mbrookhart commented on issue #12502: [NGRAPH] MXNet - nGraph initial 
integration
URL: https://github.com/apache/incubator-mxnet/pull/12502#issuecomment-455385471
 
 
   On an AWS C5.18xlarge instance, running `MXNET_SUBGRAPH_BACKEND="ngraph" 
OMP_NUM_THREADS=18 KMP_BLOCKTIME=1 KMP_AFFINITY=granularity=fine,compact,1,0 
python example/image-classification/benchmark_score.py` we see:
   
   ```
   INFO:root:It may take some time to run all models, set --network to run a 
specific one
   INFO:root:run batchsize [1, 32, 64, 128, 256] by default, set --batch-size 
to run a specific one
   INFO:root:network: alexnet
   INFO:root:device: cpu(0)
   /home/ubuntu/mxnet-upstream/python/mxnet/module/base_module.py:67: 
UserWarning: Data provided by label_shapes don't match names specified by 
label_names ([] vs. ['softmax_label'])
     warnings.warn(msg)
   INFO:root:batch size  1, dtype float32, images/sec: 248.445345
   INFO:root:batch size 32, dtype float32, images/sec: 690.425029
   INFO:root:batch size 64, dtype float32, images/sec: 729.949982
   INFO:root:batch size 128, dtype float32, images/sec: 762.059816
   INFO:root:batch size 256, dtype float32, images/sec: 775.141168
   INFO:root:network: vgg-16
   INFO:root:device: cpu(0)
   INFO:root:batch size  1, dtype float32, images/sec: 53.914065
   INFO:root:batch size 32, dtype float32, images/sec: 78.701371
   INFO:root:batch size 64, dtype float32, images/sec: 79.244626
   INFO:root:batch size 128, dtype float32, images/sec: 79.563567
   INFO:root:batch size 256, dtype float32, images/sec: 79.929285
   INFO:root:network: resnetv1-50
   INFO:root:device: cpu(0)
   INFO:root:batch size  1, dtype float32, images/sec: 169.282840
   INFO:root:batch size 32, dtype float32, images/sec: 309.324862
   INFO:root:batch size 64, dtype float32, images/sec: 313.505546
   INFO:root:batch size 128, dtype float32, images/sec: 310.872866
   INFO:root:batch size 256, dtype float32, images/sec: 304.673203
   INFO:root:network: resnet-50
   INFO:root:device: cpu(0)
   INFO:root:batch size  1, dtype float32, images/sec: 129.465415
   INFO:root:batch size 32, dtype float32, images/sec: 228.144305
   INFO:root:batch size 64, dtype float32, images/sec: 230.074768
   INFO:root:batch size 128, dtype float32, images/sec: 229.630909
   INFO:root:batch size 256, dtype float32, images/sec: 227.750606
   INFO:root:network: resnet-152
   INFO:root:device: cpu(0)
   INFO:root:batch size  1, dtype float32, images/sec: 52.347661
   INFO:root:batch size 32, dtype float32, images/sec: 91.885424
   INFO:root:batch size 64, dtype float32, images/sec: 91.905888
   INFO:root:batch size 128, dtype float32, images/sec: 90.593244
   INFO:root:batch size 256, dtype float32, images/sec: 87.036963
   INFO:root:network: inception-bn
   INFO:root:device: cpu(0)
   INFO:root:batch size  1, dtype float32, images/sec: 227.770573
   INFO:root:batch size 32, dtype float32, images/sec: 470.279520
   INFO:root:batch size 64, dtype float32, images/sec: 450.336728
   INFO:root:batch size 128, dtype float32, images/sec: 451.288855
   INFO:root:batch size 256, dtype float32, images/sec: 445.435831
   INFO:root:network: inception-v3
   INFO:root:device: cpu(0)
   INFO:root:batch size  1, dtype float32, images/sec: 111.587438
   INFO:root:batch size 32, dtype float32, images/sec: 176.759147
   INFO:root:batch size 64, dtype float32, images/sec: 178.181525
   INFO:root:batch size 128, dtype float32, images/sec: 177.111329
   INFO:root:batch size 256, dtype float32, images/sec: 175.750856
   INFO:root:network: inception-v4
   INFO:root:device: cpu(0)
   INFO:root:batch size  1, dtype float32, images/sec: 56.575425
   INFO:root:batch size 32, dtype float32, images/sec: 81.981927
   INFO:root:batch size 64, dtype float32, images/sec: 83.634385
   INFO:root:batch size 128, dtype float32, images/sec: 84.176398
   INFO:root:batch size 256, dtype float32, images/sec: 83.734302
   INFO:root:network: inception-resnet-v2
   INFO:root:device: cpu(0)
   INFO:root:batch size  1, dtype float32, images/sec: 67.571594
   INFO:root:batch size 32, dtype float32, images/sec: 160.976709
   INFO:root:batch size 64, dtype float32, images/sec: 162.998386
   INFO:root:batch size 128, dtype float32, images/sec: 160.510979
   INFO:root:batch size 256, dtype float32, images/sec: 157.681065
   INFO:root:network: mobilenet
   INFO:root:device: cpu(0)
   INFO:root:batch size  1, dtype float32, images/sec: 637.248211
   INFO:root:batch size 32, dtype float32, images/sec: 1067.549185
   INFO:root:batch size 64, dtype float32, images/sec: 1069.801105
   INFO:root:batch size 128, dtype float32, images/sec: 1071.474370
   INFO:root:batch size 256, dtype float32, images/sec: 1075.118513
   INFO:root:network: densenet121
   INFO:root:network: densenet121 is converted from gluon modelzoo
   INFO:root:you can run benchmark/python/gluon/benchmark_gluon.py for more 
models
   INFO:root:device: cpu(0)
   INFO:root:batch size  1, dtype float32, images/sec: 122.715461
   INFO:root:batch size 32, dtype float32, images/sec: 197.825720
   INFO:root:batch size 64, dtype float32, images/sec: 199.889640
   INFO:root:batch size 128, dtype float32, images/sec: 202.583578
   INFO:root:batch size 256, dtype float32, images/sec: 197.577774
   INFO:root:network: squeezenet1.1
   INFO:root:network: squeezenet1.1 is converted from gluon modelzoo
   INFO:root:you can run benchmark/python/gluon/benchmark_gluon.py for more 
models
   INFO:root:device: cpu(0)
   INFO:root:batch size  1, dtype float32, images/sec: 1084.528107
   INFO:root:batch size 32, dtype float32, images/sec: 1271.417794
   INFO:root:batch size 64, dtype float32, images/sec: 1257.443801
   INFO:root:batch size 128, dtype float32, images/sec: 1272.551532
   INFO:root:batch size 256, dtype float32, images/sec: 1258.965189
   ```

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
[email protected]


With regards,
Apache Git Services

Reply via email to