A simple tool for parsing the profile.json file of mxnet
export MXNET_EXEC_BULK_EXEC_INFERENCE=0
export MXNET_EXEC_BULK_EXEC_TRAIN=0
export MXNET_PROFILER_AUTOSTART=1
$ python mxProfileParser.py --file demo.json
Time of each OP:
_copy 2855.707 ms 27.1972095238 ms/call 105 calls
Convolution 29936.515 ms 3.03308156028 ms/call 9870 calls
Pooling 4265.345 ms 2.9015952381 ms/call 1470 calls
SoftmaxOutput 8.093 ms 0.0770761904762 ms/call 105 calls
Activation 6724.473 ms 0.681304255319 ms/call 9870 calls
BatchNorm 10322.554 ms 1.0458514691 ms/call 9870 calls
FullyConnected 253.268 ms 2.41207619048 ms/call 105 calls
Concat 3792.917 ms 3.28391082251 ms/call 1155 calls
Flatten 36.932 ms 0.351733333333 ms/call 105 calls
Total OP Time: 58195.80400000 ms
$ python mxProfileParser.py --file demo.json --op BatchNorm
Time of BatchNorm:
BatchNorm 10322.554 ms 1.0458514691 ms/call 9870 calls
$ python mxProfileParser.py --file demo.json --iterations 105
Time of each OP:
_copy 2855.707 ms 27.1972095238 ms/call 105 calls 1 calls/iter
Convolution 29936.515 ms 3.03308156028 ms/call 9870 calls 94 calls/iter
Pooling 4265.345 ms 2.9015952381 ms/call 1470 calls 14 calls/iter
SoftmaxOutput 8.093 ms 0.0770761904762 ms/call 105 calls 1 calls/iter
Activation 6724.473 ms 0.681304255319 ms/call 9870 calls 94 calls/iter
BatchNorm 10322.554 ms 1.0458514691 ms/call 9870 calls 94 calls/iter
FullyConnected 253.268 ms 2.41207619048 ms/call 105 calls 1 calls/iter
Concat 3792.917 ms 3.28391082251 ms/call 1155 calls 11 calls/iter
Flatten 36.932 ms 0.351733333333 ms/call 105 calls 1 calls/iter
Total OP Time: 58195.80400000 ms
Iteration Time: 554.24575238 ms
- First column are operator names. Second column are the total execution time of each operator. Third column are the average execution time of each operator. Third column are the number of calls of each operator. The last column are the number of calls in one iteration.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent