ChaiBapchya/mxnet_operator_benchmark_results_cpu.md

## mxnet_operator_benchmark_results_cpu.md

      
    Raw
  

              mxnet_operator_benchmark_results_cpu.md
            
          
    Runtime Features


BLAS_APPLE : ✖ BLAS_APPLE
BLAS_ATLAS : ✖ BLAS_ATLAS
BLAS_MKL : ✖ BLAS_MKL
BLAS_OPEN : ✔ BLAS_OPEN
CAFFE : ✖ CAFFE
CPU_AVX : ✔ CPU_AVX
CPU_AVX2 : ✖ CPU_AVX2
CPU_SSE : ✔ CPU_SSE
CPU_SSE2 : ✔ CPU_SSE2
CPU_SSE3 : ✔ CPU_SSE3
CPU_SSE4A : ✖ CPU_SSE4A
CPU_SSE4_1 : ✔ CPU_SSE4_1
CPU_SSE4_2 : ✔ CPU_SSE4_2
CUDA : ✔ CUDA
CUDA_RTC : ✔ CUDA_RTC
CUDNN : ✔ CUDNN
CXX14 : ✖ CXX14
DEBUG : ✖ DEBUG
DIST_KVSTORE : ✖ DIST_KVSTORE
F16C : ✔ F16C
INT64_TENSOR_SIZE : ✖ INT64_TENSOR_SIZE
JEMALLOC : ✔ JEMALLOC
LAPACK : ✔ LAPACK
MKLDNN : ✔ MKLDNN
NCCL : ✖ NCCL
OPENCV : ✔ OPENCV
OPENMP : ✔ OPENMP
PROFILER : ✖ PROFILER
SIGNAL_HANDLER : ✔ SIGNAL_HANDLER
SSE : ✖ SSE
TENSORRT : ✖ TENSORRT
TVM_OP : ✖ TVM_OP

Benchmark Results


Operator
Avg Forward Time (ms)
Avg. Backward Time (ms)
Max Mem Usage (Storage) (Bytes)
Inputs


BatchNorm
2.5248
1.2922
25165.8359
{'data': (32, 3, 256, 256), 'gamma': (3,), 'beta': (3,), 'moving_mean': (3,), 'moving_var': (3,)}


BatchNorm
5.8611
1.524
38400.0117
{'data': (32, 3, 10000, 10), 'gamma': (3,), 'beta': (3,), 'moving_mean': (3,), 'moving_var': (3,)}


BlockGrad
0.3609
---
2097.1521
{'data': (1024, 1024)}


BlockGrad
0.0089
---
20.0
{'data': (10000, 1)}


BlockGrad
0.3618
---
2000.0
{'data': (10000, 100)}


Convolution
0.5094
0.9317
2080.7681
{'data': (32, 3, 256), 'weight': (64, 3, 3), 'bias': (64,), 'kernel': (3,), 'stride': (1,), 'dilate': (1,), 'pad': (0,), 'num_filter': 64, 'layout': 'NCW'}


Deconvolution
2.2693
6.6616
2704448.25
{'data': (32, 3, 256), 'weight': (3, 64, 3), 'bias': (64,), 'kernel': (3,), 'stride': (1,), 'dilate': (1,), 'pad': (0,), 'adj': (0,), 'num_filter': 64, 'no_bias': False, 'layout': 'NCW'}


Dropout
2.6628
0.5423
37748.7344
{'data': (32, 3, 256, 256), 'p': 0.5, 'mode': 'always'}


Dropout
0.0831
0.0349
600.0
{'data': (10000, 10), 'p': 0.5, 'mode': 'always'}


FullyConnected
1.5243
98.0013
8.192
{'data': (32, 3, 256, 256), 'num_hidden': 64, 'weight': (64, 196608), 'bias': (64,), 'flatten': True}


FullyConnected
1.0998
30.8501
6291.4561
{'data': (32, 3, 256, 256), 'num_hidden': 64, 'weight': (64, 256), 'bias': (64,), 'flatten': False}


LeakyReLU
0.0667
0.0676
4194.3042
{'data': (1024, 1024), 'act_type': 'leaky', 'slope': 0.1}


LeakyReLU
0.0534
0.049
20.0
{'data': (10000, 1), 'act_type': 'leaky', 'slope': 0.1}


LeakyReLU
0.0865
0.0811
2000.0
{'data': (10000, 100), 'act_type': 'leaky', 'slope': 0.1}


LeakyReLU
0.1206
0.7309
2097.1521
{'data': (1024, 1024), 'act_type': 'elu', 'slope': 0.1}


LeakyReLU
0.0448
0.04
20.0
{'data': (10000, 1), 'act_type': 'elu', 'slope': 0.1}


LeakyReLU
0.1212
0.7229
2000.0
{'data': (10000, 100), 'act_type': 'elu', 'slope': 0.1}


LeakyReLU
0.704
0.2679
2097.1521
{'data': (1024, 1024), 'act_type': 'selu'}


LeakyReLU
0.0319
0.0287
20.0
{'data': (10000, 1), 'act_type': 'selu'}


LeakyReLU
0.6942
0.2591
2000.0
{'data': (10000, 100), 'act_type': 'selu'}


LeakyReLU
0.4296
3.7836
4194.3042
{'data': (1024, 1024), 'act_type': 'prelu', 'gamma': (1, 1024)}


LeakyReLU
0.0412
0.2741
20.0
{'data': (10000, 1), 'act_type': 'prelu', 'gamma': (1, 1)}


