A/A noise characterization

Same tenferro-rs build in both arms, balanced order. Use these spreads to declare noise.max_aa_relative_spread, noise.max_cov and the relative threshold in the confirmation config before any candidate run. No verdicts are produced here.

Case A/A statistic Spread Max CoV Rounds Round ratios
bdot_f64_b1024_m4n4k4_canonical_alloc_auto[faer]@t1 0.8942 0.1058 0.0112 4 0.7864, 0.7827, 1.0020, 1.2814
bdot_f64_b1024_m4n4k4_canonical_alloc_auto[faer]@t4 1.0003 0.0003 0.0116 4 1.2810, 0.9993, 1.0002, 1.0004
bdot_f64_b1024_m4n4k4_canonical_into_auto[faer]@t1 0.9958 0.0042 0.0089 4 1.0035, 1.0017, 0.9900, 0.9866
bdot_f64_b1024_m4n4k4_canonical_into_auto[faer]@t4 0.9874 0.0126 0.1386 4 0.9677, 1.0128, 0.8391, 1.0071
bdot_f64_b1024_m4n4k4_direct_alloc_auto[faer]@t1 1.0014 0.0014 0.0929 4 0.9714, 0.9992, 1.0036, 1.1872
bdot_f64_b1024_m4n4k4_direct_alloc_auto[faer]@t4 0.9975 0.0025 0.2404 4 1.1042, 0.9938, 1.0008, 0.9942
bdot_f64_b1024_m4n4k4_direct_into_auto[faer]@t1 1.0030 0.0030 0.0727 4 0.9938, 0.9924, 1.0122, 1.0174
bdot_f64_b1024_m4n4k4_direct_into_auto[faer]@t4 0.9400 0.0600 0.0900 4 0.9213, 0.9993, 0.9587, 0.8994
bdot_f64_b16_m64n64k64_direct_alloc_auto[faer]@t1 1.0001 0.0001 0.0916 4 0.8366, 0.9981, 1.0056, 1.0021
bdot_f64_b16_m64n64k64_direct_alloc_auto[faer]@t4 1.0005 0.0005 0.0109 4 0.9981, 1.0092, 0.9934, 1.0029
bdot_f64_b16_m64n64k64_direct_into_auto[faer]@t1 1.0063 0.0063 0.0663 4 0.9957, 1.0114, 1.2144, 1.0012
bdot_f64_b16_m64n64k64_direct_into_auto[faer]@t4 0.8004 0.1996 0.2313 4 1.2965, 0.5840, 1.0168, 0.5037
bdot_f64_b64_m4n4k4_direct_alloc_auto[faer]@t1 1.0233 0.0233 0.0078 4 1.0000, 1.0655, 0.9567, 1.0465
bdot_f64_b64_m4n4k4_direct_alloc_auto[faer]@t4 1.0084 0.0084 0.0271 4 1.0114, 1.0092, 1.0076, 0.9977
bdot_f64_b64_m4n4k4_direct_into_auto[faer]@t1 0.9993 0.0007 0.0110 4 1.0275, 0.9859, 1.0010, 0.9975
bdot_f64_b64_m4n4k4_direct_into_auto[faer]@t4 1.0033 0.0033 0.0978 4 0.8027, 0.9974, 1.0092, 1.0177
bdot_f64_b64_m64n1k64_direct_alloc_auto[faer]@t1 1.0264 0.0264 0.0752 4 1.0217, 1.0311, 1.0448, 1.0141
bdot_f64_b64_m64n1k64_direct_alloc_auto[faer]@t4 0.8888 0.1112 0.1008 4 0.8161, 0.7775, 0.9615, 1.0478
bdot_f64_b64_m64n1k64_direct_into_auto[faer]@t1 1.0075 0.0075 0.0236 4 1.0348, 0.9599, 1.0363, 0.9801
