[Numpy-discussion] Fwd: Fwd: numpy error with mkl 10.1

Bruce Southey bsouthey@gmail....
Thu Nov 3 07:09:08 CDT 2011


On Wed, Nov 2, 2011 at 10:33 PM, akshar bhosale
<akshar.bhosale@gmail.com> wrote:
> any other means of getting it fixed?
>
> ---------- Forwarded message ----------
> From: Olivier Delalleau <shish@keba.be>
> Date: Thu, Nov 3, 2011 at 8:58 AM
> Subject: Re: [Numpy-discussion] Fwd: numpy error with mkl 10.1
> To: Discussion of Numerical Python <numpy-discussion@scipy.org>
>
>
> I'm sorry, I don't know. I'm not at all an expert about numpy installation,
> I just happened to have access to a computer with a working install of numpy
> with MKL, so I thought it might help... but I'm afraid that's all I can do
> for you :/
>
> -=- Olivier
>
> 2011/11/2 akshar bhosale <akshar.bhosale@gmail.com>
>>
>> thanks once again. how can i use ur distutils directly (ur supplied files
>> etc)
>>
>> On Thu, Nov 3, 2011 at 8:15 AM, Olivier Delalleau <shish@keba.be> wrote:
>>>
>>> Yes they pass:
>>>
>>> Ran 2030 tests in 6.672s
>>>
>>> OK (KNOWNFAIL=1)
>>> <nose.result.TextTestResult run=2030 errors=0 failures=0>
>>>
>>> Google "einsum hang".... I know you're not the first one with this bug
>>> (but I'm not sure exactly how to fix it).
>>>
>>> -=- Olivier
>>>
>>> 2011/11/2 akshar bhosale <akshar.bhosale@gmail.com>
>>>>
>>>> my numpy.test hangs here
>>>>
>>>> Test whether equivalent subarray dtypes hash the same. ... ok
>>>> Test whether different subarray dtypes hash differently. ... ok
>>>> Test some data types that are equal ... ok
>>>> Test some more complicated cases that shouldn't be equal ... ok
>>>> Test some simple cases that shouldn't be equal ... ok
>>>> test_single_subarray (test_dtype.TestSubarray) ... ok
>>>> test_einsum_errors (test_einsum.TestEinSum) ... ok
>>>> test_einsum_sums_cfloat128 (test_einsum.TestEinSum) ...
>>>>
>>>>
>>>> ---------- Forwarded message ----------
>>>> From: akshar bhosale <akshar.bhosale@gmail.com>
>>>> Date: Thu, Nov 3, 2011 at 8:00 AM
>>>> Subject: Re: [Numpy-discussion] numpy error with mkl 10.1
>>>> To: Discussion of Numerical Python <numpy-discussion@scipy.org>
>>>>
>>>>
>>>> hi,
>>>> on your machine, is numpy.test passed?
>>>>
>>>> On Thu, Nov 3, 2011 at 6:54 AM, Olivier Delalleau <shish@keba.be> wrote:
>>>>>
>>>>> I believe they are optional, but I'm not sure.
>>>>>
>>>>> Note that this system I copied the distutils from is not one I setup
>>>>> myself so I'm not sure how numpy was installed exactly. I just know it works
>>>>> and it's using MKL :) It's an 8 core Intel(R) Xeon(R) CPU E5462  @ 2.80GHz
>>>>> (x86_64 as well).
>>>>>
>>>>> -=- Olivier
>>>>>
>>>>> 2011/11/2 akshar bhosale <akshar.bhosale@gmail.com>
>>>>>>
>>>>>> hi,
>>>>>> are amd,fftw and umfpack required? my architecture is : intel xeon
>>>>>> x_86_64.
>>>>>>
>>>>>> On Thu, Nov 3, 2011 at 6:30 AM, Olivier Delalleau <shish@keba.be>
>>>>>> wrote:
>>>>>>>
>>>>>>> It's inside the distutils folder.
>>>>>>>
>>>>>>> -=- Olivier
>>>>>>>
>>>>>>> 2011/11/2 akshar bhosale <akshar.bhosale@gmail.com>
>>>>>>>>
>>>>>>>> thanks..what about site.cfg?
>>>>>>>>
>>>>>>>> On Thu, Nov 3, 2011 at 1:27 AM, Olivier Delalleau <shish@keba.be>
>>>>>>>> wrote:
>>>>>>>>>
>>>>>>>>> Sorry, no clue :/
>>>>>>>>>
>>>>>>>>> I made a tarball with my distutils folder here:
>>>>>>>>> http://www.iro.umontreal.ca/~delallea/tmp/distutils.tar.bz2
>>>>>>>>> Hope this helps...
>>>>>>>>>
>>>>>>>>> -=- Olivier
>>>>>>>>>
>>>>>>>>> 2011/11/2 akshar bhosale <akshar.bhosale@gmail.com>
>>>>>>>>>>
>>>>>>>>>> hi,
>>>>>>>>>> thanks for the reply
>>>>>>>>>> none of them is linked with libmkl_lapack.so.
>>>>>>>>>> what can be the problem?
>>>>>>>>>> libmkl_lapack.so is present in its location.
>>>>>>>>>>
>>>>>>>>>> if possible, can you send your site.cfg and other related files
>>>>>>>>>> like intelccompilers.py / system_config.py etc...and configure/build
>>>>>>>>>> options, i will try the same?
