nnicolosi/Raptor
1
0
Fork 0
Code Issues Pull Requests Actions 9 Packages Projects Releases Wiki Activity

big refactor
Validate Operations / validate-operations (push) Has been cancelled
Details

Browse Source
This commit is contained in:
NiccoloN
2026-08-04 11:28:05 +02:00
parent f4a3b012cc
commit 10b6ee6c32
150 changed files with 6737 additions and 4816 deletions
Download Patch File Download Diff File Expand all files Collapse all files
validation/operations/validation_results.csv
+168 -165
View File
@@ -1,166 +1,169 @@
Operation,Result,Compile,Host mem,Cores mem,Cores,Xbars,Latency,Power,Energy
add/after_gemm,PASS,0.141 s,0.01 MiB,0.01 MiB,5,4,0.007784 ms,104.703618 mW,815012.960000 pJ
add/basic,PASS,0.114 s,0.00 MiB,0.00 MiB,1,0,0.000323 ms,78.222910 mW,25266.000000 pJ
add/broadcast_row,PASS,0.110 s,0.00 MiB,0.00 MiB,1,0,0.000323 ms,78.222910 mW,25266.000000 pJ
add/channel_broadcast_1024,PASS,0.105 s,0.02 MiB,0.01 MiB,1,0,0.006913 ms,78.118038 mW,540030.000000 pJ
add/leading_dimension_broadcast,PASS,0.137 s,0.00 MiB,0.00 MiB,1,0,0.000323 ms,78.222910 mW,25266.000000 pJ
concat/channel_axis,PASS,0.127 s,0.00 MiB,0.00 MiB,1,0,0.000457 ms,78.157549 mW,35718.000000 pJ
concat/negative_axis,PASS,0.102 s,0.00 MiB,0.00 MiB,1,0,0.001043 ms,78.092042 mW,81450.000000 pJ
concat/three_inputs_channel_axis,PASS,0.127 s,0.00 MiB,0.00 MiB,1,0,0.000644 ms,78.149068 mW,50328.000000 pJ
conv/batch_2,PASS,0.102 s,0.00 MiB,0.00 MiB,2,2,0.013694 ms,82.623885 mW,1131451.480000 pJ
conv/batch_4_pointwise,PASS,0.110 s,0.00 MiB,0.01 MiB,5,4,0.003932 ms,116.078576 mW,456420.960000 pJ
conv/depthwise_1024_channels,PASS,0.167 s,0.19 MiB,0.38 MiB,129,128,0.220751 ms,178.454307 mW,39393966.720000 pJ
conv/depthwise_grouped,PASS,0.109 s,0.01 MiB,0.00 MiB,5,4,0.006024 ms,108.326521 mW,652558.960000 pJ
conv/dilated_3x3,PASS,0.143 s,0.00 MiB,0.00 MiB,3,3,0.004045 ms,110.234541 mW,445898.720000 pJ
conv/dynamic,PASS,0.107 s,0.00 MiB,0.00 MiB,5,0,0.001835 ms,92.281199 mW,169336.000000 pJ
conv/explicit_padding,PASS,0.139 s,0.00 MiB,0.00 MiB,4,4,0.004327 ms,115.794768 mW,501043.960000 pJ
conv/grouped_many_groups,PASS,0.616 s,0.05 MiB,0.09 MiB,65,64,0.181845 ms,142.210104 mW,25860196.360000 pJ
conv/grouped_two_groups,PASS,0.145 s,0.00 MiB,0.00 MiB,3,2,0.005360 ms,101.459418 mW,543822.480000 pJ
conv/huge_pointwise_1024,PASS,0.749 s,0.01 MiB,0.01 MiB,1,64,0.028261 ms,133.488743 mW,3772525.360000 pJ
conv/huge_pointwise_1024_dynamic,PASS,0.097 s,8.04 MiB,12.61 MiB,168,0,2.627964 ms,169.518697 mW,445489032.000000 pJ
conv/kernel_3x3,PASS,0.155 s,0.00 MiB,0.00 MiB,3,3,0.003091 ms,115.862414 mW,358130.720000 pJ
conv/kernel_equals_input_spatial,PASS,0.156 s,0.00 MiB,0.00 MiB,1,2,0.008443 ms,83.863849 mW,708062.480000 pJ
conv/large_input_channels_1x1,PASS,0.160 s,0.01 MiB,0.01 MiB,1,8,0.017167 ms,89.416900 mW,1535019.920000 pJ
conv/large_output_channels_1x1,PASS,0.141 s,0.00 MiB,0.01 MiB,1,8,0.004964 ms,117.628106 mW,583905.920000 pJ
conv/large_spatial,PASS,0.141 s,0.00 MiB,0.01 MiB,6,6,0.004096 ms,129.015000 mW,528445.440000 pJ
conv/multi_channel,PASS,0.143 s,0.00 MiB,0.00 MiB,3,3,0.005148 ms,106.453520 mW,548022.720000 pJ
conv/non_square_kernel_1x3,PASS,0.127 s,0.00 MiB,0.00 MiB,5,5,0.004029 ms,123.600943 mW,497988.200000 pJ
conv/non_square_kernel_3x1,PASS,0.085 s,0.00 MiB,0.00 MiB,3,3,0.005526 ms,105.464843 mW,582798.720000 pJ
conv/non_uniform_stride,PASS,0.120 s,0.00 MiB,0.00 MiB,4,4,0.005808 ms,110.081433 mW,639352.960000 pJ
conv/pointwise_1x1,PASS,0.139 s,0.00 MiB,0.00 MiB,4,4,0.004539 ms,114.835858 mW,521239.960000 pJ
conv/pointwise_tiled_chain,PASS,0.943 s,0.01 MiB,0.02 MiB,2,80,0.084437 ms,102.289307 mW,8637002.200000 pJ
conv/real_asymmetric_padding,PASS,0.115 s,0.00 MiB,0.00 MiB,4,4,0.005232 ms,111.870214 mW,585304.960000 pJ
conv/relu_conv_store,PASS,0.107 s,0.05 MiB,0.08 MiB,32,32,0.062978 ms,246.649827 mW,15533512.800000 pJ
conv/same_lower_3x3,PASS,0.094 s,0.00 MiB,0.00 MiB,5,5,0.004700 ms,119.232170 mW,560391.200000 pJ
conv/same_padding_3x3,PASS,0.126 s,0.00 MiB,0.00 MiB,5,5,0.004700 ms,119.232170 mW,560391.200000 pJ
conv/simple,PASS,0.125 s,0.00 MiB,0.00 MiB,2,2,0.003148 ms,94.665972 mW,298008.480000 pJ
conv/stride_2,PASS,0.146 s,0.00 MiB,0.00 MiB,2,2,0.002827 ms,96.393873 mW,272505.480000 pJ
conv/with_bias_3x3,PASS,0.121 s,0.00 MiB,0.00 MiB,3,3,0.004898 ms,107.176546 mW,524950.720000 pJ
conv/with_constant,PASS,0.106 s,0.00 MiB,0.00 MiB,3,3,0.004273 ms,109.362677 mW,467306.720000 pJ
conv/without_kernel_shape_attr,PASS,0.105 s,0.00 MiB,0.00 MiB,3,3,0.003091 ms,115.862414 mW,358130.720000 pJ
conv/yolo11n_stem,PASS,3.091 s,29.20 MiB,31.38 MiB,168,488,9.799213 ms,361.075203 mW,3538252825.000010 pJ
div/after_gemm,PASS,0.130 s,0.01 MiB,0.01 MiB,5,4,0.007784 ms,104.703618 mW,815012.960000 pJ
div/basic,PASS,0.116 s,0.00 MiB,0.00 MiB,1,0,0.000323 ms,78.222910 mW,25266.000000 pJ
div/channel_broadcast_1024,PASS,0.121 s,0.02 MiB,0.01 MiB,1,0,0.006913 ms,78.118038 mW,540030.000000 pJ
div/leading_dimension_broadcast,PASS,0.095 s,0.00 MiB,0.00 MiB,1,0,0.000323 ms,78.222910 mW,25266.000000 pJ
div/runtime_scalar_rhs,PASS,0.126 s,0.02 MiB,0.01 MiB,1,0,0.006913 ms,78.118038 mW,540030.000000 pJ
div/scalar_constant,PASS,0.069 s,0.00 MiB,0.00 MiB,1,0,0.000323 ms,78.222910 mW,25266.000000 pJ
gather/3d_input_axis1,PASS,0.121 s,0.00 MiB,0.00 MiB,1,0,0.000589 ms,78.081494 mW,45990.000000 pJ
gather/axis0_matrix_indices,PASS,0.088 s,0.00 MiB,0.00 MiB,1,0,0.000697 ms,78.068867 mW,54414.000000 pJ
gather/axis1,PASS,0.123 s,0.00 MiB,0.00 MiB,1,0,0.000801 ms,78.059925 mW,62526.000000 pJ
gather/negative_axis,PASS,0.087 s,0.00 MiB,0.00 MiB,1,0,0.001437 ms,78.033403 mW,112134.000000 pJ
gather/negative_indices,PASS,0.115 s,0.00 MiB,0.00 MiB,1,0,0.000376 ms,78.127660 mW,29376.000000 pJ
gemm/alpha_beta,PASS,0.121 s,0.01 MiB,0.01 MiB,5,4,0.007456 ms,105.272125 mW,784908.960000 pJ
gemm/bias_rank2_broadcast,PASS,0.127 s,0.00 MiB,0.01 MiB,5,4,0.007072 ms,105.979208 mW,749484.960000 pJ
gemm/dynamic,PASS,0.084 s,0.00 MiB,0.00 MiB,5,0,0.002421 ms,91.480793 mW,221475.000000 pJ
gemm/dynamic_alpha,PASS,0.104 s,0.00 MiB,0.00 MiB,5,0,0.003262 ms,91.415696 mW,298198.000000 pJ
gemm/dynamic_beta,PASS,0.110 s,0.00 MiB,0.00 MiB,5,0,0.004365 ms,91.316151 mW,398595.000000 pJ
gemm/dynamic_bias,PASS,0.092 s,0.00 MiB,0.00 MiB,5,0,0.002665 ms,91.445779 mW,243703.000000 pJ
gemm/dynamic_bias_alpha_beta,PASS,0.089 s,0.00 MiB,0.00 MiB,5,0,0.005629 ms,91.279268 mW,513811.000000 pJ
gemm/dynamic_transB,PASS,0.077 s,0.00 MiB,0.00 MiB,5,0,0.001301 ms,91.378171 mW,118883.000000 pJ
gemm/huge_1024,PASS,0.219 s,0.01 MiB,0.10 MiB,73,64,0.017522 ms,215.037402 mW,3767885.360000 pJ
gemm/large,PASS,0.097 s,0.02 MiB,0.03 MiB,17,16,0.011229 ms,140.152181 mW,1573768.840000 pJ
gemm/large_k_small_n,PASS,0.106 s,0.01 MiB,0.01 MiB,9,8,0.004748 ms,133.481449 mW,633769.920000 pJ
gemm/non_square,PASS,0.102 s,0.00 MiB,0.01 MiB,5,4,0.003527 ms,118.958310 mW,419565.960000 pJ
gemm/scalar_bias,PASS,0.111 s,0.00 MiB,0.01 MiB,5,4,0.007072 ms,105.979208 mW,749484.960000 pJ
gemm/simple,PASS,0.135 s,0.03 MiB,0.08 MiB,42,40,0.021640 ms,151.774196 mW,3284393.600000 pJ
gemm/small,PASS,0.074 s,0.00 MiB,0.00 MiB,2,2,0.004420 ms,90.144000 mW,398436.480000 pJ
gemm/small_k_large_n,PASS,0.129 s,0.01 MiB,0.02 MiB,17,8,0.007962 ms,131.005014 mW,1043061.920000 pJ
gemm/transA,PASS,0.089 s,0.00 MiB,0.01 MiB,5,4,0.005762 ms,109.140743 mW,628868.960000 pJ
gemm/transA_transB,PASS,0.115 s,0.00 MiB,0.01 MiB,5,4,0.005762 ms,109.140743 mW,628868.960000 pJ
gemm/transB,PASS,0.068 s,0.00 MiB,0.01 MiB,5,4,0.003527 ms,118.958310 mW,419565.960000 pJ
gemm/transB_with_bias,PASS,0.069 s,0.01 MiB,0.01 MiB,5,4,0.005046 ms,110.546762 mW,557818.960000 pJ
gemm/with_bias,PASS,0.113 s,0.01 MiB,0.01 MiB,5,4,0.005562 ms,108.767882 mW,604966.960000 pJ
gemv/constant,PASS,0.112 s,0.00 MiB,0.00 MiB,0,0,0.000000 ms,2.000000 mW,0.000000 pJ
gemv/simple,PASS,0.136 s,0.00 MiB,0.01 MiB,6,4,0.005160 ms,111.150380 mW,573535.960000 pJ
gemv/with_heterogeneous_constant,PASS,0.138 s,0.00 MiB,0.01 MiB,6,4,0.005549 ms,109.816536 mW,609371.960000 pJ
gemv/with_homogeneous_constant,PASS,0.140 s,0.00 MiB,0.01 MiB,6,4,0.005549 ms,109.816536 mW,609371.960000 pJ
gemv/with_scalar_constant,PASS,0.124 s,0.00 MiB,0.01 MiB,6,4,0.005549 ms,109.816536 mW,609371.960000 pJ
matmul/basic,PASS,0.081 s,0.00 MiB,0.00 MiB,2,2,0.004420 ms,90.144000 mW,398436.480000 pJ
matmul/batched_3d,PASS,0.133 s,0.00 MiB,0.01 MiB,5,4,0.005958 ms,108.588949 mW,646972.960000 pJ
matmul/batched_3d_dynamic,PASS,0.105 s,0.00 MiB,0.00 MiB,9,0,0.003525 ms,92.330213 mW,325464.000000 pJ
matmul/batched_left_constant,PASS,0.136 s,0.00 MiB,0.02 MiB,9,8,0.008822 ms,114.385164 mW,1009105.920000 pJ
matmul/batched_lhs_broadcast,PASS,0.133 s,0.00 MiB,0.01 MiB,5,4,0.005681 ms,109.389361 mW,621440.960000 pJ
matmul/batched_rhs_broadcast,PASS,0.134 s,0.00 MiB,0.01 MiB,5,4,0.005958 ms,108.588949 mW,646972.960000 pJ
matmul/dynamic,PASS,0.093 s,0.00 MiB,0.00 MiB,5,0,0.001621 ms,91.421962 mW,148195.000000 pJ
matmul/huge_1024,PASS,0.277 s,0.01 MiB,0.10 MiB,73,64,0.017522 ms,215.037402 mW,3767885.360000 pJ
matmul/left_constant,PASS,0.123 s,0.00 MiB,0.01 MiB,5,4,0.005853 ms,108.861944 mW,637168.960000 pJ
matmul/matrix_vector,PASS,0.150 s,0.52 MiB,0.78 MiB,168,173,0.384660 ms,202.131271 mW,77751814.880000 pJ
matmul/vector_matrix,PASS,0.186 s,0.01 MiB,0.01 MiB,9,8,0.007409 ms,118.680243 mW,879301.920000 pJ
mul/after_conv,PASS,0.127 s,0.00 MiB,0.00 MiB,4,3,0.005453 ms,107.639046 mW,586955.720000 pJ
mul/basic,PASS,0.070 s,0.00 MiB,0.00 MiB,1,0,0.000323 ms,78.222910 mW,25266.000000 pJ
mul/channel_broadcast_1024,PASS,0.065 s,0.02 MiB,0.01 MiB,1,0,0.006913 ms,78.118038 mW,540030.000000 pJ
mul/leading_dimension_broadcast,PASS,0.108 s,0.00 MiB,0.00 MiB,1,0,0.000323 ms,78.222910 mW,25266.000000 pJ
mul/scalar_constant,PASS,0.082 s,0.00 MiB,0.00 MiB,1,0,0.000323 ms,78.222910 mW,25266.000000 pJ
pool/avg_basic,PASS,0.070 s,0.00 MiB,0.00 MiB,1,0,0.011939 ms,78.022112 mW,931506.000000 pJ
pool/avg_ceil_mode,PASS,0.107 s,0.00 MiB,0.00 MiB,1,0,0.004359 ms,78.033035 mW,340146.000000 pJ
pool/avg_explicit_padding,PASS,0.111 s,0.00 MiB,0.00 MiB,1,0,0.008822 ms,78.027205 mW,688356.000000 pJ
pool/avg_include_pad,PASS,0.075 s,0.00 MiB,0.00 MiB,1,0,0.008506 ms,78.016929 mW,663612.000000 pJ
pool/avg_large_channels,PASS,0.064 s,0.04 MiB,0.02 MiB,1,0,0.178249 ms,78.280327 mW,13953390.000000 pJ
pool/avg_non_uniform_stride,PASS,0.068 s,0.00 MiB,0.00 MiB,1,0,0.014513 ms,78.016537 mW,1132254.000000 pJ
pool/avg_real_asymmetric_padding,PASS,0.064 s,0.00 MiB,0.00 MiB,1,0,0.025206 ms,78.024756 mW,1966692.000000 pJ
pool/max_after_conv,PASS,0.070 s,0.00 MiB,0.00 MiB,6,4,0.006452 ms,96.374606 mW,621808.960000 pJ
pool/max_basic,PASS,0.063 s,0.00 MiB,0.00 MiB,3,0,0.001634 ms,92.132191 mW,150544.000000 pJ
pool/max_ceil_mode,PASS,0.065 s,0.00 MiB,0.00 MiB,2,0,0.001297 ms,79.111025 mW,102607.000000 pJ
pool/max_global_style_kernel_equals_input,PASS,0.089 s,0.00 MiB,0.00 MiB,1,0,0.004366 ms,78.010994 mW,340596.000000 pJ
pool/max_non_square_kernel,PASS,0.103 s,0.00 MiB,0.00 MiB,4,0,0.003409 ms,93.253447 mW,317901.000000 pJ
pool/max_real_asymmetric_padding,PASS,0.088 s,0.00 MiB,0.00 MiB,4,0,0.003078 ms,93.124756 mW,286638.000000 pJ
pool/max_same_upper,PASS,0.065 s,0.00 MiB,0.00 MiB,3,0,0.003024 ms,92.095238 mW,278496.000000 pJ
pool/max_stride2_multichannel,PASS,0.086 s,0.00 MiB,0.00 MiB,3,0,0.004012 ms,92.269192 mW,370184.000000 pJ
reduce_mean/4d_spatial,PASS,0.061 s,0.00 MiB,0.00 MiB,3,0,0.000321 ms,92.448598 mW,29676.000000 pJ
reduce_mean/4d_spatial_keepdims_0,PASS,0.067 s,0.00 MiB,0.00 MiB,4,0,0.000655 ms,94.352672 mW,61801.000000 pJ
reduce_mean/after_conv,PASS,0.115 s,0.00 MiB,0.00 MiB,5,3,0.005342 ms,106.951089 mW,571332.720000 pJ
reduce_mean/all_axes_keepdims_0,PASS,0.103 s,0.00 MiB,0.00 MiB,2,0,0.000391 ms,79.237852 mW,30982.000000 pJ
reduce_mean/all_axes_keepdims_1,PASS,0.086 s,0.00 MiB,0.00 MiB,1,0,0.000221 ms,78.217195 mW,17286.000000 pJ
reduce_mean/basic,PASS,0.099 s,0.00 MiB,0.00 MiB,4,0,0.000373 ms,93.514745 mW,34881.000000 pJ
reduce_mean/channel_axis_nchw,PASS,0.078 s,0.03 MiB,0.02 MiB,4,0,0.164926 ms,93.596631 mW,15436518.000000 pJ
reduce_mean/keepdims_0,PASS,0.073 s,0.00 MiB,0.00 MiB,5,0,0.000748 ms,91.401070 mW,68368.000000 pJ
reduce_mean/large_dimension_1024,PASS,0.115 s,0.01 MiB,0.00 MiB,1,0,0.002785 ms,78.017235 mW,217278.000000 pJ
reduce_mean/legacy_axes_1_2_keepdims_1,PASS,0.074 s,0.00 MiB,0.00 MiB,2,0,0.000271 ms,79.354244 mW,21505.000000 pJ
reduce_mean/legacy_axis1_keepdims_0,PASS,0.111 s,0.00 MiB,0.00 MiB,9,0,0.001986 ms,92.501511 mW,183708.000000 pJ
reduce_mean/legacy_axis1_keepdims_1,PASS,0.086 s,0.00 MiB,0.00 MiB,8,0,0.001373 ms,94.559359 mW,129830.000000 pJ
reduce_mean/legacy_empty_axes_noop,PASS,0.115 s,0.00 MiB,0.00 MiB,1,0,0.000221 ms,78.217195 mW,17286.000000 pJ
reduce_mean/legacy_nchw_spatial,PASS,0.090 s,0.00 MiB,0.00 MiB,3,0,0.000321 ms,92.448598 mW,29676.000000 pJ
reduce_mean/legacy_negative_axis,PASS,0.110 s,0.00 MiB,0.00 MiB,6,0,0.000553 ms,93.520796 mW,51717.000000 pJ
reduce_mean/legacy_reduce_all_keepdims_1,PASS,0.107 s,0.00 MiB,0.00 MiB,1,0,0.000221 ms,78.217195 mW,17286.000000 pJ
reduce_mean/negative_axis,PASS,0.120 s,0.00 MiB,0.00 MiB,6,0,0.000553 ms,93.520796 mW,51717.000000 pJ
relu/4d,PASS,0.116 s,0.00 MiB,0.00 MiB,1,0,0.000521 ms,78.184261 mW,40734.000000 pJ
relu/after_conv,PASS,0.078 s,0.00 MiB,0.00 MiB,3,3,0.004956 ms,106.998935 mW,530286.720000 pJ
relu/after_gemm,PASS,0.077 s,0.01 MiB,0.01 MiB,5,4,0.007513 ms,105.158653 mW,790056.960000 pJ
relu/basic,PASS,0.066 s,0.00 MiB,0.00 MiB,1,0,0.000221 ms,78.217195 mW,17286.000000 pJ
reshape/4d_to_2d_flatten,PASS,0.098 s,0.00 MiB,0.00 MiB,1,0,0.000258 ms,78.279070 mW,20196.000000 pJ
reshape/infer_dim_minus_one,PASS,0.105 s,0.00 MiB,0.00 MiB,1,0,0.000162 ms,78.296296 mW,12684.000000 pJ
reshape/same_rank,PASS,0.057 s,0.00 MiB,0.00 MiB,1,0,0.000162 ms,78.296296 mW,12684.000000 pJ
reshape/zero_copies_input_dim,PASS,0.090 s,0.00 MiB,0.00 MiB,1,0,0.000162 ms,78.296296 mW,12684.000000 pJ