bdot_f64_b64_m64n1k64_direct_into_auto[faer]@t4 1.0656 0.0656 0.1041 4 1.0842, 0.9640, 1.1700, 1.0470
beinsum_f64_b1024_m4n4k4_direct_alloc_auto[faer]@t1 0.9968 0.0032 0.0521 4 1.0056, 1.0006, 0.9929, 0.9820
beinsum_f64_b1024_m4n4k4_direct_alloc_auto[faer]@t4 1.0055 0.0055 0.0236 4 1.0110, 1.0000, 1.0138, 0.9945
beinsum_f64_b1024_m4n4k4_direct_into_auto[faer]@t1 1.0058 0.0058 0.0495 4 1.0107, 1.0094, 0.9968, 1.0022
beinsum_f64_b1024_m4n4k4_direct_into_auto[faer]@t4 1.0054 0.0054 0.1048 4 1.0018, 1.0767, 1.0090, 0.9862
chain3_f64_n4_einsum_alloc[faer]@t1 0.9959 0.0041 0.4051 4 1.0178, 0.9782, 1.0137, 0.9162
chain3_f64_n4_einsum_alloc[faer]@t4 1.0172 0.0172 0.0518 4 1.0300, 0.8407, 1.0043, 1.2198
hadamard_f64_m64n64_dot_alloc[faer]@t1 0.9412 0.0588 0.0900 4 0.9979, 0.9921, 0.8309, 0.8902
hadamard_f64_m64n64_dot_alloc[faer]@t4 1.0026 0.0026 0.0048 4 1.0010, 1.0072, 1.0042, 0.9964
hadamard_f64_m64n64_einsum_alloc[faer]@t1 0.9547 0.0453 0.0901 4 0.9807, 0.9288, 0.9157, 0.9925
hadamard_f64_m64n64_einsum_alloc[faer]@t4 0.9937 0.0063 0.0237 4 0.9973, 1.0150, 0.9902, 0.9826
matmul_n16_count1024[blas]@t1 1.0026 0.0026 0.0875 4 0.9981, 1.0007, 1.1173, 1.0045
matmul_n16_count1024[blas]@t4 0.9976 0.0024 0.0832 4 0.9948, 1.0005, 0.9780, 1.0043
matmul_n16_count1024[faer]@t1 0.9705 0.0295 0.0907 4 0.9467, 0.9335, 0.9942, 1.0698
matmul_n16_count1024[faer]@t4 0.9917 0.0083 0.0135 4 0.9917, 0.9862, 0.9916, 1.0031
matmul_n16_count1024[pytorch]@t1 1.0070 0.0070 0.0885 4 1.0051, 1.0117, 1.0090, 0.9892
matmul_n16_count1024[pytorch]@t4 0.9324 0.0676 0.1198 4 0.8212, 1.0152, 0.8781, 0.9867
matmul_n2_count1024[blas]@t1 1.0010 0.0010 0.0916 4 0.9975, 1.1977, 0.9333, 1.0046
matmul_n2_count1024[blas]@t4 0.9984 0.0016 0.0102 4 1.0028, 0.9975, 0.9756, 0.9994
matmul_n2_count1024[faer]@t1 0.9998 0.0002 0.0643 4 0.9913, 1.0512, 0.9984, 1.0012
matmul_n2_count1024[faer]@t4 0.9049 0.0951 0.2829 4 1.0033, 0.6888, 1.0140, 0.8066
matmul_n2_count1024[pytorch]@t1 0.9951 0.0049 0.1422 4 0.9969, 1.0341, 0.9934, 0.9884
matmul_n2_count1024[pytorch]@t4 0.9835 0.0165 0.0853 4 0.9862, 0.9458, 0.9808, 1.0276
matmul_n32_count1024[blas]@t1 0.9567 0.0433 0.0758 4 0.8926, 0.9449, 0.9992, 0.9685
matmul_n32_count1024[blas]@t4 0.9901 0.0099 0.0646 4 0.9974, 1.0040, 0.9633, 0.9827
matmul_n32_count1024[faer]@t1 1.0116 0.0116 0.0795 4 0.9032, 1.0921, 1.0277, 0.9955