LeakyReLU
0.4179
3.2919
4000.0
{'data': (10000, 100), 'act_type': 'prelu', 'gamma': (1, 100)}


Pooling
0.2986
0.3777
98.304
{'data': (32, 3, 256), 'kernel': 3, 'pool_type': 'avg', 'global_pool': 0, 'stride': 1, 'pad': 1}


abs
0.0513
0.3088
4194.3042
{'data': (1024, 1024)}


abs
0.0323
0.0748
20.0
{'data': (10000, 1)}


abs
0.5531
0.7512
4000.0
{'data': (10000, 100)}


adam_update
0.4475
---
2639212.0
{'weight': (1024, 1024), 'grad': (1024, 1024), 'mean': (1024, 1024), 'var': (1024, 1024), 'lr': 0.1, 'beta1': 0.1, 'beta2': 0.1, 'epsilon': 1e-08, 'rescale_grad': 0.4, 'lazy_update': 0}


adam_update
0.0538
---
2637154.75
{'weight': (10000, 1), 'grad': (10000, 1), 'mean': (10000, 1), 'var': (10000, 1), 'lr': 0.5, 'beta1': 0.5, 'beta2': 0.5, 'epsilon': 1e-08, 'rescale_grad': 0.4, 'lazy_update': 0}


adam_update
0.6825
---
2641134.75
{'weight': (10000, 100), 'grad': (10000, 100), 'mean': (10000, 100), 'var': (10000, 100), 'lr': 0.9, 'beta1': 0.9, 'beta2': 0.9, 'epsilon': 1e-08, 'rescale_grad': 0.4, 'lazy_update': 0}


arccos
1.0392
0.6424
2097.1521
{'data': (1024, 1024)}


arccos
0.0344
0.0292
20.0
{'data': (10000, 1)}


arccos
0.769
0.6015
2000.0
{'data': (10000, 100)}


arccosh
0.6852
0.7021
2097.1521
{'data': (1024, 1024)}


arccosh
0.04
0.0237
20.0
{'data': (10000, 1)}


arccosh
0.6658
0.671
2000.0
{'data': (10000, 100)}


arcsin
0.9986
0.7568
2097.1521
{'data': (1024, 1024)}


arcsin
0.0267
0.0241
20.0
{'data': (10000, 1)}


arcsin
0.7345
0.6657
2000.0
{'data': (10000, 100)}


arcsinh
1.4487
0.4922
2097.1521
{'data': (1024, 1024)}


arcsinh
0.0401
0.0318
20.0
{'data': (10000, 1)}


arcsinh
1.6481
0.4774
2000.0
{'data': (10000, 100)}


arctan
0.8992
0.0822
2097.1521
{'data': (1024, 1024)}


arctan
0.0323
0.0258
20.0
{'data': (10000, 1)}


arctan
0.94
0.0845
2000.0
{'data': (10000, 100)}


arctanh
1.3642
0.0813
4194.3042
{'data': (1024, 1024)}


arctanh
0.0345
0.0247
20.0
{'data': (10000, 1)}


arctanh
1.3843
0.1367
2000.0
{'data': (10000, 100)}


argmax
10.5619
---
4.096
{'data': (1024, 1024), 'axis': 0}


argmax
0.0614
---
0.002
{'data': (10000, 1), 'axis': 0}


argmax
6.553
---
0.2
{'data': (10000, 100), 'axis': 0}


argmax_channel
0.2887
---
2.048
{'data': (1024, 1024)}


argmax_channel
0.0364
---
20.0
{'data': (10000, 1)}


argmax_channel
0.3419
---
40.0
{'data': (10000, 100)}


argmin
7.2113
---
2.048
{'data': (1024, 1024), 'axis': 0}


argmin
0.0877
---
0.002
{'data': (10000, 1), 'axis': 0}


argmin
6.1747
---
0.2
{'data': (10000, 100), 'axis': 0}


argsort
36.1523
---
4194.3042
{'data': (1024, 1024), 'axis': 0}


argsort
1.1341
---
40.0
{'data': (10000, 1), 'axis': 0}


argsort
28.6094
---
4000.0
{'data': (10000, 100), 'axis': 0}


batch_dot
566.924
---
67108.8672
{'lhs': (32, 1024, 1024), 'rhs': (32, 1024, 1024)}


batch_dot
381.4693
---
128000.0
{'lhs': (32, 1000, 10), 'rhs': (32, 1000, 10), 'transpose_b': True}


batch_dot
3.2809
---
12.8
{'lhs': (32, 1000, 1), 'rhs': (32, 100, 1000), 'transpose_a': True, 'transpose_b': True}


