Stopping the profiler is very expensive

ezyang / pytorch-unattached

Tensors and Dynamic neural networks in Python with strong GPU acceleration

http://pytorch.org

Other

20 stars 8 forks source link

Closed ezyang closed 7 years ago

ezyang commented 7 years ago

Profiler processing is done in Python and this is too slow for real world profiling. @zdevito triggered this case. We need to optimize it somehow.

apaszke commented 7 years ago

Solved by pytorch/pytorch#2942