resize/height_only,PASS,0.104 s,0.00 MiB,0.00 MiB,1,0,0.000795 ms,78.060377 mW,62058.000000 pJ
resize/nearest_2x,PASS,0.104 s,0.00 MiB,0.00 MiB,1,0,0.001422 ms,78.033755 mW,110964.000000 pJ
resize/nearest_downsample,PASS,0.059 s,0.00 MiB,0.00 MiB,1,0,0.000481 ms,78.099792 mW,37566.000000 pJ
resize/non_uniform,PASS,0.063 s,0.00 MiB,0.00 MiB,1,0,0.002108 ms,78.034156 mW,164496.000000 pJ
resize/width_only,PASS,0.068 s,0.00 MiB,0.00 MiB,1,0,0.000792 ms,78.060606 mW,61824.000000 pJ
resize/with_sizes,PASS,0.056 s,0.00 MiB,0.00 MiB,1,0,0.000951 ms,78.050473 mW,74226.000000 pJ
sigmoid/4d,PASS,0.104 s,0.00 MiB,0.00 MiB,1,0,0.000521 ms,78.184261 mW,40734.000000 pJ
sigmoid/after_gemm,PASS,0.118 s,0.01 MiB,0.01 MiB,5,4,0.007513 ms,105.158653 mW,790056.960000 pJ
sigmoid/basic,PASS,0.083 s,0.00 MiB,0.00 MiB,1,0,0.000221 ms,78.217195 mW,17286.000000 pJ
slice/2d_basic,PASS,0.067 s,0.00 MiB,0.00 MiB,1,0,0.000242 ms,78.297521 mW,18948.000000 pJ
slice/after_conv,PASS,0.111 s,0.00 MiB,0.01 MiB,7,6,0.011296 ms,118.190765 mW,1335082.880000 pJ
slice/default_axes,PASS,0.065 s,0.00 MiB,0.00 MiB,1,0,0.000242 ms,78.297521 mW,18948.000000 pJ
slice/large_channel_1024,PASS,0.060 s,0.01 MiB,0.00 MiB,1,0,0.002832 ms,78.144068 mW,221304.000000 pJ
slice/nchw_spatial_crop,PASS,0.067 s,0.00 MiB,0.00 MiB,1,0,0.001302 ms,78.239631 mW,101868.000000 pJ
slice/negative_axis,PASS,0.098 s,0.00 MiB,0.00 MiB,1,0,0.000562 ms,78.298932 mW,44004.000000 pJ
slice/negative_indices,PASS,0.106 s,0.00 MiB,0.00 MiB,1,0,0.000322 ms,78.298137 mW,25212.000000 pJ
slice/step2,PASS,0.068 s,0.00 MiB,0.00 MiB,1,0,0.002042 ms,78.293830 mW,159876.000000 pJ
softmax/3d_last_axis,PASS,0.072 s,0.00 MiB,0.00 MiB,1,0,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED
softmax/basic,PASS,0.088 s,0.00 MiB,0.00 MiB,1,0,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED
softmax/channel_axis,PASS,0.110 s,0.00 MiB,0.00 MiB,1,0,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED
softmax/large_dimension_1024,PASS,0.104 s,0.01 MiB,0.01 MiB,1,0,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED
softmax/negative_axis,PASS,0.118 s,0.00 MiB,0.00 MiB,1,0,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED
split/basic,PASS,0.077 s,0.00 MiB,0.00 MiB,1,0,0.000403 ms,78.297767 mW,31554.000000 pJ
split/equal_three_way,PASS,0.107 s,0.00 MiB,0.00 MiB,1,0,0.000564 ms,78.297872 mW,44160.000000 pJ
split/negative_axis,PASS,0.111 s,0.00 MiB,0.00 MiB,1,0,0.001083 ms,78.288089 mW,84786.000000 pJ
split/uneven_channel_axis_4d,PASS,0.061 s,0.00 MiB,0.00 MiB,1,0,0.000242 ms,78.297521 mW,18948.000000 pJ
sub/after_gemm,PASS,0.068 s,0.01 MiB,0.01 MiB,5,4,0.007784 ms,104.703618 mW,815012.960000 pJ
sub/basic,PASS,0.063 s,0.00 MiB,0.00 MiB,1,0,0.000323 ms,78.222910 mW,25266.000000 pJ
sub/broadcast_row,PASS,0.075 s,0.00 MiB,0.00 MiB,1,0,0.000323 ms,78.222910 mW,25266.000000 pJ
sub/channel_broadcast_1024,PASS,0.052 s,0.02 MiB,0.01 MiB,1,0,0.006913 ms,78.118038 mW,540030.000000 pJ
sub/constant_lhs_broadcast,PASS,0.057 s,0.00 MiB,0.00 MiB,1,0,0.000322 ms,78.223602 mW,25188.000000 pJ
sub/leading_dimension_broadcast,PASS,0.053 s,0.00 MiB,0.00 MiB,1,0,0.000323 ms,78.222910 mW,25266.000000 pJ
add/after_gemm,PASS,0.063 s,0.01 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
add/basic,PASS,0.057 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
add/broadcast_row,PASS,0.057 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
add/channel_broadcast_1024,PASS,0.061 s,0.02 MiB,0.01 MiB,1,0,SKIP,SKIP,SKIP
add/leading_dimension_broadcast,PASS,0.057 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
concat/channel_axis,PASS,0.069 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
concat/negative_axis,PASS,0.061 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
concat/three_inputs_channel_axis,PASS,0.058 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
conv/batch_2,PASS,0.066 s,0.00 MiB,0.00 MiB,2,2,SKIP,SKIP,SKIP
conv/batch_4_pointwise,PASS,0.065 s,0.00 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
conv/depthwise_1024_channels,PASS,0.083 s,0.19 MiB,0.38 MiB,129,128,SKIP,SKIP,SKIP
conv/depthwise_grouped,PASS,0.071 s,0.01 MiB,0.00 MiB,5,4,SKIP,SKIP,SKIP
conv/dilated_3x3,PASS,0.068 s,0.01 MiB,0.01 MiB,10,9,SKIP,SKIP,SKIP
conv/dynamic,PASS,0.065 s,0.00 MiB,0.00 MiB,5,0,SKIP,SKIP,SKIP
conv/explicit_padding,PASS,0.062 s,0.01 MiB,0.02 MiB,17,16,SKIP,SKIP,SKIP
conv/grouped_many_groups,PASS,0.442 s,0.05 MiB,0.09 MiB,65,64,SKIP,SKIP,SKIP
conv/grouped_two_groups,PASS,0.060 s,0.00 MiB,0.00 MiB,3,2,SKIP,SKIP,SKIP
conv/huge_pointwise_1024,PASS,0.156 s,0.01 MiB,0.11 MiB,73,64,SKIP,SKIP,SKIP
conv/huge_pointwise_1024_dynamic,PASS,0.079 s,8.04 MiB,12.61 MiB,168,0,SKIP,SKIP,SKIP
conv/kernel_3x3,PASS,0.062 s,0.01 MiB,0.01 MiB,10,9,SKIP,SKIP,SKIP
conv/kernel_equals_input_spatial,PASS,0.064 s,0.00 MiB,0.00 MiB,2,2,SKIP,SKIP,SKIP
conv/large_input_channels_1x1,PASS,0.089 s,0.01 MiB,0.02 MiB,9,8,SKIP,SKIP,SKIP
conv/large_output_channels_1x1,PASS,0.098 s,0.01 MiB,0.02 MiB,17,8,SKIP,SKIP,SKIP
conv/large_spatial,PASS,0.069 s,0.01 MiB,0.04 MiB,37,36,SKIP,SKIP,SKIP
conv/multi_channel,PASS,0.066 s,0.00 MiB,0.00 MiB,4,3,SKIP,SKIP,SKIP
conv/non_square_kernel_1x3,PASS,0.061 s,0.00 MiB,0.00 MiB,3,2,SKIP,SKIP,SKIP
conv/non_square_kernel_3x1,PASS,0.064 s,0.00 MiB,0.00 MiB,3,2,SKIP,SKIP,SKIP
conv/non_uniform_stride,PASS,0.062 s,0.00 MiB,0.00 MiB,4,3,SKIP,SKIP,SKIP
conv/pointwise_1x1,PASS,0.059 s,0.00 MiB,0.00 MiB,1,1,SKIP,SKIP,SKIP
conv/pointwise_tiled_chain,PASS,0.604 s,0.01 MiB,0.04 MiB,20,80,SKIP,SKIP,SKIP
conv/real_asymmetric_padding,PASS,0.060 s,0.01 MiB,0.03 MiB,29,28,SKIP,SKIP,SKIP
conv/relu_conv_store,PASS,0.091 s,0.16 MiB,0.67 MiB,168,184,SKIP,SKIP,SKIP
conv/same_lower_3x3,PASS,0.078 s,0.01 MiB,0.02 MiB,26,25,SKIP,SKIP,SKIP
conv/same_padding_3x3,PASS,0.070 s,0.01 MiB,0.02 MiB,26,25,SKIP,SKIP,SKIP
conv/simple,PASS,0.064 s,0.00 MiB,0.00 MiB,1,1,SKIP,SKIP,SKIP
conv/stride_2,PASS,0.064 s,0.01 MiB,0.00 MiB,5,4,SKIP,SKIP,SKIP
conv/with_bias_3x3,PASS,0.067 s,0.00 MiB,0.01 MiB,4,3,SKIP,SKIP,SKIP
conv/with_constant,PASS,0.070 s,0.00 MiB,0.00 MiB,1,1,SKIP,SKIP,SKIP
conv/without_kernel_shape_attr,PASS,0.069 s,0.01 MiB,0.01 MiB,10,9,SKIP,SKIP,SKIP
conv/yolo11n_depthwise_head,PASS,1.482 s,8.66 MiB,34.24 MiB,168,255,SKIP,SKIP,SKIP
conv/yolo11n_heavy,PASS,0.431 s,4.82 MiB,19.10 MiB,161,800,SKIP,SKIP,SKIP
conv/yolo11n_stem,PASS,0.783 s,12.86 MiB,37.59 MiB,168,488,SKIP,SKIP,SKIP
div/after_gemm,PASS,0.072 s,0.01 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
div/basic,PASS,0.061 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
div/channel_broadcast_1024,PASS,0.066 s,0.02 MiB,0.01 MiB,1,0,SKIP,SKIP,SKIP
div/leading_dimension_broadcast,PASS,0.059 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
div/runtime_scalar_rhs,PASS,0.056 s,0.02 MiB,0.01 MiB,1,0,SKIP,SKIP,SKIP
div/scalar_constant,PASS,0.073 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
gather/3d_input_axis1,PASS,0.060 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