matmul_n32_count1024[faer]@t4 0.9643 0.0357 0.0660 4 0.9625, 0.9661, 0.9534, 1.1301
matmul_n32_count1024[pytorch]@t1 1.0179 0.0179 0.1445 4 1.0206, 0.9031, 1.0229, 1.0152
matmul_n32_count1024[pytorch]@t4 0.9806 0.0194 0.0629 4 1.0087, 0.9502, 0.9976, 0.9635
matmul_n4_count1024[blas]@t1 0.9225 0.0775 0.0941 4 0.8302, 0.8547, 1.0029, 0.9904
matmul_n4_count1024[blas]@t4 0.9869 0.0131 0.0106 4 0.9748, 0.9908, 0.9963, 0.9831
matmul_n4_count1024[faer]@t1 1.0046 0.0046 0.0811 4 0.9980, 1.0458, 0.9991, 1.0101
matmul_n4_count1024[faer]@t4 0.9976 0.0024 0.0112 4 0.9813, 0.9974, 1.0071, 0.9978
matmul_n4_count1024[pytorch]@t1 0.9731 0.0269 0.1187 4 0.9694, 0.8734, 0.9767, 1.0093
matmul_n4_count1024[pytorch]@t4 1.0017 0.0017 0.0638 4 1.0171, 0.9733, 1.0298, 0.9863
matmul_n8_count1024[blas]@t1 0.8423 0.1577 0.0887 4 0.8376, 0.8440, 0.8407, 1.0071
matmul_n8_count1024[blas]@t4 0.9949 0.0051 0.0188 4 0.9960, 0.9996, 0.9356, 0.9939
matmul_n8_count1024[faer]@t1 0.9624 0.0376 0.0909 4 0.9791, 0.9457, 0.9437, 1.2166
matmul_n8_count1024[faer]@t4 0.9998 0.0002 0.0088 4 0.9978, 1.0023, 0.9668, 1.0017
matmul_n8_count1024[pytorch]@t1 0.9989 0.0011 0.0905 4 1.0054, 0.9887, 1.0390, 0.9925
matmul_n8_count1024[pytorch]@t4 1.0165 0.0165 0.1351 4 0.8175, 1.0792, 1.0640, 0.9690
solve_n16_count1024[blas]@t1 0.9976 0.0024 0.0783 4 0.8343, 0.9929, 1.0042, 1.0024
solve_n16_count1024[blas]@t4 1.0066 0.0066 0.0653 4 0.9543, 1.0049, 1.0587, 1.0082
solve_n16_count1024[faer]@t1 0.8950 0.1050 0.0807 4 0.8559, 0.9342, 0.8391, 1.0045
solve_n16_count1024[faer]@t4 1.0044 0.0044 0.0448 4 0.9925, 1.0043, 1.0046, 1.0554
solve_n16_count1024[pytorch]@t1 0.9641 0.0359 0.0887 4 0.8311, 0.9341, 0.9942, 1.0191
solve_n16_count1024[pytorch]@t4 1.0026 0.0026 0.0776 4 1.0067, 1.0545, 0.9682, 0.9984
solve_n2_count1024[blas]@t1 0.9893 0.0107 0.0926 4 1.2004, 0.9772, 1.0014, 0.8379
solve_n2_count1024[blas]@t4 1.0552 0.0552 0.0576 4 1.0883, 1.0238, 1.0867, 1.0089
solve_n2_count1024[faer]@t1 0.9265 0.0735 0.0861 4 1.0508, 0.9338, 0.8418, 0.9193
solve_n2_count1024[faer]@t4 0.9906 0.0094 0.1762 4 0.9925, 0.9887, 0.9748, 0.9957
solve_n2_count1024[pytorch]@t1 0.9792 0.0208 0.0805 4 0.9829, 0.9755, 0.9897, 0.9541
solve_n2_count1024[pytorch]@t4 0.9896 0.0104 0.0752 4 0.8304, 0.9895, 1.0006, 0.9897
