[Unit Test]test_numpy_op.py:test_np_mixed_precision_binary_funcs randomly failed due to assertion error

[heaseny]

Description

test_numpy_op.py:test_np_mixed_precision_binary_funcs randomly failed due to assertion error, and can be reproduced by test seed.
Test command:
cd tests/python/unittest/
MXNET_TEST_SEED=445183722 nosetests -v -s test_numpy_op.py:test_np_mixed_precision_binary_funcs

Error Message

test_numpy_op.test_np_mixed_precision_binary_funcs ... [INFO] Setting test np/mx/python random seeds, use MXNET_TEST_SEED=445183722 to reproduce.

*** Maximum errors for vector of size 6: rtol=0.01, atol=0.001

1: Error 1.470077 Location of error: (1, 1), a=0.02921580, b=0.02734375
FAIL

======================================================================
FAIL: test_numpy_op.test_np_mixed_precision_binary_funcs

Traceback (most recent call last):
File "/home/mxnet/anacondaenv/anaconda3_mxnet/envs/mxnet_daily_py3/lib/python3.6/site-packages/nose/case.py", line 197, in runTest
self.test(*self.arg)
File "/home/mxnet/workspace/source/incubator-mxnet/tests/python/unittest/common.py", line 223, in test_new
orig_test(*args, **kwargs)
File "/home/mxnet/anacondaenv/anaconda3_mxnet/envs/mxnet_daily_py3/lib/python3.6/site-packages/mxnet-2.0.0-py3.6.egg/mxnet/util.py", line 297, in _with_np_shape
return func(*args, **kwargs)
File "/home/mxnet/anacondaenv/anaconda3_mxnet/envs/mxnet_daily_py3/lib/python3.6/site-packages/mxnet-2.0.0-py3.6.egg/mxnet/util.py", line 481, in _with_np_array
return func(*args, **kwargs)
File "/home/mxnet/workspace/source/incubator-mxnet/tests/python/unittest/test_numpy_op.py", line 2655, in test_np_mixed_precision_binary_funcs
check_mixed_precision_binary_func(func, low, high, lshape, rshape, lgrad, rgrad, type1, type2)
File "/home/mxnet/workspace/source/incubator-mxnet/tests/python/unittest/test_numpy_op.py", line 2595, in check_mixed_precision_binary_func
use_broadcast=False, equal_nan=True)
File "/home/mxnet/anacondaenv/anaconda3_mxnet/envs/mxnet_daily_py3/lib/python3.6/site-packages/mxnet-2.0.0-py3.6.egg/mxnet/test_utils.py", line 637, in assert_almost_equal
raise AssertionError(msg)
AssertionError:
Error 1.470077 exceeds tolerance rtol=1.000000e-02, atol=1.000000e-03 (mismatch 16.666667%).
Location of maximum error: (1, 1), a=0.02921580, b=0.02734375
ACTUAL: array([[2.87304688, 1.95996094, 1.35742188],
[3.35351562, 0.0292158 , 3.74609375]])
DESIRED: array([[2.87304688, 1.95996094, 1.35742188],
[3.35351562, 0.02734375, 3.74609375]])
-------------------- >> begin captured logging << --------------------
common: INFO: Setting test np/mx/python random seeds, use MXNET_TEST_SEED=445183722 to reproduce.
root: INFO: NumPy-shape semantics has been activated in your code. This is required for creating and manipulating scalar and zero-size tensors, which were not supported in MXNet before, as in the official NumPy library. Please DO NOT manually deactivate this semantics while using mxnet.numpy and mxnet.numpy_extension modules.
--------------------- >> end captured logging << ---------------------

---

Ran 1 test in 2.560s

FAILED (failures=1)

To Reproduce

Test command:
build with latest commit on master branch and run with the below commands:
cd tests/python/unittest/
MXNET_TEST_SEED=445183722 nosetests -v -s

Steps to reproduce

(Paste the commands you ran that produced the error.)

1. Build with mkl and install mxnet
   make -j USE_MKLDNN=1 USE_OPENCV=1 USE_BLAS=mkl USE_GPERFTOOLS=0 USE_INTEL_PATH=/opt/intel/
2. Run with the below command:
   cd tests/python/unittest/
   MXNET_TEST_SEED=445183722 nosetests -v -s test_numpy_op.py:test_np_mixed_precision_binary_funcs

Environment

Centos7.6
python3.6.8
GCC7.3.1
cmake:3.14.0
CPU

[leezu]

Thanks @heaseny . This is a duplicate of #16848