gather/axis0_matrix_indices,PASS,0.056 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
gather/axis1,PASS,0.058 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
gather/negative_axis,PASS,0.058 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
gather/negative_indices,PASS,0.063 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
gemm/alpha_beta,PASS,0.067 s,0.01 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
gemm/bias_rank2_broadcast,PASS,0.060 s,0.00 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
gemm/dynamic,PASS,0.064 s,0.00 MiB,0.00 MiB,5,0,SKIP,SKIP,SKIP
gemm/dynamic_alpha,PASS,0.062 s,0.00 MiB,0.00 MiB,5,0,SKIP,SKIP,SKIP
gemm/dynamic_beta,PASS,0.060 s,0.00 MiB,0.00 MiB,5,0,SKIP,SKIP,SKIP
gemm/dynamic_bias,PASS,0.058 s,0.00 MiB,0.00 MiB,5,0,SKIP,SKIP,SKIP
gemm/dynamic_bias_alpha_beta,PASS,0.067 s,0.00 MiB,0.00 MiB,5,0,SKIP,SKIP,SKIP
gemm/dynamic_transB,PASS,0.062 s,0.00 MiB,0.00 MiB,5,0,SKIP,SKIP,SKIP
gemm/huge_1024,PASS,0.147 s,0.01 MiB,0.10 MiB,73,64,SKIP,SKIP,SKIP
gemm/large,PASS,0.068 s,0.02 MiB,0.03 MiB,17,16,SKIP,SKIP,SKIP
gemm/large_k_small_n,PASS,0.095 s,0.01 MiB,0.01 MiB,9,8,SKIP,SKIP,SKIP
gemm/non_square,PASS,0.063 s,0.00 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
gemm/scalar_bias,PASS,0.060 s,0.00 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
gemm/simple,PASS,0.072 s,0.03 MiB,0.08 MiB,42,40,SKIP,SKIP,SKIP
gemm/small,PASS,0.065 s,0.00 MiB,0.00 MiB,2,2,SKIP,SKIP,SKIP
gemm/small_k_large_n,PASS,0.097 s,0.01 MiB,0.02 MiB,17,8,SKIP,SKIP,SKIP
gemm/transA,PASS,0.064 s,0.00 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
gemm/transA_transB,PASS,0.069 s,0.00 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
gemm/transB,PASS,0.062 s,0.00 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
gemm/transB_with_bias,PASS,0.055 s,0.01 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
gemm/with_bias,PASS,0.067 s,0.01 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
gemv/constant,PASS,0.064 s,0.00 MiB,0.00 MiB,0,0,SKIP,SKIP,SKIP
gemv/simple,PASS,0.069 s,0.00 MiB,0.01 MiB,6,4,SKIP,SKIP,SKIP
gemv/with_heterogeneous_constant,PASS,0.066 s,0.00 MiB,0.01 MiB,6,4,SKIP,SKIP,SKIP
gemv/with_homogeneous_constant,PASS,0.070 s,0.00 MiB,0.01 MiB,6,4,SKIP,SKIP,SKIP
gemv/with_scalar_constant,PASS,0.070 s,0.00 MiB,0.01 MiB,6,4,SKIP,SKIP,SKIP
matmul/basic,PASS,0.062 s,0.00 MiB,0.00 MiB,2,2,SKIP,SKIP,SKIP
matmul/batched_3d,PASS,0.066 s,0.00 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
matmul/batched_3d_dynamic,PASS,0.057 s,0.00 MiB,0.00 MiB,4,0,SKIP,SKIP,SKIP
matmul/batched_left_constant,PASS,0.070 s,0.00 MiB,0.02 MiB,9,8,SKIP,SKIP,SKIP
matmul/batched_lhs_broadcast,PASS,0.063 s,0.00 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
matmul/batched_rhs_broadcast,PASS,0.062 s,0.00 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
matmul/dynamic,PASS,0.060 s,0.00 MiB,0.00 MiB,5,0,SKIP,SKIP,SKIP
matmul/huge_1024,PASS,0.145 s,0.01 MiB,0.10 MiB,73,64,SKIP,SKIP,SKIP
matmul/left_constant,PASS,0.051 s,0.00 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
matmul/matrix_vector,PASS,0.095 s,0.52 MiB,0.78 MiB,168,173,SKIP,SKIP,SKIP
matmul/vector_matrix,PASS,0.087 s,0.01 MiB,0.01 MiB,9,8,SKIP,SKIP,SKIP
matmul/yolo_attention,PASS,0.385 s,1.02 MiB,43.44 MiB,168,0,SKIP,SKIP,SKIP
mul/after_conv,PASS,0.055 s,0.00 MiB,0.00 MiB,4,3,SKIP,SKIP,SKIP
mul/basic,PASS,0.055 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
mul/channel_broadcast_1024,PASS,0.058 s,0.02 MiB,0.01 MiB,1,0,SKIP,SKIP,SKIP
mul/leading_dimension_broadcast,PASS,0.059 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
mul/scalar_constant,PASS,0.056 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
pool/avg_basic,PASS,0.063 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
pool/avg_ceil_mode,PASS,0.063 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
pool/avg_explicit_padding,PASS,0.061 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
pool/avg_include_pad,PASS,0.049 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
pool/avg_large_channels,PASS,0.059 s,0.04 MiB,0.02 MiB,1,0,SKIP,SKIP,SKIP
pool/avg_non_uniform_stride,PASS,0.066 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
pool/avg_real_asymmetric_padding,PASS,0.055 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
pool/max_after_conv,PASS,0.069 s,0.00 MiB,0.00 MiB,5,4,SKIP,SKIP,SKIP
pool/max_basic,PASS,0.064 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
pool/max_ceil_mode,PASS,0.058 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
pool/max_global_style_kernel_equals_input,PASS,0.063 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
pool/max_non_square_kernel,PASS,0.067 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
pool/max_real_asymmetric_padding,PASS,0.063 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
pool/max_same_upper,PASS,0.061 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
pool/max_stride2_multichannel,PASS,0.076 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
reduce_mean/4d_spatial,PASS,0.058 s,0.00 MiB,0.00 MiB,3,0,SKIP,SKIP,SKIP
reduce_mean/4d_spatial_keepdims_0,PASS,0.068 s,0.00 MiB,0.00 MiB,4,0,SKIP,SKIP,SKIP
reduce_mean/after_conv,PASS,0.067 s,0.00 MiB,0.00 MiB,5,3,SKIP,SKIP,SKIP
reduce_mean/all_axes_keepdims_0,PASS,0.057 s,0.00 MiB,0.00 MiB,2,0,SKIP,SKIP,SKIP
reduce_mean/all_axes_keepdims_1,PASS,0.053 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
reduce_mean/basic,PASS,0.058 s,0.00 MiB,0.00 MiB,4,0,SKIP,SKIP,SKIP
reduce_mean/channel_axis_nchw,PASS,0.063 s,0.03 MiB,0.02 MiB,4,0,SKIP,SKIP,SKIP
reduce_mean/keepdims_0,PASS,0.064 s,0.00 MiB,0.00 MiB,5,0,SKIP,SKIP,SKIP
reduce_mean/large_dimension_1024,PASS,0.066 s,0.01 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
reduce_mean/legacy_axes_1_2_keepdims_1,PASS,0.069 s,0.00 MiB,0.00 MiB,2,0,SKIP,SKIP,SKIP
reduce_mean/legacy_axis1_keepdims_0,PASS,0.067 s,0.00 MiB,0.00 MiB,9,0,SKIP,SKIP,SKIP
reduce_mean/legacy_axis1_keepdims_1,PASS,0.057 s,0.00 MiB,0.00 MiB,8,0,SKIP,SKIP,SKIP
reduce_mean/legacy_empty_axes_noop,PASS,0.058 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
reduce_mean/legacy_nchw_spatial,PASS,0.059 s,0.00 MiB,0.00 MiB,3,0,SKIP,SKIP,SKIP
reduce_mean/legacy_negative_axis,PASS,0.052 s,0.00 MiB,0.00 MiB,6,0,SKIP,SKIP,SKIP
reduce_mean/legacy_reduce_all_keepdims_1,PASS,0.057 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
reduce_mean/negative_axis,PASS,0.055 s,0.00 MiB,0.00 MiB,6,0,SKIP,SKIP,SKIP
relu/4d,PASS,0.059 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
relu/after_conv,PASS,0.062 s,0.00 MiB,0.00 MiB,4,3,SKIP,SKIP,SKIP
relu/after_gemm,PASS,0.062 s,0.01 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
relu/basic,PASS,0.062 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
reshape/4d_to_2d_flatten,PASS,0.059 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
reshape/infer_dim_minus_one,PASS,0.059 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