broadcast_add
0.0693
0.0825
2097.1521
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_add
0.038
0.0458
200.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_add
0.0297
0.0327
20.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_div
0.072
0.1186
2097.1521
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_div
0.0365
0.0493
400.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_div
0.0327
0.0549
20.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_equal
0.1043
---
4194.3042
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_equal
0.0389
---
200.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_equal
0.0329
---
20.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_greater
0.1042
---
2097.1521
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_greater
0.0366
---
400.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_greater
0.0373
---
20.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_greater_equal
0.0708
---
2097.1521
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_greater_equal
0.042
---
200.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_greater_equal
0.0437
---
20.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_hypot
0.4855
1.0332
4194.3042
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_hypot
0.0746
0.1288
200.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_hypot
0.0403
0.0546
20.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_lesser
0.2353
---
4194.3042
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_lesser
0.0346
---
400.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_lesser
0.0355
---
40.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_lesser_equal
0.1437
---
4194.3042
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_lesser_equal
0.0337
---
400.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_lesser_equal
0.0307
---
40.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_logical_and
0.0988
---
2097.1521
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_logical_and
0.0309
---
200.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_logical_and
0.0328
---
20.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_logical_or
0.1435
---
4194.3042
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_logical_or
0.033
---
200.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_logical_or
0.0358
---
40.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_logical_xor
0.2049
---
4194.3042
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_logical_xor
0.0409
---
400.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_logical_xor
0.0416
---
40.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_maximum
0.1068
0.124
2097.1521
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_maximum
0.0385
0.0452
200.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_maximum
0.0386
0.0631
20.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_minimum
0.0786
0.1034
2097.1521
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_minimum
0.0415
0.0547
200.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_minimum
0.0378
0.0525
20.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_minus
0.0741
---
2097.1521
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_minus
0.04
---
200.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_minus
0.032
---
20.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_mod
1.4273
0.1419
2097.1521
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_mod
0.2105
0.0541
200.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_mod
0.051
0.0662
20.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_mul
0.0721
0.1085
2097.1521
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_mul
0.038
0.0516
400.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_mul
0.0327
0.0416
20.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_not_equal
0.3561
---
4194.3042
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_not_equal
0.0382
---
400.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_not_equal
0.0323
---
40.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_plus
0.1912
---
4194.3042
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_plus
0.0372
---
400.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_plus
0.0293
---
40.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_power
1.8165
3.9704
2097.1521
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_power
0.2463
0.416
200.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_power
0.0529
0.0814
20.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


broadcast_sub
0.0642
0.0722
2097.1521
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


broadcast_sub
0.0417
0.0541
200.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


broadcast_sub
0.0349
0.0393
20.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


cbrt
1.3988
0.0834
2097.1521
{'data': (1024, 1024)}


cbrt
0.039
0.028
20.0
{'data': (10000, 1)}


cbrt
1.3426
0.0825
2000.0
{'data': (10000, 100)}


ceil
0.1624
---
4194.3042
{'data': (1024, 1024)}


ceil
0.0484
---
20.0
{'data': (10000, 1)}


ceil
0.1578
---
2000.0
{'data': (10000, 100)}


cos
0.5093
0.5672
2097.1521
{'data': (1024, 1024)}


cos
0.1051
0.0349
20.0
{'data': (10000, 1)}


cos
0.4706
0.5068
2000.0
{'data': (10000, 100)}


cosh
0.8828
1.5931
2097.1521
{'data': (1024, 1024)}


cosh
0.0329
0.0353
20.0
{'data': (10000, 1)}


cosh
0.778
1.3467
4000.0
{'data': (10000, 100)}


degrees
0.0497
0.0461
2097.1521
{'data': (1024, 1024)}


degrees
0.0274
0.0259
20.0
{'data': (10000, 1)}


degrees
0.058
0.0506
2000.0
{'data': (10000, 100)}


depth_to_space
0.0482
---
0.064
{'data': (1, 4, 2, 4), 'block_size': 2}


depth_to_space
0.4207
---
500.0
{'data': (10, 25, 10, 100), 'block_size': 5}


dot
3.4368
5.7041
4194.3042
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


dot
0.4234
4.5814
2000.0
{'lhs': (1000, 10), 'rhs': (1000, 10), 'transpose_b': True}


dot
0.1543
0.2657
0.2
{'lhs': (1000, 1), 'rhs': (100, 1000), 'transpose_a': True, 'transpose_b': True}


elemwise_add
0.1565
---
4194.3042
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


elemwise_add
0.0459
---
400.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


elemwise_add
0.0424
---
40.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


elemwise_div
0.1586
---
4194.3042
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


elemwise_div
0.0305
---
200.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


elemwise_div
0.0248
---
40.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


elemwise_mul
0.0938
---
4194.3042
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


elemwise_mul
0.0316
---
200.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


