[GitHub] [incubator-mxnet] RuRo commented on issue #18090: Deadlock with ThreadedEngine

[GitBox <gi...@apache.org>]

RuRo commented on issue #18090:
URL: https://github.com/apache/incubator-mxnet/issues/18090#issuecomment-617413655


   > @RuRo, @reminisce helped point out an error in your reproducer script. You use zero-shape tensor, but didn't enable numpy shape semantics. It should be
   
   Ah, I see. I had the `npx.set_np()` call, when I started reproducing this issue, but I removed it, since it made no difference for the hanging behaviour. I was under the impression, that scalar arrays are enabled by default for the `mx.np` kind of array. My bad.
   
   Also, the hang happens with or without the `mx.nd.waitall()` call. Also, as previously mentioned, you don't need the GPU context. So the short version is
   ```python
   import mxnet as mx
   mx.npx.set_np()
   while True:
       for _ in range(100):
           mx_data = mx.np.array(0)
           mx_x = mx.np.empty_like(mx_data)
       print(end='.', flush=True)
   ```


----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org