reshape/same_rank,PASS,0.056 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
reshape/zero_copies_input_dim,PASS,0.077 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
resize/height_only,PASS,0.059 s,0.00 MiB,0.00 MiB,4,0,SKIP,SKIP,SKIP
resize/nearest_2x,PASS,0.066 s,0.00 MiB,0.00 MiB,4,0,SKIP,SKIP,SKIP
resize/nearest_downsample,PASS,0.062 s,0.00 MiB,0.00 MiB,2,0,SKIP,SKIP,SKIP
resize/non_uniform,PASS,0.069 s,0.00 MiB,0.00 MiB,6,0,SKIP,SKIP,SKIP
resize/width_only,PASS,0.055 s,0.00 MiB,0.00 MiB,2,0,SKIP,SKIP,SKIP
resize/with_sizes,PASS,0.060 s,0.00 MiB,0.00 MiB,3,0,SKIP,SKIP,SKIP
sigmoid/4d,PASS,0.060 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
sigmoid/after_gemm,PASS,0.059 s,0.01 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
sigmoid/basic,PASS,0.053 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
slice/2d_basic,PASS,0.058 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
slice/after_conv,PASS,0.070 s,0.00 MiB,0.01 MiB,7,6,SKIP,SKIP,SKIP
slice/default_axes,PASS,0.066 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
slice/large_channel_1024,PASS,0.064 s,0.01 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
slice/nchw_spatial_crop,PASS,0.055 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
slice/negative_axis,PASS,0.060 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
slice/negative_indices,PASS,0.053 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
slice/step2,PASS,0.058 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
softmax/3d_last_axis,PASS,0.050 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
softmax/basic,PASS,0.061 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
softmax/channel_axis,PASS,0.058 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
softmax/large_dimension_1024,PASS,0.061 s,0.01 MiB,0.01 MiB,1,0,SKIP,SKIP,SKIP
softmax/negative_axis,PASS,0.058 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
split/basic,PASS,0.059 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
split/equal_three_way,PASS,0.063 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
split/negative_axis,PASS,0.059 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
split/uneven_channel_axis_4d,PASS,0.061 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
sub/after_gemm,PASS,0.064 s,0.01 MiB,0.01 MiB,5,4,SKIP,SKIP,SKIP
sub/basic,PASS,0.054 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
sub/broadcast_row,PASS,0.064 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
sub/channel_broadcast_1024,PASS,0.063 s,0.02 MiB,0.01 MiB,1,0,SKIP,SKIP,SKIP
sub/constant_lhs_broadcast,PASS,0.054 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
sub/leading_dimension_broadcast,PASS,0.054 s,0.00 MiB,0.00 MiB,1,0,SKIP,SKIP,SKIP
1 Operation Result Compile Host mem Cores mem Cores Xbars Latency Power Energy
2 add/after_gemm PASS 0.141 s 0.063 s 0.01 MiB 0.01 MiB 5 4 0.007784 ms SKIP 104.703618 mW SKIP 815012.960000 pJ SKIP
3 add/basic PASS 0.114 s 0.057 s 0.00 MiB 0.00 MiB 1 0 0.000323 ms SKIP 78.222910 mW SKIP 25266.000000 pJ SKIP
4 add/broadcast_row PASS 0.110 s 0.057 s 0.00 MiB 0.00 MiB 1 0 0.000323 ms SKIP 78.222910 mW SKIP 25266.000000 pJ SKIP
5 add/channel_broadcast_1024 PASS 0.105 s 0.061 s 0.02 MiB 0.01 MiB 1 0 0.006913 ms SKIP 78.118038 mW SKIP 540030.000000 pJ SKIP
6 add/leading_dimension_broadcast PASS 0.137 s 0.057 s 0.00 MiB 0.00 MiB 1 0 0.000323 ms SKIP 78.222910 mW SKIP 25266.000000 pJ SKIP
7 concat/channel_axis PASS 0.127 s 0.069 s 0.00 MiB 0.00 MiB 1 0 0.000457 ms SKIP 78.157549 mW SKIP 35718.000000 pJ SKIP
8 concat/negative_axis PASS 0.102 s 0.061 s 0.00 MiB 0.00 MiB 1 0 0.001043 ms SKIP 78.092042 mW SKIP 81450.000000 pJ SKIP
9 concat/three_inputs_channel_axis PASS 0.127 s 0.058 s 0.00 MiB 0.00 MiB 1 0 0.000644 ms SKIP 78.149068 mW SKIP 50328.000000 pJ SKIP
10 conv/batch_2 PASS 0.102 s 0.066 s 0.00 MiB 0.00 MiB 2 2 0.013694 ms SKIP 82.623885 mW SKIP 1131451.480000 pJ SKIP
11 conv/batch_4_pointwise PASS 0.110 s 0.065 s 0.00 MiB 0.01 MiB 5 4 0.003932 ms SKIP 116.078576 mW SKIP 456420.960000 pJ SKIP
12 conv/depthwise_1024_channels PASS 0.167 s 0.083 s 0.19 MiB 0.38 MiB 129 128 0.220751 ms SKIP 178.454307 mW SKIP 39393966.720000 pJ SKIP
13 conv/depthwise_grouped PASS 0.109 s 0.071 s 0.01 MiB 0.00 MiB 5 4 0.006024 ms SKIP 108.326521 mW SKIP 652558.960000 pJ SKIP
14 conv/dilated_3x3 PASS 0.143 s 0.068 s 0.00 MiB 0.01 MiB 0.00 MiB 0.01 MiB 3 10 3 9 0.004045 ms SKIP 110.234541 mW SKIP 445898.720000 pJ SKIP
15 conv/dynamic PASS 0.107 s 0.065 s 0.00 MiB 0.00 MiB 5 0 0.001835 ms SKIP 92.281199 mW SKIP 169336.000000 pJ SKIP
16 conv/explicit_padding PASS 0.139 s 0.062 s 0.00 MiB 0.01 MiB 0.00 MiB 0.02 MiB 4 17 4 16 0.004327 ms SKIP 115.794768 mW SKIP 501043.960000 pJ SKIP
17 conv/grouped_many_groups PASS 0.616 s 0.442 s 0.05 MiB 0.09 MiB 65 64 0.181845 ms SKIP 142.210104 mW SKIP 25860196.360000 pJ SKIP
18 conv/grouped_two_groups PASS 0.145 s 0.060 s 0.00 MiB 0.00 MiB 3 2 0.005360 ms SKIP 101.459418 mW SKIP 543822.480000 pJ SKIP
19 conv/huge_pointwise_1024 PASS 0.749 s 0.156 s 0.01 MiB 0.01 MiB 0.11 MiB 1 73 64 0.028261 ms SKIP 133.488743 mW SKIP 3772525.360000 pJ SKIP
20 conv/huge_pointwise_1024_dynamic PASS 0.097 s 0.079 s 8.04 MiB 12.61 MiB 168 0 2.627964 ms SKIP 169.518697 mW SKIP 445489032.000000 pJ SKIP
21 conv/kernel_3x3 PASS 0.155 s 0.062 s 0.00 MiB 0.01 MiB 0.00 MiB 0.01 MiB 3 10 3 9 0.003091 ms SKIP 115.862414 mW SKIP 358130.720000 pJ SKIP
22 conv/kernel_equals_input_spatial PASS 0.156 s 0.064 s 0.00 MiB 0.00 MiB 1 2 2 0.008443 ms SKIP 83.863849 mW SKIP 708062.480000 pJ SKIP
23 conv/large_input_channels_1x1 PASS 0.160 s 0.089 s 0.01 MiB 0.01 MiB 0.02 MiB 1 9 8 0.017167 ms SKIP 89.416900 mW SKIP 1535019.920000 pJ SKIP
24 conv/large_output_channels_1x1 PASS 0.141 s 0.098 s 0.00 MiB 0.01 MiB 0.01 MiB 0.02 MiB 1 17 8 0.004964 ms SKIP 117.628106 mW SKIP 583905.920000 pJ SKIP
25 conv/large_spatial PASS 0.141 s 0.069 s 0.00 MiB 0.01 MiB 0.01 MiB 0.04 MiB 6 37 6 36 0.004096 ms SKIP 129.015000 mW SKIP 528445.440000 pJ SKIP
26 conv/multi_channel PASS 0.143 s 0.066 s 0.00 MiB 0.00 MiB 3 4 3 0.005148 ms SKIP 106.453520 mW SKIP 548022.720000 pJ SKIP
27 conv/non_square_kernel_1x3 PASS 0.127 s 0.061 s 0.00 MiB 0.00 MiB 5 3 5 2 0.004029 ms SKIP 123.600943 mW SKIP 497988.200000 pJ SKIP
28 conv/non_square_kernel_3x1 PASS 0.085 s 0.064 s 0.00 MiB 0.00 MiB 3 3 2 0.005526 ms SKIP 105.464843 mW SKIP 582798.720000 pJ SKIP
29 conv/non_uniform_stride PASS 0.120 s 0.062 s 0.00 MiB 0.00 MiB 4 4 3 0.005808 ms SKIP 110.081433 mW SKIP 639352.960000 pJ SKIP
30 conv/pointwise_1x1 PASS 0.139 s 0.059 s 0.00 MiB 0.00 MiB 4 1 4 1 0.004539 ms SKIP 114.835858 mW SKIP 521239.960000 pJ SKIP
31 conv/pointwise_tiled_chain PASS 0.943 s 0.604 s 0.01 MiB 0.02 MiB 0.04 MiB 2 20 80 0.084437 ms SKIP 102.289307 mW SKIP 8637002.200000 pJ SKIP
32 conv/real_asymmetric_padding PASS 0.115 s 0.060 s 0.00 MiB 0.01 MiB 0.00 MiB 0.03 MiB 4 29 4 28 0.005232 ms SKIP 111.870214 mW SKIP 585304.960000 pJ SKIP