elemwise_mul
0.0236
---
20.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


elemwise_sub
0.0667
---
2097.1521
{'lhs': (1024, 1024), 'rhs': (1024, 1024)}


elemwise_sub
0.0323
---
400.0
{'lhs': (10000, 10), 'rhs': (10000, 10)}


elemwise_sub
0.0228
---
40.0
{'lhs': (10000, 1), 'rhs': (10000, 1)}


erf
0.9298
0.7784
2097.1521
{'data': (1024, 1024)}


erf
0.0365
0.0306
20.0
{'data': (10000, 1)}


erf
0.9764
0.7217
4000.0
{'data': (10000, 100)}


erfinv
3.06
0.6606
4194.3042
{'data': (1024, 1024)}


erfinv
0.0726
0.0397
20.0
{'data': (10000, 1)}


erfinv
2.3107
0.477
4000.0
{'data': (10000, 100)}


exp
0.3763
---
2097.1521
{'data': (1024, 1024)}


exp
0.0351
---
40.0
{'data': (10000, 1)}


exp
0.362
---
2000.0
{'data': (10000, 100)}


expm1
0.9999
0.6668
4194.3042
{'data': (1024, 1024)}


expm1
0.0323
0.0273
20.0
{'data': (10000, 1)}


expm1
1.0027
0.6459
4000.0
{'data': (10000, 100)}


fix
0.3574
---
4194.3042
{'data': (1024, 1024)}


fix
0.0355
---
40.0
{'data': (10000, 1)}


fix
0.3398
---
4000.0
{'data': (10000, 100)}


flatten
0.0807
---
4194.3042
{'data': (1024, 1024)}


flatten
0.0623
---
40.0
{'data': (10000, 1)}


flatten
0.0835
---
4000.0
{'data': (10000, 100)}


flip
0.4665
---
2097.1521
{'data': (1024, 1024), 'axis': 0}


flip
0.0395
---
20.0
{'data': (10000, 1), 'axis': 0}


flip
0.4598
---
4000.0
{'data': (10000, 100), 'axis': 0}


floor
0.2024
---
4194.3042
{'data': (1024, 1024)}


floor
0.0647
---
40.0
{'data': (10000, 1)}


floor
0.2069
---
4000.0
{'data': (10000, 100)}


ftml_update
0.8405
---
2645426.25
{'weight': (1024, 1024), 'grad': (1024, 1024), 'd': (1024, 1024), 'v': (1024, 1024), 'z': (1024, 1024), 'lr': 0.1, 'beta1': 0.1, 'beta2': 0.1, 'epsilon': 1e-08, 't': 1, 'rescale_grad': 0.4, 'clip_grad': -1.0}


ftml_update
0.0474
---
2643369.25
{'weight': (10000, 1), 'grad': (10000, 1), 'd': (10000, 1), 'v': (10000, 1), 'z': (10000, 1), 'lr': 0.5, 'beta1': 0.5, 'beta2': 0.5, 'epsilon': 1e-08, 't': 1, 'rescale_grad': 0.4, 'clip_grad': -1.0}


ftml_update
0.8968
---
2647349.25
{'weight': (10000, 100), 'grad': (10000, 100), 'd': (10000, 100), 'v': (10000, 100), 'z': (10000, 100), 'lr': 0.9, 'beta1': 0.9, 'beta2': 0.9, 'epsilon': 1e-08, 't': 1, 'rescale_grad': 0.4, 'clip_grad': -1.0}


ftrl_update
1.002
---
2651640.5
{'weight': (1024, 1024), 'grad': (1024, 1024), 'z': (1024, 1024), 'n': (1024, 1024), 'lr': 0.1, 'rescale_grad': 0.4}


ftrl_update
0.0456
---
2649583.5
{'weight': (10000, 1), 'grad': (10000, 1), 'z': (10000, 1), 'n': (10000, 1), 'lr': 0.5, 'rescale_grad': 0.4}


ftrl_update
1.1469
---
2653563.5
{'weight': (10000, 100), 'grad': (10000, 100), 'z': (10000, 100), 'n': (10000, 100), 'lr': 0.9, 'rescale_grad': 0.4}


gamma
2.9704
5.2562
4194.3042
{'data': (1024, 1024)}


gamma
0.0655
0.095
20.0
{'data': (10000, 1)}


gamma
3.7014
6.6354
4000.0
{'data': (10000, 100)}


gammaln
29.6384
2.4416
4194.3042
{'data': (1024, 1024)}


gammaln
0.3629
0.0522
20.0
{'data': (10000, 1)}


gammaln
27.7632
3.2959
4000.0
{'data': (10000, 100)}


hard_sigmoid
0.0984
0.0984
2097.1521
{'data': (1024, 1024), 'alpha': 0.25, 'beta': 0.5}


hard_sigmoid
0.0381
0.0331
20.0
{'data': (10000, 1), 'alpha': 0.25, 'beta': 0.5}


hard_sigmoid
0.0932
0.0944
2000.0
{'data': (10000, 100), 'alpha': 0.25, 'beta': 0.5}


identity
---
---
4194.3042
{'data': (1024, 1024)}


identity
---
---
40.0
{'data': (10000, 1)}


identity
---
---
4000.0
{'data': (10000, 100)}


log
0.8534
0.0808
2097.1521
{'data': (1024, 1024)}


log
0.0376
0.0299
20.0
{'data': (10000, 1)}


log
0.8084
0.0775
2000.0
{'data': (10000, 100)}


log10
0.9989
0.0825
2097.1521
{'data': (1024, 1024)}


log10
0.0517
0.052
20.0
{'data': (10000, 1)}


log10
0.7217
0.0465
2000.0
{'data': (10000, 100)}


log1p
1.0008
0.085
2097.1521
{'data': (1024, 1024)}


log1p
0.0392
0.0333
20.0
{'data': (10000, 1)}


log1p
0.954
0.0813
2000.0
{'data': (10000, 100)}


log2
0.8624
0.0797
2097.1521
{'data': (1024, 1024)}


log2
0.0449
0.0318
20.0
{'data': (10000, 1)}


log2
0.9151
0.0703
4000.0
{'data': (10000, 100)}


log_softmax
1.6722
0.8149
2097.1521
{'data': (1024, 1024), 'axis': -1, 'temperature': 0.5}


log_softmax
0.042
0.036
20.0
{'data': (10000, 1), 'axis': -1, 'temperature': 0.5}


log_softmax
1.1919
0.5942
2000.0
{'data': (10000, 100), 'axis': -1, 'temperature': 0.5}