solve_n32_count1024[blas]@t1 0.9875 0.0125 0.0789 4 0.8236, 0.9993, 0.9779, 0.9971
solve_n32_count1024[blas]@t4 0.9978 0.0022 0.0527 4 0.9686, 1.0172, 0.9879, 1.0078
solve_n32_count1024[faer]@t1 0.9941 0.0059 0.0706 4 0.9489, 1.0845, 0.9880, 1.0001
solve_n32_count1024[faer]@t4 0.9986 0.0014 0.0788 4 0.9983, 1.0329, 0.9647, 0.9989
solve_n32_count1024[pytorch]@t1 0.9999 0.0001 0.0624 4 0.8401, 1.0504, 1.0092, 0.9906
solve_n32_count1024[pytorch]@t4 0.9947 0.0053 0.0485 4 0.9909, 1.0243, 0.9863, 0.9986
solve_n4_count1024[blas]@t1 1.1207 0.1207 0.0864 4 1.0526, 0.8916, 1.1888, 1.2096
solve_n4_count1024[blas]@t4 1.0179 0.0179 0.0647 4 1.0127, 1.0231, 0.9052, 1.1266
solve_n4_count1024[faer]@t1 0.9984 0.0016 0.0846 4 0.9954, 0.8317, 1.0014, 1.1948
solve_n4_count1024[faer]@t4 0.9970 0.0030 0.2273 4 1.0023, 0.9674, 1.0065, 0.9917
solve_n4_count1024[pytorch]@t1 0.9986 0.0014 0.0765 4 0.9806, 1.0045, 0.9973, 0.9999
solve_n4_count1024[pytorch]@t4 1.0039 0.0039 0.0569 4 0.8544, 1.0071, 1.0006, 1.0108
solve_n8_count1024[blas]@t1 1.0083 0.0083 0.0817 4 1.0058, 1.0122, 0.8299, 1.0108
solve_n8_count1024[blas]@t4 1.0439 0.0439 0.0572 4 1.0752, 1.1147, 1.0127, 0.9965
solve_n8_count1024[faer]@t1 0.9348 0.0652 0.0922 4 0.8731, 0.8475, 0.9964, 1.2036
solve_n8_count1024[faer]@t4 0.9845 0.0155 0.0184 4 0.9845, 1.0100, 0.9846, 0.9834
solve_n8_count1024[pytorch]@t1 0.9656 0.0344 0.0650 4 0.8483, 0.9634, 0.9679, 1.0018
solve_n8_count1024[pytorch]@t4 0.9804 0.0196 0.0702 4 0.8675, 0.9808, 0.9801, 0.9978
stream_f64_fixed_len32_einsum_alloc[faer]@t1 1.0164 0.0164 0.0201 4 1.0428, 1.0039, 1.0289, 1.0032
stream_f64_fixed_len32_einsum_alloc[faer]@t4 0.9875 0.0125 0.0950 4 0.9957, 1.0062, 0.8028, 0.9792
stream_f64_fresh_len32_einsum_alloc[faer]@t1 0.9910 0.0090 0.0260 4 0.9171, 0.9967, 1.0205, 0.9852
stream_f64_fresh_len32_einsum_alloc[faer]@t4 0.9951 0.0049 0.0873 4 1.0019, 0.9852, 0.9883, 1.0137
stream_f64_mixed_len32_einsum_alloc[faer]@t1 0.9987 0.0013 0.0214 4 1.0720, 0.9861, 0.9985, 0.9989
stream_f64_mixed_len32_einsum_alloc[faer]@t4 0.9924 0.0076 0.0421 4 1.0363, 0.9881, 0.9966, 0.9802
stream_f64_strides_len32_einsum_alloc[faer]@t1 1.0064 0.0064 0.0292 4 1.0446, 0.9864, 0.9985, 1.0143
stream_f64_strides_len32_einsum_alloc[faer]@t4 1.0223 0.0223 0.0896 4 1.0448, 1.0263, 1.0173, 1.0182