33 conv/relu_conv_store PASS 0.107 s 0.091 s 0.05 MiB 0.16 MiB 0.08 MiB 0.67 MiB 32 168 32 184 0.062978 ms SKIP 246.649827 mW SKIP 15533512.800000 pJ SKIP
34 conv/same_lower_3x3 PASS 0.094 s 0.078 s 0.00 MiB 0.01 MiB 0.00 MiB 0.02 MiB 5 26 5 25 0.004700 ms SKIP 119.232170 mW SKIP 560391.200000 pJ SKIP
35 conv/same_padding_3x3 PASS 0.126 s 0.070 s 0.00 MiB 0.01 MiB 0.00 MiB 0.02 MiB 5 26 5 25 0.004700 ms SKIP 119.232170 mW SKIP 560391.200000 pJ SKIP
36 conv/simple PASS 0.125 s 0.064 s 0.00 MiB 0.00 MiB 2 1 2 1 0.003148 ms SKIP 94.665972 mW SKIP 298008.480000 pJ SKIP
37 conv/stride_2 PASS 0.146 s 0.064 s 0.00 MiB 0.01 MiB 0.00 MiB 2 5 2 4 0.002827 ms SKIP 96.393873 mW SKIP 272505.480000 pJ SKIP
38 conv/with_bias_3x3 PASS 0.121 s 0.067 s 0.00 MiB 0.00 MiB 0.01 MiB 3 4 3 0.004898 ms SKIP 107.176546 mW SKIP 524950.720000 pJ SKIP
39 conv/with_constant PASS 0.106 s 0.070 s 0.00 MiB 0.00 MiB 3 1 3 1 0.004273 ms SKIP 109.362677 mW SKIP 467306.720000 pJ SKIP
40 conv/without_kernel_shape_attr PASS 0.105 s 0.069 s 0.00 MiB 0.01 MiB 0.00 MiB 0.01 MiB 3 10 3 9 0.003091 ms SKIP 115.862414 mW SKIP 358130.720000 pJ SKIP
41 conv/yolo11n_stem conv/yolo11n_depthwise_head PASS 3.091 s 1.482 s 29.20 MiB 8.66 MiB 31.38 MiB 34.24 MiB 168 488 255 9.799213 ms SKIP 361.075203 mW SKIP 3538252825.000010 pJ SKIP
42 div/after_gemm conv/yolo11n_heavy PASS 0.130 s 0.431 s 0.01 MiB 4.82 MiB 0.01 MiB 19.10 MiB 5 161 4 800 0.007784 ms SKIP 104.703618 mW SKIP 815012.960000 pJ SKIP
43 div/basic conv/yolo11n_stem PASS 0.116 s 0.783 s 0.00 MiB 12.86 MiB 0.00 MiB 37.59 MiB 1 168 0 488 0.000323 ms SKIP 78.222910 mW SKIP 25266.000000 pJ SKIP
44 div/channel_broadcast_1024 div/after_gemm PASS 0.121 s 0.072 s 0.02 MiB 0.01 MiB 0.01 MiB 1 5 0 4 0.006913 ms SKIP 78.118038 mW SKIP 540030.000000 pJ SKIP
45 div/leading_dimension_broadcast div/basic PASS 0.095 s 0.061 s 0.00 MiB 0.00 MiB 1 0 0.000323 ms SKIP 78.222910 mW SKIP 25266.000000 pJ SKIP
46 div/runtime_scalar_rhs div/channel_broadcast_1024 PASS 0.126 s 0.066 s 0.02 MiB 0.01 MiB 1 0 0.006913 ms SKIP 78.118038 mW SKIP 540030.000000 pJ SKIP
47 div/scalar_constant div/leading_dimension_broadcast PASS 0.069 s 0.059 s 0.00 MiB 0.00 MiB 1 0 0.000323 ms SKIP 78.222910 mW SKIP 25266.000000 pJ SKIP
48 gather/3d_input_axis1 div/runtime_scalar_rhs PASS 0.121 s 0.056 s 0.00 MiB 0.02 MiB 0.00 MiB 0.01 MiB 1 0 0.000589 ms SKIP 78.081494 mW SKIP 45990.000000 pJ SKIP
49 gather/axis0_matrix_indices div/scalar_constant PASS 0.088 s 0.073 s 0.00 MiB 0.00 MiB 1 0 0.000697 ms SKIP 78.068867 mW SKIP 54414.000000 pJ SKIP
50 gather/axis1 gather/3d_input_axis1 PASS 0.123 s 0.060 s 0.00 MiB 0.00 MiB 1 0 0.000801 ms SKIP 78.059925 mW SKIP 62526.000000 pJ SKIP
51 gather/negative_axis gather/axis0_matrix_indices PASS 0.087 s 0.056 s 0.00 MiB 0.00 MiB 1 0 0.001437 ms SKIP 78.033403 mW SKIP 112134.000000 pJ SKIP
52 gather/negative_indices gather/axis1 PASS 0.115 s 0.058 s 0.00 MiB 0.00 MiB 1 0 0.000376 ms SKIP 78.127660 mW SKIP 29376.000000 pJ SKIP
53 gemm/alpha_beta gather/negative_axis PASS 0.121 s 0.058 s 0.01 MiB 0.00 MiB 0.01 MiB 0.00 MiB 5 1 4 0 0.007456 ms SKIP 105.272125 mW SKIP 784908.960000 pJ SKIP
54 gemm/bias_rank2_broadcast gather/negative_indices PASS 0.127 s 0.063 s 0.00 MiB 0.01 MiB 0.00 MiB 5 1 4 0 0.007072 ms SKIP 105.979208 mW SKIP 749484.960000 pJ SKIP
55 gemm/dynamic gemm/alpha_beta PASS 0.084 s 0.067 s 0.00 MiB 0.01 MiB 0.00 MiB 0.01 MiB 5 0 4 0.002421 ms SKIP 91.480793 mW SKIP 221475.000000 pJ SKIP
56 gemm/dynamic_alpha gemm/bias_rank2_broadcast PASS 0.104 s 0.060 s 0.00 MiB 0.00 MiB 0.01 MiB 5 0 4 0.003262 ms SKIP 91.415696 mW SKIP 298198.000000 pJ SKIP
57 gemm/dynamic_beta gemm/dynamic PASS 0.110 s 0.064 s 0.00 MiB 0.00 MiB 5 0 0.004365 ms SKIP 91.316151 mW SKIP 398595.000000 pJ SKIP
58 gemm/dynamic_bias gemm/dynamic_alpha PASS 0.092 s 0.062 s 0.00 MiB 0.00 MiB 5 0 0.002665 ms SKIP 91.445779 mW SKIP 243703.000000 pJ SKIP
59 gemm/dynamic_bias_alpha_beta gemm/dynamic_beta PASS 0.089 s 0.060 s 0.00 MiB 0.00 MiB 5 0 0.005629 ms SKIP 91.279268 mW SKIP 513811.000000 pJ SKIP
60 gemm/dynamic_transB gemm/dynamic_bias PASS 0.077 s 0.058 s 0.00 MiB 0.00 MiB 5 0 0.001301 ms SKIP 91.378171 mW SKIP 118883.000000 pJ SKIP
61 gemm/huge_1024 gemm/dynamic_bias_alpha_beta PASS 0.219 s 0.067 s 0.01 MiB 0.00 MiB 0.10 MiB 0.00 MiB 73 5 64 0 0.017522 ms SKIP 215.037402 mW SKIP 3767885.360000 pJ SKIP
62 gemm/large gemm/dynamic_transB PASS 0.097 s 0.062 s 0.02 MiB 0.00 MiB 0.03 MiB 0.00 MiB 17 5 16 0 0.011229 ms SKIP 140.152181 mW SKIP 1573768.840000 pJ SKIP
63 gemm/large_k_small_n gemm/huge_1024 PASS 0.106 s 0.147 s 0.01 MiB 0.01 MiB 0.10 MiB 9 73 8 64 0.004748 ms SKIP 133.481449 mW SKIP 633769.920000 pJ SKIP
64 gemm/non_square gemm/large PASS 0.102 s 0.068 s 0.00 MiB 0.02 MiB 0.01 MiB 0.03 MiB 5 17 4 16 0.003527 ms SKIP 118.958310 mW SKIP 419565.960000 pJ SKIP
65 gemm/scalar_bias gemm/large_k_small_n PASS 0.111 s 0.095 s 0.00 MiB 0.01 MiB 0.01 MiB 5 9 4 8 0.007072 ms SKIP 105.979208 mW SKIP 749484.960000 pJ SKIP
66 gemm/simple gemm/non_square PASS 0.135 s 0.063 s 0.03 MiB 0.00 MiB 0.08 MiB 0.01 MiB 42 5 40 4 0.021640 ms SKIP 151.774196 mW SKIP 3284393.600000 pJ SKIP
67 gemm/small gemm/scalar_bias PASS 0.074 s 0.060 s 0.00 MiB 0.00 MiB 0.01 MiB 2 5 2 4 0.004420 ms SKIP 90.144000 mW SKIP 398436.480000 pJ SKIP
68 gemm/small_k_large_n gemm/simple PASS 0.129 s 0.072 s 0.01 MiB 0.03 MiB 0.02 MiB 0.08 MiB 17 42 8 40 0.007962 ms SKIP 131.005014 mW SKIP 1043061.920000 pJ SKIP
69 gemm/transA gemm/small PASS 0.089 s 0.065 s 0.00 MiB 0.01 MiB 0.00 MiB 5 2 4 2 0.005762 ms SKIP 109.140743 mW SKIP 628868.960000 pJ SKIP
70 gemm/transA_transB gemm/small_k_large_n PASS 0.115 s 0.097 s 0.00 MiB 0.01 MiB 0.01 MiB 0.02 MiB 5 17 4 8 0.005762 ms SKIP 109.140743 mW SKIP 628868.960000 pJ SKIP
71 gemm/transB gemm/transA PASS 0.068 s 0.064 s 0.00 MiB 0.01 MiB 5 4 0.003527 ms SKIP 118.958310 mW SKIP 419565.960000 pJ SKIP
72 gemm/transB_with_bias gemm/transA_transB PASS 0.069 s 0.01 MiB 0.00 MiB 0.01 MiB 5 4 0.005046 ms SKIP 110.546762 mW SKIP 557818.960000 pJ SKIP
73 gemm/with_bias gemm/transB PASS 0.113 s 0.062 s 0.01 MiB 0.00 MiB 0.01 MiB 5 4 0.005562 ms SKIP 108.767882 mW SKIP 604966.960000 pJ SKIP
74 gemv/constant gemm/transB_with_bias PASS 0.112 s 0.055 s 0.00 MiB 0.01 MiB 0.00 MiB 0.01 MiB 0 5 0 4 0.000000 ms SKIP 2.000000 mW SKIP 0.000000 pJ SKIP
75 gemv/simple gemm/with_bias PASS 0.136 s 0.067 s 0.00 MiB 0.01 MiB 0.01 MiB 6 5 4 0.005160 ms SKIP 111.150380 mW SKIP 573535.960000 pJ SKIP
76 gemv/with_heterogeneous_constant gemv/constant PASS 0.138 s 0.064 s 0.00 MiB 0.01 MiB 0.00 MiB 6 0 4 0 0.005549 ms SKIP 109.816536 mW SKIP 609371.960000 pJ SKIP
77 gemv/with_homogeneous_constant gemv/simple PASS 0.140 s 0.069 s 0.00 MiB 0.01 MiB 6 4 0.005549 ms SKIP 109.816536 mW SKIP 609371.960000 pJ SKIP
78 gemv/with_scalar_constant gemv/with_heterogeneous_constant PASS 0.124 s 0.066 s 0.00 MiB 0.01 MiB 6 4 0.005549 ms SKIP 109.816536 mW SKIP 609371.960000 pJ SKIP