logical_not
0.0574
---
2097.1521
{'data': (1024, 1024)}


logical_not
0.0302
---
20.0
{'data': (10000, 1)}


logical_not
0.054
---
2000.0
{'data': (10000, 100)}


make_loss
0.3905
---
2097.1521
{'data': (1024, 1024)}


make_loss
0.0192
---
20.0
{'data': (10000, 1)}


make_loss
0.3912
---
4000.0
{'data': (10000, 100)}


max
0.7253
0.8652
2.048
{'data': (1024, 1024), 'axis': 0}


max
0.1004
0.0333
0.002
{'data': (10000, 1), 'axis': 0}


max
0.5944
0.8481
0.2
{'data': (10000, 100), 'axis': 0}


max_axis
0.7089
---
2.048
{'data': (1024, 1024), 'axis': 0}


max_axis
0.1393
---
0.004
{'data': (10000, 1), 'axis': 0}


max_axis
0.6674
---
0.2
{'data': (10000, 100), 'axis': 0}


mean
0.6215
1.1323
4.096
{'data': (1024, 1024), 'axis': 0}


mean
0.1909
0.0532
0.002
{'data': (10000, 1), 'axis': 0}


mean
0.9125
1.4182
0.4
{'data': (10000, 100), 'axis': 0}


min
0.4774
0.4824
2.048
{'data': (1024, 1024), 'axis': 0}


min
0.1494
0.0412
0.002
{'data': (10000, 1), 'axis': 0}


min
0.5034
0.4568
0.2
{'data': (10000, 100), 'axis': 0}


min_axis
0.7033
---
2.048
{'data': (1024, 1024), 'axis': 0}


min_axis
0.1374
---
0.002
{'data': (10000, 1), 'axis': 0}


min_axis
0.6716
---
0.4
{'data': (10000, 100), 'axis': 0}


mp_sgd_mom_update
0.2236
---
2657855.0
{'weight': (1024, 1024), 'grad': (1024, 1024), 'mom': (1024, 1024), 'weight32': (1024, 1024), 'lr': 0.1, 'rescale_grad': 0.4, 'lazy_update': 0}


mp_sgd_mom_update
0.0371
---
2655797.75
{'weight': (10000, 1), 'grad': (10000, 1), 'mom': (10000, 1), 'weight32': (10000, 1), 'lr': 0.5, 'rescale_grad': 0.4, 'lazy_update': 0}


mp_sgd_mom_update
0.1963
---
2659777.75
{'weight': (10000, 100), 'grad': (10000, 100), 'mom': (10000, 100), 'weight32': (10000, 100), 'lr': 0.9, 'rescale_grad': 0.4, 'lazy_update': 0}


mp_sgd_update
0.1748
---
2664069.25
{'weight': (1024, 1024), 'grad': (1024, 1024), 'weight32': (1024, 1024), 'lr': 0.1, 'rescale_grad': 0.4, 'lazy_update': 0}


mp_sgd_update
0.0315
---
2662012.0
{'weight': (10000, 1), 'grad': (10000, 1), 'weight32': (10000, 1), 'lr': 0.5, 'rescale_grad': 0.4, 'lazy_update': 0}


mp_sgd_update
0.1708
---
2665992.0
{'weight': (10000, 100), 'grad': (10000, 100), 'weight32': (10000, 100), 'lr': 0.9, 'rescale_grad': 0.4, 'lazy_update': 0}


nanprod
0.7865
0.7833
2.048
{'data': (1024, 1024), 'axis': 0}


nanprod
0.0939
0.0319
0.002
{'data': (10000, 1), 'axis': 0}


nanprod
0.586
0.765
0.4
{'data': (10000, 100), 'axis': 0}


nansum
0.7961
0.7197
2.048
{'data': (1024, 1024), 'axis': 0}


nansum
0.1528
0.0357
0.002
{'data': (10000, 1), 'axis': 0}


nansum
0.7476
0.697
0.4
{'data': (10000, 100), 'axis': 0}


negative
0.0515
---
2097.1521
{'data': (1024, 1024)}


negative
0.026
---
40.0
{'data': (10000, 1)}


negative
0.0528
---
4000.0
{'data': (10000, 100)}


ones_like
0.0416
---
4194.3042
{'data': (1024, 1024)}


ones_like
0.0228
---
40.0
{'data': (10000, 1)}


ones_like
0.0423
---
4000.0
{'data': (10000, 100)}


prod
0.7739
0.7642
2.048
{'data': (1024, 1024), 'axis': 0}


prod
0.1388
0.0387
0.002
{'data': (10000, 1), 'axis': 0}


prod
0.6413
0.6501
0.2
{'data': (10000, 100), 'axis': 0}


radians
0.0717
0.0498
2097.1521
{'data': (1024, 1024)}


radians
0.0275
0.0261
40.0
{'data': (10000, 1)}


radians
0.0816
0.0475
4000.0
{'data': (10000, 100)}


random_exponential
2.1319
---
4194.3042
{'shape': (1024, 1024)}


random_exponential
0.0516
---
20.0
{'shape': (10000, 1)}


random_exponential
2.1279
---
2000.0
{'shape': (10000, 100)}


random_gamma
7.0137
---
4194.3042
{'shape': (1024, 1024)}


random_gamma
0.1017
---
20.0
{'shape': (10000, 1)}


random_gamma
6.6949
---
4000.0
{'shape': (10000, 100)}


random_generalized_negative_binomial
9.5154
---
4194.3042
{'shape': (1024, 1024)}


random_generalized_negative_binomial
0.1137
---
20.0
{'shape': (10000, 1)}


random_generalized_negative_binomial
6.0115
---
2000.0
{'shape': (10000, 100)}


random_negative_binomial
5.0877
---
2097.1521
{'k': 1, 'p': 1, 'shape': (1024, 1024)}


random_negative_binomial
0.0879
---
20.0
{'k': 1, 'p': 1, 'shape': (10000, 1)}


random_negative_binomial
4.8165
---
2000.0
{'k': 1, 'p': 1, 'shape': (10000, 100)}


