v16i8)b1, (v16i8)b0); \ 112 dst = (v16u8)__msa_pckev_b((v16i8)b0, (v16i8)b0); \ 116 const v16u8 zero = { 0 }; \ 119 const v16u8 in_y = LD_UB(y); \ 120 const v16u8 in_u = LD_UB(u); [all...] |