79 matmul/basic gemv/with_homogeneous_constant PASS 0.081 s 0.070 s 0.00 MiB 0.00 MiB 0.01 MiB 2 6 2 4 0.004420 ms SKIP 90.144000 mW SKIP 398436.480000 pJ SKIP
80 matmul/batched_3d gemv/with_scalar_constant PASS 0.133 s 0.070 s 0.00 MiB 0.01 MiB 5 6 4 0.005958 ms SKIP 108.588949 mW SKIP 646972.960000 pJ SKIP
81 matmul/batched_3d_dynamic matmul/basic PASS 0.105 s 0.062 s 0.00 MiB 0.00 MiB 9 2 0 2 0.003525 ms SKIP 92.330213 mW SKIP 325464.000000 pJ SKIP
82 matmul/batched_left_constant matmul/batched_3d PASS 0.136 s 0.066 s 0.00 MiB 0.02 MiB 0.01 MiB 9 5 8 4 0.008822 ms SKIP 114.385164 mW SKIP 1009105.920000 pJ SKIP
83 matmul/batched_lhs_broadcast matmul/batched_3d_dynamic PASS 0.133 s 0.057 s 0.00 MiB 0.01 MiB 0.00 MiB 5 4 4 0 0.005681 ms SKIP 109.389361 mW SKIP 621440.960000 pJ SKIP
84 matmul/batched_rhs_broadcast matmul/batched_left_constant PASS 0.134 s 0.070 s 0.00 MiB 0.01 MiB 0.02 MiB 5 9 4 8 0.005958 ms SKIP 108.588949 mW SKIP 646972.960000 pJ SKIP
85 matmul/dynamic matmul/batched_lhs_broadcast PASS 0.093 s 0.063 s 0.00 MiB 0.00 MiB 0.01 MiB 5 0 4 0.001621 ms SKIP 91.421962 mW SKIP 148195.000000 pJ SKIP
86 matmul/huge_1024 matmul/batched_rhs_broadcast PASS 0.277 s 0.062 s 0.01 MiB 0.00 MiB 0.10 MiB 0.01 MiB 73 5 64 4 0.017522 ms SKIP 215.037402 mW SKIP 3767885.360000 pJ SKIP
87 matmul/left_constant matmul/dynamic PASS 0.123 s 0.060 s 0.00 MiB 0.01 MiB 0.00 MiB 5 4 0 0.005853 ms SKIP 108.861944 mW SKIP 637168.960000 pJ SKIP
88 matmul/matrix_vector matmul/huge_1024 PASS 0.150 s 0.145 s 0.52 MiB 0.01 MiB 0.78 MiB 0.10 MiB 168 73 173 64 0.384660 ms SKIP 202.131271 mW SKIP 77751814.880000 pJ SKIP
89 matmul/vector_matrix matmul/left_constant PASS 0.186 s 0.051 s 0.01 MiB 0.00 MiB 0.01 MiB 9 5 8 4 0.007409 ms SKIP 118.680243 mW SKIP 879301.920000 pJ SKIP
90 mul/after_conv matmul/matrix_vector PASS 0.127 s 0.095 s 0.00 MiB 0.52 MiB 0.00 MiB 0.78 MiB 4 168 3 173 0.005453 ms SKIP 107.639046 mW SKIP 586955.720000 pJ SKIP
91 mul/basic matmul/vector_matrix PASS 0.070 s 0.087 s 0.00 MiB 0.01 MiB 0.00 MiB 0.01 MiB 1 9 0 8 0.000323 ms SKIP 78.222910 mW SKIP 25266.000000 pJ SKIP
92 mul/channel_broadcast_1024 matmul/yolo_attention PASS 0.065 s 0.385 s 0.02 MiB 1.02 MiB 0.01 MiB 43.44 MiB 1 168 0 0.006913 ms SKIP 78.118038 mW SKIP 540030.000000 pJ SKIP
93 mul/leading_dimension_broadcast mul/after_conv PASS 0.108 s 0.055 s 0.00 MiB 0.00 MiB 1 4 0 3 0.000323 ms SKIP 78.222910 mW SKIP 25266.000000 pJ SKIP
94 mul/scalar_constant mul/basic PASS 0.082 s 0.055 s 0.00 MiB 0.00 MiB 1 0 0.000323 ms SKIP 78.222910 mW SKIP 25266.000000 pJ SKIP
95 pool/avg_basic mul/channel_broadcast_1024 PASS 0.070 s 0.058 s 0.00 MiB 0.02 MiB 0.00 MiB 0.01 MiB 1 0 0.011939 ms SKIP 78.022112 mW SKIP 931506.000000 pJ SKIP
96 pool/avg_ceil_mode mul/leading_dimension_broadcast PASS 0.107 s 0.059 s 0.00 MiB 0.00 MiB 1 0 0.004359 ms SKIP 78.033035 mW SKIP 340146.000000 pJ SKIP
97 pool/avg_explicit_padding mul/scalar_constant PASS 0.111 s 0.056 s 0.00 MiB 0.00 MiB 1 0 0.008822 ms SKIP 78.027205 mW SKIP 688356.000000 pJ SKIP
98 pool/avg_include_pad pool/avg_basic PASS 0.075 s 0.063 s 0.00 MiB 0.00 MiB 1 0 0.008506 ms SKIP 78.016929 mW SKIP 663612.000000 pJ SKIP
99 pool/avg_large_channels pool/avg_ceil_mode PASS 0.064 s 0.063 s 0.04 MiB 0.00 MiB 0.02 MiB 0.00 MiB 1 0 0.178249 ms SKIP 78.280327 mW SKIP 13953390.000000 pJ SKIP
100 pool/avg_non_uniform_stride pool/avg_explicit_padding PASS 0.068 s 0.061 s 0.00 MiB 0.00 MiB 1 0 0.014513 ms SKIP 78.016537 mW SKIP 1132254.000000 pJ SKIP
101 pool/avg_real_asymmetric_padding pool/avg_include_pad PASS 0.064 s 0.049 s 0.00 MiB 0.00 MiB 1 0 0.025206 ms SKIP 78.024756 mW SKIP 1966692.000000 pJ SKIP
102 pool/max_after_conv pool/avg_large_channels PASS 0.070 s 0.059 s 0.00 MiB 0.04 MiB 0.00 MiB 0.02 MiB 6 1 4 0 0.006452 ms SKIP 96.374606 mW SKIP 621808.960000 pJ SKIP
103 pool/max_basic pool/avg_non_uniform_stride PASS 0.063 s 0.066 s 0.00 MiB 0.00 MiB 3 1 0 0.001634 ms SKIP 92.132191 mW SKIP 150544.000000 pJ SKIP
104 pool/max_ceil_mode pool/avg_real_asymmetric_padding PASS 0.065 s 0.055 s 0.00 MiB 0.00 MiB 2 1 0 0.001297 ms SKIP 79.111025 mW SKIP 102607.000000 pJ SKIP
105 pool/max_global_style_kernel_equals_input pool/max_after_conv PASS 0.089 s 0.069 s 0.00 MiB 0.00 MiB 1 5 0 4 0.004366 ms SKIP 78.010994 mW SKIP 340596.000000 pJ SKIP
106 pool/max_non_square_kernel pool/max_basic PASS 0.103 s 0.064 s 0.00 MiB 0.00 MiB 4 1 0 0.003409 ms SKIP 93.253447 mW SKIP 317901.000000 pJ SKIP
107 pool/max_real_asymmetric_padding pool/max_ceil_mode PASS 0.088 s 0.058 s 0.00 MiB 0.00 MiB 4 1 0 0.003078 ms SKIP 93.124756 mW SKIP 286638.000000 pJ SKIP
108 pool/max_same_upper pool/max_global_style_kernel_equals_input PASS 0.065 s 0.063 s 0.00 MiB 0.00 MiB 3 1 0 0.003024 ms SKIP 92.095238 mW SKIP 278496.000000 pJ SKIP
109 pool/max_stride2_multichannel pool/max_non_square_kernel PASS 0.086 s 0.067 s 0.00 MiB 0.00 MiB 3 1 0 0.004012 ms SKIP 92.269192 mW SKIP 370184.000000 pJ SKIP
110 reduce_mean/4d_spatial pool/max_real_asymmetric_padding PASS 0.061 s 0.063 s 0.00 MiB 0.00 MiB 3 1 0 0.000321 ms SKIP 92.448598 mW SKIP 29676.000000 pJ SKIP
111 reduce_mean/4d_spatial_keepdims_0 pool/max_same_upper PASS 0.067 s 0.061 s 0.00 MiB 0.00 MiB 4 1 0 0.000655 ms SKIP 94.352672 mW SKIP 61801.000000 pJ SKIP
112 reduce_mean/after_conv pool/max_stride2_multichannel PASS 0.115 s 0.076 s 0.00 MiB 0.00 MiB 5 1 3 0 0.005342 ms SKIP 106.951089 mW SKIP 571332.720000 pJ SKIP
113 reduce_mean/all_axes_keepdims_0 reduce_mean/4d_spatial PASS 0.103 s 0.058 s 0.00 MiB 0.00 MiB 2 3 0 0.000391 ms SKIP 79.237852 mW SKIP 30982.000000 pJ SKIP
114 reduce_mean/all_axes_keepdims_1 reduce_mean/4d_spatial_keepdims_0 PASS 0.086 s 0.068 s 0.00 MiB 0.00 MiB 1 4 0 0.000221 ms SKIP 78.217195 mW SKIP 17286.000000 pJ SKIP
115 reduce_mean/basic reduce_mean/after_conv PASS 0.099 s 0.067 s 0.00 MiB 0.00 MiB 4 5 0 3 0.000373 ms SKIP 93.514745 mW SKIP 34881.000000 pJ SKIP
116 reduce_mean/channel_axis_nchw reduce_mean/all_axes_keepdims_0 PASS 0.078 s 0.057 s 0.03 MiB 0.00 MiB 0.02 MiB 0.00 MiB 4 2 0 0.164926 ms SKIP 93.596631 mW SKIP 15436518.000000 pJ SKIP
117 reduce_mean/keepdims_0 reduce_mean/all_axes_keepdims_1 PASS 0.073 s 0.053 s 0.00 MiB 0.00 MiB 5 1 0 0.000748 ms SKIP 91.401070 mW SKIP 68368.000000 pJ SKIP
118 reduce_mean/large_dimension_1024 reduce_mean/basic PASS 0.115 s 0.058 s 0.01 MiB 0.00 MiB 0.00 MiB 1 4 0 0.002785 ms SKIP 78.017235 mW SKIP 217278.000000 pJ SKIP
119 reduce_mean/legacy_axes_1_2_keepdims_1 reduce_mean/channel_axis_nchw PASS 0.074 s 0.063 s 0.00 MiB 0.03 MiB 0.00 MiB 0.02 MiB 2 4 0 0.000271 ms SKIP 79.354244 mW SKIP 21505.000000 pJ SKIP
120 reduce_mean/legacy_axis1_keepdims_0 reduce_mean/keepdims_0 PASS 0.111 s 0.064 s 0.00 MiB 0.00 MiB 9 5 0 0.001986 ms SKIP 92.501511 mW SKIP 183708.000000 pJ SKIP
121 reduce_mean/legacy_axis1_keepdims_1 reduce_mean/large_dimension_1024 PASS 0.086 s 0.066 s 0.00 MiB 0.01 MiB 0.00 MiB 8 1 0 0.001373 ms SKIP 94.559359 mW SKIP 129830.000000 pJ SKIP