random_normal
1.593
---
2097.1521
{'shape': (1024, 1024)}


random_normal
0.0506
---
20.0
{'shape': (10000, 1)}


random_normal
1.9039
---
2000.0
{'shape': (10000, 100)}


random_pdf_dirichlet
0.0371
---
0.002
{'sample': (2,), 'alpha': [0.0, 2.5]}


random_pdf_exponential
0.0345
---
0.004
{'sample': (2,), 'lam': [1.0, 8.5]}


random_pdf_gamma
0.037
---
0.004
{'sample': (2,), 'alpha': [0.0, 2.5], 'beta': [1.0, 0.7]}


random_pdf_generalized_negative_binomial
0.0364
---
0.004
{'sample': (2,), 'mu': [2.0, 2.5], 'alpha': [0.0, 2.5]}


random_pdf_negative_binomial
0.038
---
0.008
{'sample': (2,), 'k': [20, 49], 'p': [0.4, 0.77]}


random_pdf_normal
0.0532
---
0.008
{'sample': (2,), 'mu': [2.0, 2.5], 'sigma': [1.0, 3.7]}


random_pdf_poisson
0.0355
---
0.004
{'sample': (2,), 'lam': [1.0, 8.5]}


random_pdf_uniform
0.0296
---
0.008
{'sample': (2,), 'low': [0.0, 2.5], 'high': [1.0, 3.7]}


random_poisson
1.9602
---
2097.1521
{'shape': (1024, 1024)}


random_poisson
0.0493
---
20.0
{'shape': (10000, 1)}


random_poisson
1.6006
---
4000.0
{'shape': (10000, 100)}


random_randint
1.0988
---
4194.3042
{'low': 0, 'high': 5, 'shape': (1024, 1024)}


random_randint
0.05
---
20.0
{'low': 0, 'high': 5, 'shape': (10000, 1)}


random_randint
1.3119
---
4000.0
{'low': 0, 'high': 5, 'shape': (10000, 100)}


random_uniform
0.5839
---
2097.1521
{'low': 0, 'high': 5, 'shape': (1024, 1024)}


random_uniform
0.0374
---
20.0
{'low': 0, 'high': 5, 'shape': (10000, 1)}


random_uniform
0.5641
---
4000.0
{'low': 0, 'high': 5, 'shape': (10000, 100)}


rcbrt
1.3523
1.4009
4194.3042
{'data': (1024, 1024)}


rcbrt
0.0432
0.0394
20.0
{'data': (10000, 1)}


rcbrt
1.2844
1.3316
2000.0
{'data': (10000, 100)}


reciprocal
0.0811
0.0619
2097.1521
{'data': (1024, 1024)}


reciprocal
0.0298
0.0327
20.0
{'data': (10000, 1)}


reciprocal
0.0735
0.0563
2000.0
{'data': (10000, 100)}


relu
0.0938
0.0975
4194.3042
{'data': (1024, 1024)}


relu
0.0289
0.0357
20.0
{'data': (10000, 1)}


relu
0.0781
0.0827
4000.0
{'data': (10000, 100)}


rint
0.3149
---
2097.1521
{'data': (1024, 1024)}


rint
0.0384
---
20.0
{'data': (10000, 1)}


rint
0.1954
---
2000.0
{'data': (10000, 100)}


rmsprop_update
0.4186
---
2670283.5
{'weight': (1024, 1024), 'grad': (1024, 1024), 'n': (1024, 1024), 'lr': 0.1, 'gamma1': 0.1, 'epsilon': 1e-08, 'rescale_grad': 0.4}


rmsprop_update
0.0387
---
2668226.25
{'weight': (10000, 1), 'grad': (10000, 1), 'n': (10000, 1), 'lr': 0.5, 'gamma1': 0.5, 'epsilon': 1e-08, 'rescale_grad': 0.4}


rmsprop_update
0.5333
---
2672206.25
{'weight': (10000, 100), 'grad': (10000, 100), 'n': (10000, 100), 'lr': 0.9, 'gamma1': 0.9, 'epsilon': 1e-08, 'rescale_grad': 0.4}


rmspropalex_update
0.642
---
2676497.75
{'weight': (1024, 1024), 'grad': (1024, 1024), 'n': (1024, 1024), 'g': (1024, 1024), 'delta': (1024, 1024), 'lr': 0.1, 'gamma1': 0.1, 'gamma2': 0.1, 'epsilon': 1e-08, 'rescale_grad': 0.4}


rmspropalex_update
0.0488
---
2674440.75
{'weight': (10000, 1), 'grad': (10000, 1), 'n': (10000, 1), 'g': (10000, 1), 'delta': (10000, 1), 'lr': 0.5, 'gamma1': 0.5, 'gamma2': 0.5, 'epsilon': 1e-08, 'rescale_grad': 0.4}


rmspropalex_update
1.1672
---
2678420.75
{'weight': (10000, 100), 'grad': (10000, 100), 'n': (10000, 100), 'g': (10000, 100), 'delta': (10000, 100), 'lr': 0.9, 'gamma1': 0.9, 'gamma2': 0.9, 'epsilon': 1e-08, 'rescale_grad': 0.4}


round
0.2487
---
2097.1521
{'data': (1024, 1024)}


round
0.0968
---
20.0
{'data': (10000, 1)}


round
0.2385
---
2000.0
{'data': (10000, 100)}


rsqrt
0.6115
0.5608
2097.1521
{'data': (1024, 1024)}


rsqrt
0.0909
0.0355
20.0
{'data': (10000, 1)}


rsqrt
0.6664
0.6492
4000.0
{'data': (10000, 100)}


sample_exponential
4.1897
---
8388.6084
{'lam': [1.0, 8.5], 'shape': (1024, 1024)}


