scipy: Failures on 0.17.x with MKL

Time for this release candidate’s set of MKL failures. Do other people test using MKL, or should I try to make a habit of doing this check before each release?

icc --version is icc (ICC) 16.0.1 20151021

numpy on maintenance/1.10.x branch, version 1.10.0.dev0+075cc98, numpy.test('full') passing; scipy.test('full') gives:

>>> scipy.test('full')
Running unit tests for scipy
NumPy version 1.10.0.dev0+075cc98
NumPy relaxed strides checking option: False
NumPy is installed in /home/larsoner/.local/lib/python2.7/site-packages/numpy
SciPy version 0.17.0rc1
SciPy is installed in /home/larsoner/.local/lib/python2.7/site-packages/scipy
Python version 2.7.10 (default, Oct 14 2015, 16:09:02) [GCC 5.2.1 20151010]
nose version 1.3.6

...

======================================================================
FAIL: test_lorentz (test_odr.TestODR)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/larsoner/.local/lib/python2.7/site-packages/scipy/odr/tests/test_odr.py", line 293, in test_lorentz
    3.7798193600109009e+00]),
  File "/home/larsoner/.local/lib/python2.7/site-packages/numpy/testing/utils.py", line 892, in assert_array_almost_equal
    precision=decimal)
  File "/home/larsoner/.local/lib/python2.7/site-packages/numpy/testing/utils.py", line 713, in assert_array_compare
    raise AssertionError(msg)
AssertionError: 
Arrays are not almost equal to 6 decimals

(mismatch 100.0%)
 x: array([  1.000000e+03,   1.000000e-01,   3.800000e+00])
 y: array([  1.430678e+03,   1.339051e-01,   3.779819e+00])

======================================================================
FAIL: test_multi (test_odr.TestODR)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/larsoner/.local/lib/python2.7/site-packages/scipy/odr/tests/test_odr.py", line 190, in test_multi
    0.5101147161764654, 0.5173902330489161]),
  File "/home/larsoner/.local/lib/python2.7/site-packages/numpy/testing/utils.py", line 892, in assert_array_almost_equal
    precision=decimal)
  File "/home/larsoner/.local/lib/python2.7/site-packages/numpy/testing/utils.py", line 713, in assert_array_compare
    raise AssertionError(msg)
AssertionError: 
Arrays are not almost equal to 6 decimals

(mismatch 100.0%)
 x: array([ 4. ,  2. ,  7. ,  0.4,  0.5])
 y: array([ 4.379988,  2.433306,  8.002885,  0.510115,  0.51739 ])

======================================================================
FAIL: test_pearson (test_odr.TestODR)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/larsoner/.local/lib/python2.7/site-packages/scipy/odr/tests/test_odr.py", line 236, in test_pearson
    np.array([5.4767400299231674, -0.4796082367610305]),
  File "/home/larsoner/.local/lib/python2.7/site-packages/numpy/testing/utils.py", line 892, in assert_array_almost_equal
    precision=decimal)
  File "/home/larsoner/.local/lib/python2.7/site-packages/numpy/testing/utils.py", line 713, in assert_array_compare
    raise AssertionError(msg)
AssertionError: 
Arrays are not almost equal to 6 decimals

(mismatch 100.0%)
 x: array([ 1.,  1.])
 y: array([ 5.47674 , -0.479608])

----------------------------------------------------------------------
Ran 21565 tests in 1190.894s

FAILED (KNOWNFAIL=130, SKIP=1811, failures=3)
<nose.result.TextTestResult run=21565 errors=0 failures=3>
>>>

I can also try with latest numpy master, but I figured we should try to clear up the 1.10 failures to see if that gets everything working first.

About this issue

Original URL
State: closed
Created 9 years ago
Reactions: 1
Comments: 20 (20 by maintainers)

Links to this issue

from the performance point of - Intel Community

Most upvoted comments

I’d also try to downgrade numpy to 1.6.2 or 1.7.2: it could be due to some recent tightening in numpy.

It’s too easy for me to say “I would” when I’m not doing the work, though. So I’d rather say thank you for testing it! 24.12.2015 0:52 пользователь “Eric Larson” notifications@github.com написал:

Could be it’s broken on the latest Intel MKL compiler, then, since that’s also been updated by me since the last check. I’ll give 0.16.x a try to see if it fails similarly, since I know that did pass last time I ran it (which was with the previous MKL release).

— Reply to this email directly or view it on GitHub https://github.com/scipy/scipy/issues/5621#issuecomment-166998064.

ev-br on Dec 23, 2015

ordqz passes, yay! 😃

should I try to make a habit of doing this check before each release?

That’d be great. A monthly cron job?

ev-br on Dec 23, 2015