Linux64bit Numpy 1.6 test_nonsero_byteswap & test failed

Linux64bit Numpy 1.6 test_nonsero_byteswap & test failed

Hi,
I am using MKL version 10.3 and Python 2.7.2 and compiled numpy1.6.1 using custom library method.
I ran numpy.test('full') and here is the output.

compiler options used "icc -O2 -g -fPIC -fp-model strict -openmp -xhost "

Couple of tests failed. Has anyone noticed similar problems. Is there a fix for these. These errors don't appear when numpy is compiled with gcc.

test_nonzero_bytesswap
======================================================================
FAIL: test_nonzero_byteswap (test_regression.TestRegression)
----------------------------------------------------------------------
Traceback (most recent call last):
File "/home/user/python27-install/lib/python2.7/site-packages/numpy/core/tests/test_regression.py", line 1450, in test_nonzero_byteswap
assert_equal(a.nonzero()[0], [1])
File "/home/user/python27-install/lib/python2.7/site-packages/numpy/testing/utils.py", line 256, in assert_equal
return assert_array_equal(actual, desired, err_msg, verbose)
File "/home/user/python27-install/lib/python2.7/site-packages/numpy/testing/utils.py", line 707, in assert_array_equal
verbose=verbose, header='Arrays are not equal')
File "/home/user/python27-install/lib/python2.7/site-packages/numpy/testing/utils.py", line 600, in assert_array_compare
raise AssertionError(msg)
AssertionError:
Arrays are not equal

(shapes (0,), (1,) mismatch)
x: array([], dtype=int64)
y: array([1])

======================================================================
FAIL: test_kind.TestKind.test_all
----------------------------------------------------------------------
Traceback (most recent call last):
File "/home/user/python27-install/lib/python2.7/site-packages/nose/case.py", line 197, in runTest
self.test(*self.arg)
File "/home/user/python27-install/lib/python2.7/site-packages/numpy/f2py/tests/test_kind.py", line 30, in test_all
'selectedrealkind(%s): expected %r but got %r' % (i, selected_real_kind(i), selectedrealkind(i)))
File "/home/user/python27-install/lib/python2.7/site-packages/numpy/testing/utils.py", line 34, in assert_
raise AssertionError(msg)
AssertionError: selectedrealkind(16): expected 10 but got 16

----------------------------------------------------------------------
Ran 3552 tests in 52.984s

FAILED (KNOWNFAIL=3, SKIP=1, failures=2)

-thanks,
Vyas

6 posts / 0 new
Last post
For more complete information about compiler optimizations, see our Optimization Notice.

What libs did you link with? (32 bit or 64 bit)?

--Vipin

Hi,

I linked against 64bit libs. Here is how I generated my custom libs

In tools/builder directory

make libintel64 name=/home/user/libmkl_blas export=blas_example_list
make libintel64 name=/home/user/libmkl_cbals export=cblas_example_list
make libintel64 name=/home/user/libmkl_lapack export=lapack_example_list

Hi,

I have some questions related to used MKL libraries.

Quoting vyasdmake libintel64 name=/home/user/libmkl_blas export=blas_example_list
make libintel64 name=/home/user/libmkl_cbals export=cblas_example_list
make libintel64 name=/home/user/libmkl_lapack export=lapack_example_list

What exactly custom library was used to link with Numpy? Did you change example-lists?

Thanks,
-- Victor

Hi,Here is my Numpy 1.6.1 site.cfg

[mkl]
library_dirs = /home/vduggira:/cm/shared/apps/intel/composerxe/2011.5.220/compiler/lib/intel64:/cm/shared/apps/intel/composerxe/2011.5.220/mkl/lib/intel64
include_dirs = /cm/shared/apps/intel/composerxe/2011.5.220/mkl/include
mkl_libs = mkl_blas, mkl_cblas
lapack_libs = mkl_lapack

The example-lists haven't been edited in any way.(This is numpy 1.6.1, the latest release. Not 1.6 as refered to in the subject line)
thanks,Vyas

Hi,

I'd recommend to create the only one custom library (withall needed functions together: blas, cblas, lapack) to be used in site.cfg

Thanks,
-- Victor

Leave a Comment

Please sign in to add a comment. Not a member? Join today