sample_exponential
0.0779
---
40.0
{'lam': [1.0, 8.5], 'shape': (10000, 1)}


sample_exponential
4.2833
---
8000.0
{'lam': [1.0, 8.5], 'shape': (10000, 100)}


sample_gamma
16.064
---
8388.6084
{'alpha': [0.0, 2.5], 'shape': (1024, 1024), 'beta': [1.0, 0.7]}


sample_gamma
0.1433
---
40.0
{'alpha': [0.0, 2.5], 'shape': (10000, 1), 'beta': [1.0, 0.7]}


sample_gamma
9.8149
---
4000.0
{'alpha': [0.0, 2.5], 'shape': (10000, 100), 'beta': [1.0, 0.7]}


sample_generalized_negative_binomial
30.0658
---
8388.6084
{'mu': [2.0, 2.5], 'shape': (1024, 1024), 'alpha': [0.0, 2.5]}


sample_generalized_negative_binomial
0.3096
---
40.0
{'mu': [2.0, 2.5], 'shape': (10000, 1), 'alpha': [0.0, 2.5]}


sample_generalized_negative_binomial
28.0669
---
8000.0
{'mu': [2.0, 2.5], 'shape': (10000, 100), 'alpha': [0.0, 2.5]}


sample_negative_binomial
347.9748
---
4194.3042
{'k': [20, 49], 'shape': (1024, 1024), 'p': [0.4, 0.77]}


sample_negative_binomial
3.098
---
80.0
{'k': [20, 49], 'shape': (10000, 1), 'p': [0.4, 0.77]}


sample_negative_binomial
306.4638
---
4000.0
{'k': [20, 49], 'shape': (10000, 100), 'p': [0.4, 0.77]}


sample_normal
4.9098
---
8388.6084
{'mu': [2.0, 2.5], 'shape': (1024, 1024), 'sigma': [1.0, 3.7]}


sample_normal
0.0861
---
40.0
{'mu': [2.0, 2.5], 'shape': (10000, 1), 'sigma': [1.0, 3.7]}


sample_normal
4.7348
---
8000.0
{'mu': [2.0, 2.5], 'shape': (10000, 100), 'sigma': [1.0, 3.7]}


sample_poisson
10.097
---
4194.3042
{'lam': [1.0, 8.5], 'shape': (1024, 1024)}


sample_poisson
0.1211
---
40.0
{'lam': [1.0, 8.5], 'shape': (10000, 1)}


sample_poisson
9.5847
---
4000.0
{'lam': [1.0, 8.5], 'shape': (10000, 100)}


sample_uniform
1.1467
---
4194.3042
{'low': [0.0, 2.5], 'shape': (1024, 1024), 'high': [1.0, 3.7]}


sample_uniform
0.0384
---
40.0
{'low': [0.0, 2.5], 'shape': (10000, 1), 'high': [1.0, 3.7]}


sample_uniform
1.1347
---
8000.0
{'low': [0.0, 2.5], 'shape': (10000, 100), 'high': [1.0, 3.7]}


sgd_mom_update
0.2134
---
2682712.0
{'weight': (1024, 1024), 'grad': (1024, 1024), 'mom': (1024, 1024), 'lr': 0.1, 'rescale_grad': 0.4, 'lazy_update': 0}


sgd_mom_update
0.0419
---
2680655.0
{'weight': (10000, 1), 'grad': (10000, 1), 'mom': (10000, 1), 'lr': 0.5, 'rescale_grad': 0.4, 'lazy_update': 0}


sgd_mom_update
0.2044
---
2684635.0
{'weight': (10000, 100), 'grad': (10000, 100), 'mom': (10000, 100), 'lr': 0.9, 'rescale_grad': 0.4, 'lazy_update': 0}


sgd_update
0.1791
---
2688926.5
{'weight': (1024, 1024), 'grad': (1024, 1024), 'lr': 0.1, 'rescale_grad': 0.4, 'lazy_update': 0}


sgd_update
0.0367
---
2686869.25
{'weight': (10000, 1), 'grad': (10000, 1), 'lr': 0.5, 'rescale_grad': 0.4, 'lazy_update': 0}


sgd_update
0.1686
---
2690849.25
{'weight': (10000, 100), 'grad': (10000, 100), 'lr': 0.9, 'rescale_grad': 0.4, 'lazy_update': 0}


shape_array
0.0165
---
0.016
{'data': (1024, 1024)}


shape_array
0.0166
---
0.016
{'data': (10000, 1)}


shape_array
0.0161
---
0.016
{'data': (10000, 100)}


shuffle
1.1006
---
4194.3042
{'data': (1024, 1024)}


shuffle
0.4032
---
20.0
{'data': (10000, 1)}


shuffle
1.2646
---
2000.0
{'data': (10000, 100)}


sigmoid
0.8481
0.0758
2097.1521
{'data': (1024, 1024)}


sigmoid
0.0337
0.0273
20.0
{'data': (10000, 1)}


sigmoid
0.7093
0.0667
2000.0
{'data': (10000, 100)}


sign
0.2681
0.0484
2097.1521
{'data': (1024, 1024)}


sign
0.0687
0.022
20.0
{'data': (10000, 1)}


sign
0.2627
0.0448
2000.0
{'data': (10000, 100)}


signsgd_update
0.2254
---
2695140.75
{'weight': (1024, 1024), 'grad': (1024, 1024), 'lr': 0.1, 'rescale_grad': 0.4}


signsgd_update
0.0339
---
2693083.5
{'weight': (10000, 1), 'grad': (10000, 1), 'lr': 0.5, 'rescale_grad': 0.4}