122 reduce_mean/legacy_empty_axes_noop reduce_mean/legacy_axes_1_2_keepdims_1 PASS 0.115 s 0.069 s 0.00 MiB 0.00 MiB 1 2 0 0.000221 ms SKIP 78.217195 mW SKIP 17286.000000 pJ SKIP
123 reduce_mean/legacy_nchw_spatial reduce_mean/legacy_axis1_keepdims_0 PASS 0.090 s 0.067 s 0.00 MiB 0.00 MiB 3 9 0 0.000321 ms SKIP 92.448598 mW SKIP 29676.000000 pJ SKIP
124 reduce_mean/legacy_negative_axis reduce_mean/legacy_axis1_keepdims_1 PASS 0.110 s 0.057 s 0.00 MiB 0.00 MiB 6 8 0 0.000553 ms SKIP 93.520796 mW SKIP 51717.000000 pJ SKIP
125 reduce_mean/legacy_reduce_all_keepdims_1 reduce_mean/legacy_empty_axes_noop PASS 0.107 s 0.058 s 0.00 MiB 0.00 MiB 1 0 0.000221 ms SKIP 78.217195 mW SKIP 17286.000000 pJ SKIP
126 reduce_mean/negative_axis reduce_mean/legacy_nchw_spatial PASS 0.120 s 0.059 s 0.00 MiB 0.00 MiB 6 3 0 0.000553 ms SKIP 93.520796 mW SKIP 51717.000000 pJ SKIP
127 relu/4d reduce_mean/legacy_negative_axis PASS 0.116 s 0.052 s 0.00 MiB 0.00 MiB 1 6 0 0.000521 ms SKIP 78.184261 mW SKIP 40734.000000 pJ SKIP
128 relu/after_conv reduce_mean/legacy_reduce_all_keepdims_1 PASS 0.078 s 0.057 s 0.00 MiB 0.00 MiB 3 1 3 0 0.004956 ms SKIP 106.998935 mW SKIP 530286.720000 pJ SKIP
129 relu/after_gemm reduce_mean/negative_axis PASS 0.077 s 0.055 s 0.01 MiB 0.00 MiB 0.01 MiB 0.00 MiB 5 6 4 0 0.007513 ms SKIP 105.158653 mW SKIP 790056.960000 pJ SKIP
130 relu/basic relu/4d PASS 0.066 s 0.059 s 0.00 MiB 0.00 MiB 1 0 0.000221 ms SKIP 78.217195 mW SKIP 17286.000000 pJ SKIP
131 reshape/4d_to_2d_flatten relu/after_conv PASS 0.098 s 0.062 s 0.00 MiB 0.00 MiB 1 4 0 3 0.000258 ms SKIP 78.279070 mW SKIP 20196.000000 pJ SKIP
132 reshape/infer_dim_minus_one relu/after_gemm PASS 0.105 s 0.062 s 0.00 MiB 0.01 MiB 0.00 MiB 0.01 MiB 1 5 0 4 0.000162 ms SKIP 78.296296 mW SKIP 12684.000000 pJ SKIP
133 reshape/same_rank relu/basic PASS 0.057 s 0.062 s 0.00 MiB 0.00 MiB 1 0 0.000162 ms SKIP 78.296296 mW SKIP 12684.000000 pJ SKIP
134 reshape/zero_copies_input_dim reshape/4d_to_2d_flatten PASS 0.090 s 0.059 s 0.00 MiB 0.00 MiB 1 0 0.000162 ms SKIP 78.296296 mW SKIP 12684.000000 pJ SKIP
135 resize/height_only reshape/infer_dim_minus_one PASS 0.104 s 0.059 s 0.00 MiB 0.00 MiB 1 0 0.000795 ms SKIP 78.060377 mW SKIP 62058.000000 pJ SKIP
136 resize/nearest_2x reshape/same_rank PASS 0.104 s 0.056 s 0.00 MiB 0.00 MiB 1 0 0.001422 ms SKIP 78.033755 mW SKIP 110964.000000 pJ SKIP
137 resize/nearest_downsample reshape/zero_copies_input_dim PASS 0.059 s 0.077 s 0.00 MiB 0.00 MiB 1 0 0.000481 ms SKIP 78.099792 mW SKIP 37566.000000 pJ SKIP
138 resize/non_uniform resize/height_only PASS 0.063 s 0.059 s 0.00 MiB 0.00 MiB 1 4 0 0.002108 ms SKIP 78.034156 mW SKIP 164496.000000 pJ SKIP
139 resize/width_only resize/nearest_2x PASS 0.068 s 0.066 s 0.00 MiB 0.00 MiB 1 4 0 0.000792 ms SKIP 78.060606 mW SKIP 61824.000000 pJ SKIP
140 resize/with_sizes resize/nearest_downsample PASS 0.056 s 0.062 s 0.00 MiB 0.00 MiB 1 2 0 0.000951 ms SKIP 78.050473 mW SKIP 74226.000000 pJ SKIP
141 sigmoid/4d resize/non_uniform PASS 0.104 s 0.069 s 0.00 MiB 0.00 MiB 1 6 0 0.000521 ms SKIP 78.184261 mW SKIP 40734.000000 pJ SKIP
142 sigmoid/after_gemm resize/width_only PASS 0.118 s 0.055 s 0.01 MiB 0.00 MiB 0.01 MiB 0.00 MiB 5 2 4 0 0.007513 ms SKIP 105.158653 mW SKIP 790056.960000 pJ SKIP
143 sigmoid/basic resize/with_sizes PASS 0.083 s 0.060 s 0.00 MiB 0.00 MiB 1 3 0 0.000221 ms SKIP 78.217195 mW SKIP 17286.000000 pJ SKIP
144 slice/2d_basic sigmoid/4d PASS 0.067 s 0.060 s 0.00 MiB 0.00 MiB 1 0 0.000242 ms SKIP 78.297521 mW SKIP 18948.000000 pJ SKIP
145 slice/after_conv sigmoid/after_gemm PASS 0.111 s 0.059 s 0.00 MiB 0.01 MiB 0.01 MiB 7 5 6 4 0.011296 ms SKIP 118.190765 mW SKIP 1335082.880000 pJ SKIP
146 slice/default_axes sigmoid/basic PASS 0.065 s 0.053 s 0.00 MiB 0.00 MiB 1 0 0.000242 ms SKIP 78.297521 mW SKIP 18948.000000 pJ SKIP
147 slice/large_channel_1024 slice/2d_basic PASS 0.060 s 0.058 s 0.01 MiB 0.00 MiB 0.00 MiB 1 0 0.002832 ms SKIP 78.144068 mW SKIP 221304.000000 pJ SKIP
148 slice/nchw_spatial_crop slice/after_conv PASS 0.067 s 0.070 s 0.00 MiB 0.00 MiB 0.01 MiB 1 7 0 6 0.001302 ms SKIP 78.239631 mW SKIP 101868.000000 pJ SKIP
149 slice/negative_axis slice/default_axes PASS 0.098 s 0.066 s 0.00 MiB 0.00 MiB 1 0 0.000562 ms SKIP 78.298932 mW SKIP 44004.000000 pJ SKIP
150 slice/negative_indices slice/large_channel_1024 PASS 0.106 s 0.064 s 0.00 MiB 0.01 MiB 0.00 MiB 1 0 0.000322 ms SKIP 78.298137 mW SKIP 25212.000000 pJ SKIP
151 slice/step2 slice/nchw_spatial_crop PASS 0.068 s 0.055 s 0.00 MiB 0.00 MiB 1 0 0.002042 ms SKIP 78.293830 mW SKIP 159876.000000 pJ SKIP
152 softmax/3d_last_axis slice/negative_axis PASS 0.072 s 0.060 s 0.00 MiB 0.00 MiB 1 0 UNSUPPORTED SKIP UNSUPPORTED SKIP UNSUPPORTED SKIP
153 softmax/basic slice/negative_indices PASS 0.088 s 0.053 s 0.00 MiB 0.00 MiB 1 0 UNSUPPORTED SKIP UNSUPPORTED SKIP UNSUPPORTED SKIP
154 softmax/channel_axis slice/step2 PASS 0.110 s 0.058 s 0.00 MiB 0.00 MiB 1 0 UNSUPPORTED SKIP UNSUPPORTED SKIP UNSUPPORTED SKIP
155 softmax/large_dimension_1024 softmax/3d_last_axis PASS 0.104 s 0.050 s 0.01 MiB 0.00 MiB 0.01 MiB 0.00 MiB 1 0 UNSUPPORTED SKIP UNSUPPORTED SKIP UNSUPPORTED SKIP
156 softmax/negative_axis softmax/basic PASS 0.118 s 0.061 s 0.00 MiB 0.00 MiB 1 0 UNSUPPORTED SKIP UNSUPPORTED SKIP UNSUPPORTED SKIP
157 split/basic softmax/channel_axis PASS 0.077 s 0.058 s 0.00 MiB 0.00 MiB 1 0 0.000403 ms SKIP 78.297767 mW SKIP 31554.000000 pJ SKIP
158 split/equal_three_way softmax/large_dimension_1024 PASS 0.107 s 0.061 s 0.00 MiB 0.01 MiB 0.00 MiB 0.01 MiB 1 0 0.000564 ms SKIP 78.297872 mW SKIP 44160.000000 pJ SKIP
159 split/negative_axis softmax/negative_axis PASS 0.111 s 0.058 s 0.00 MiB 0.00 MiB 1 0 0.001083 ms SKIP 78.288089 mW SKIP 84786.000000 pJ SKIP
160 split/uneven_channel_axis_4d split/basic PASS 0.061 s 0.059 s 0.00 MiB 0.00 MiB 1 0 0.000242 ms SKIP 78.297521 mW SKIP 18948.000000 pJ SKIP
161 sub/after_gemm split/equal_three_way PASS 0.068 s 0.063 s 0.01 MiB 0.00 MiB 0.01 MiB 0.00 MiB 5 1 4 0 0.007784 ms SKIP 104.703618 mW SKIP 815012.960000 pJ SKIP
162 sub/basic split/negative_axis PASS 0.063 s 0.059 s 0.00 MiB 0.00 MiB 1 0 0.000323 ms SKIP 78.222910 mW SKIP 25266.000000 pJ SKIP
163 sub/broadcast_row split/uneven_channel_axis_4d PASS 0.075 s 0.061 s 0.00 MiB 0.00 MiB 1 0 0.000323 ms SKIP 78.222910 mW SKIP 25266.000000 pJ SKIP
164 sub/channel_broadcast_1024 sub/after_gemm PASS 0.052 s 0.064 s 0.02 MiB 0.01 MiB 0.01 MiB 1 5 0 4 0.006913 ms SKIP 78.118038 mW SKIP 540030.000000 pJ SKIP
165 sub/constant_lhs_broadcast sub/basic PASS 0.057 s 0.054 s 0.00 MiB 0.00 MiB 1 0 0.000322 ms SKIP 78.223602 mW SKIP 25188.000000 pJ SKIP
166 sub/leading_dimension_broadcast sub/broadcast_row PASS 0.053 s 0.064 s 0.00 MiB 0.00 MiB 1 0 0.000323 ms SKIP 78.222910 mW SKIP 25266.000000 pJ SKIP
167 sub/channel_broadcast_1024 PASS 0.063 s 0.02 MiB 0.01 MiB 1 0 SKIP SKIP SKIP
168 sub/constant_lhs_broadcast PASS 0.054 s 0.00 MiB 0.00 MiB 1 0 SKIP SKIP SKIP
169 sub/leading_dimension_broadcast PASS 0.054 s 0.00 MiB 0.00 MiB 1 0 SKIP SKIP SKIP