connorgoggins opened a new issue #17412: certain sample_ ops hanging with mx.nd.random_normal input URL: https://github.com/apache/incubator-mxnet/issues/17412 ## Description While testing sample_ operators for opperf, I came across an issue where `sample_gamma`, `sample_generalized_negative_binomial`, and `sample_generalized_negative_binomial` would hang when provided with `mx.nd.random_normal` as input args. For example, the following command hangs: `mx.nd.sample_gamma(alpha=mx.nd.random_normal(shape=(3,)),beta=mx.nd.random_normal(shape=(3,)))` Interestingly, when the `mx.nd.random_normal` call is made and the output value is assigned to a variable before that variable is used in the operator call, the call to the operator does not hang. Also, this bug appears to be specific to `random_normal`. `random_uniform` works properly in the same context. ### Error Message No error, but hangs and the process maximizes its computational resources. ### Steps to reproduce (Paste the commands you ran that produced the error.) 1. `python3` (open a Python shell) 2. `import mxnet as mx` 3. `mx.nd.sample_gamma(alpha=mx.nd.random_normal(shape=(3,)),beta=mx.nd.random_normal(shape=(3,)))` Sometimes, it takes multiple calls of Step 3 to trigger the bug. Just run the command from Step 3 repeatedly until the bug appears (shouldn't take more than 2-3 runs). ## Environment I found this bug on Mac and Ubuntu builds of MXNet, with large tensor disabled and enabled (respectively).
---------------------------------------------------------------- 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: us...@infra.apache.org With regards, Apache Git Services