signsgd_update
0.2173
---
2697063.5
{'weight': (10000, 100), 'grad': (10000, 100), 'lr': 0.9, 'rescale_grad': 0.4}


signum_update
0.3357
---
2701355.0
{'weight': (1024, 1024), 'grad': (1024, 1024), 'mom': (1024, 1024), 'lr': 0.1, 'rescale_grad': 0.4}


signum_update
0.0377
---
2699297.75
{'weight': (10000, 1), 'grad': (10000, 1), 'mom': (10000, 1), 'lr': 0.5, 'rescale_grad': 0.4}


signum_update
0.3269
---
2703277.75
{'weight': (10000, 100), 'grad': (10000, 100), 'mom': (10000, 100), 'lr': 0.9, 'rescale_grad': 0.4}


sin
0.5448
0.5685
2097.1521
{'data': (1024, 1024)}


sin
0.0298
0.0275
20.0
{'data': (10000, 1)}


sin
0.4843
0.4851
2000.0
{'data': (10000, 100)}


sinh
1.5249
0.9406
2097.1521
{'data': (1024, 1024)}


sinh
0.0452
0.0384
20.0
{'data': (10000, 1)}


sinh
1.4619
0.8909
4000.0
{'data': (10000, 100)}


size_array
0.014
---
0.008
{'data': (1024, 1024)}


size_array
0.0139
---
0.008
{'data': (10000, 1)}


size_array
0.0139
---
0.008
{'data': (10000, 100)}


softmax
1.3116
0.1007
2097.1521
{'data': (1024, 1024), 'axis': -1, 'temperature': 0.5}


softmax
0.0492
0.036
20.0
{'data': (10000, 1), 'axis': -1, 'temperature': 0.5}


softmax
1.1855
0.1003
2000.0
{'data': (10000, 100), 'axis': -1, 'temperature': 0.5}


softsign
0.0783
0.0609
2097.1521
{'data': (1024, 1024)}


softsign
0.0324
0.036
20.0
{'data': (10000, 1)}


softsign
0.076
0.059
2000.0
{'data': (10000, 100)}


sort
25.079
---
6291.4561
{'data': (1024, 1024), 'axis': 0}


sort
1.0047
---
40.0
{'data': (10000, 1), 'axis': 0}


sort
26.6501
---
6000.0
{'data': (10000, 100), 'axis': 0}


space_to_depth
0.0493
---
0.064
{'data': (1, 4, 2, 4), 'block_size': 2}


space_to_depth
0.4216
---
500.0
{'data': (10, 25, 10, 100), 'block_size': 5}


sqrt
0.9161
0.2843
2097.1521
{'data': (1024, 1024)}


sqrt
0.0399
0.0316
20.0
{'data': (10000, 1)}


sqrt
0.7284
0.1733
2000.0
{'data': (10000, 100)}


square
0.0698
0.0616
2097.1521
{'data': (1024, 1024)}


square
0.029
0.0288
40.0
{'data': (10000, 1)}


square
0.0669
0.0636
4000.0
{'data': (10000, 100)}


stop_gradient
---
---
4194.3042
{'data': (1024, 1024)}


stop_gradient
---
---
20.0
{'data': (10000, 1)}


stop_gradient
---
---
2000.0
{'data': (10000, 100)}


sum
0.8737
0.698
4.096
{'data': (1024, 1024), 'axis': 0}


sum
0.1644
0.0335
0.002
{'data': (10000, 1), 'axis': 0}


sum
0.7912
0.6712
0.2
{'data': (10000, 100), 'axis': 0}


sum_axis
0.7821
---
2.048
{'data': (1024, 1024), 'axis': 0}


sum_axis
0.1451
---
0.002
{'data': (10000, 1), 'axis': 0}


sum_axis
0.7964
---
0.4
{'data': (10000, 100), 'axis': 0}


swapaxes
1.4262
---
2097.1521
{'data': (1024, 1024), 'dim1': 0, 'dim2': 1}


swapaxes
0.0452
---
20.0
{'data': (10000, 1), 'dim1': 0, 'dim2': 1}


swapaxes
1.3266
---
4000.0
{'data': (10000, 100), 'dim1': 0, 'dim2': 1}


tan
1.1632
0.081
2097.1521
{'data': (1024, 1024)}


tan
0.0372
0.0256
20.0
{'data': (10000, 1)}


tan
1.146
0.0842
2000.0
{'data': (10000, 100)}


tanh
1.4511
0.082
2097.1521
{'data': (1024, 1024)}


tanh
0.041
0.0263
40.0
{'data': (10000, 1)}


tanh
1.3866
0.079
4000.0
{'data': (10000, 100)}


topk
29.8296
---
4.096
{'data': (1024, 1024), 'axis': 0, 'k': 1}


topk
0.2789
---
0.002
{'data': (10000, 1), 'axis': 0, 'k': 1}


topk
24.2809
---
0.4
{'data': (10000, 100), 'axis': 0, 'k': 1}


transpose
0.0961
---
4194.3042
{'data': (1024, 1024)}


transpose
0.0454
---
40.0
{'data': (10000, 1)}


transpose
0.2221
---
4000.0
{'data': (10000, 100)}


trunc
0.2947
---
4194.3042
{'data': (1024, 1024)}


trunc
0.0851
---
40.0
{'data': (10000, 1)}


trunc
0.2885
---
4000.0
{'data': (10000, 100)}


zeros_like
0.042
---
4194.3042
{'data': (1024, 1024)}


zeros_like
0.0189
---
40.0
{'data': (10000, 1)}


zeros_like
0.0454
---
4000.0
{'data': (10000, 100)}