>>>>>>>>>>
>>>>>>>>>> On Wed, Nov 2, 2011 at 11:51 PM, Olivier Delalleau <shish@keba.be>
>>>>>>>>>> wrote:
>>>>>>>>>>>
>>>>>>>>>>> Hmm that's interesting, it's not linked against libmkl_lapack. On
>>>>>>>>>>> my system with MKL it says:
>>>>>>>>>>>
>>>>>>>>>>> ldd linalg/lapack_lite.so
>>>>>>>>>>>         libmkl_lapack.so => /opt/intel/mkl/MailScanner soupçonne
>>>>>>>>>>> le lien suivant d'être une tentative de fraude de la part de "10.1.3.027"
>>>>>>>>>>> MailScanner soupçonne le lien suivant d'être une tentative de fraude de la
>>>>>>>>>>> part de "10.1.3.027" MailScanner soupçonne le lien suivant d'être une
>>>>>>>>>>> tentative de fraude de la part de "10.1.3.027" MailScanner soupçonne le lien
>>>>>>>>>>> suivant d'être une tentative de fraude de la part de "10.1.3.027"
>>>>>>>>>>> MailScanner warning: numerical links are often malicious:
>>>>>>>>>>> 10.1.3.027/lib/em64t/libmkl_lapack.so (0x00002acf0e25a000)
>>>>>>>>>>>         libmkl_intel_lp64.so => /opt/intel/mkl/MailScanner
>>>>>>>>>>> soupçonne le lien suivant d'être une tentative de fraude de la part de
>>>>>>>>>>> "10.1.3.027" MailScanner soupçonne le lien suivant d'être une tentative de
>>>>>>>>>>> fraude de la part de "10.1.3.027" MailScanner soupçonne le lien suivant
>>>>>>>>>>> d'être une tentative de fraude de la part de "10.1.3.027" MailScanner
>>>>>>>>>>> soupçonne le lien suivant d'être une tentative de fraude de la part de
>>>>>>>>>>> "10.1.3.027" MailScanner warning: numerical links are often malicious:
>>>>>>>>>>> 10.1.3.027/lib/em64t/libmkl_intel_lp64.so (0x00002acf0ebb8000)
>>>>>>>>>>>         libmkl_intel_thread.so => /opt/intel/mkl/MailScanner
>>>>>>>>>>> soupçonne le lien suivant d'être une tentative de fraude de la part de
>>>>>>>>>>> "10.1.3.027" MailScanner soupçonne le lien suivant d'être une tentative de
>>>>>>>>>>> fraude de la part de "10.1.3.027" MailScanner soupçonne le lien suivant
>>>>>>>>>>> d'être une tentative de fraude de la part de "10.1.3.027" MailScanner
>>>>>>>>>>> soupçonne le lien suivant d'être une tentative de fraude de la part de
>>>>>>>>>>> "10.1.3.027" MailScanner warning: numerical links are often malicious:
>>>>>>>>>>> 10.1.3.027/lib/em64t/libmkl_intel_thread.so (0x00002acf0ef2e000)
>>>>>>>>>>>         libmkl_def.so => /opt/intel/mkl/MailScanner soupçonne le
>>>>>>>>>>> lien suivant d'être une tentative de fraude de la part de "10.1.3.027"
>>>>>>>>>>> MailScanner soupçonne le lien suivant d'être une tentative de fraude de la
>>>>>>>>>>> part de "10.1.3.027" MailScanner soupçonne le lien suivant d'être une
>>>>>>>>>>> tentative de fraude de la part de "10.1.3.027" MailScanner soupçonne le lien
>>>>>>>>>>> suivant d'être une tentative de fraude de la part de "10.1.3.027"
>>>>>>>>>>> MailScanner warning: numerical links are often malicious:
>>>>>>>>>>> 10.1.3.027/lib/em64t/libmkl_def.so (0x00002acf0fc93000)
>>>>>>>>>>>         libmkl_core.so => /opt/intel/mkl/MailScanner soupçonne le
>>>>>>>>>>> lien suivant d'être une tentative de fraude de la part de "10.1.3.027"
>>>>>>>>>>> MailScanner soupçonne le lien suivant d'être une tentative de fraude de la
>>>>>>>>>>> part de "10.1.3.027" MailScanner soupçonne le lien suivant d'être une
>>>>>>>>>>> tentative de fraude de la part de "10.1.3.027" MailScanner soupçonne le lien
>>>>>>>>>>> suivant d'être une tentative de fraude de la part de "10.1.3.027"
>>>>>>>>>>> MailScanner warning: numerical links are often malicious:
>>>>>>>>>>> 10.1.3.027/lib/em64t/libmkl_core.so (0x00002acf1080e000)
>>>>>>>>>>>         libguide.so => /opt/intel/cce/10.1.026/lib/libguide.so
>>>>>>>>>>> (0x00002acf10a03000)
>>>>>>>>>>>         libpthread.so.0 => /lib64/libpthread.so.0
>>>>>>>>>>> (0x00002acf10ba6000)
>>>>>>>>>>>         libm.so.6 => /lib64/libm.so.6 (0x00002acf10dc1000)
>>>>>>>>>>>         libpython2.6.so.1.0 =>
>>>>>>>>>>> /opt/python64/2.6.4/lib/libpython2.6.so.1.0 (0x00002acf11044000)
>>>>>>>>>>>         libc.so.6 => /lib64/libc.so.6 (0x00002acf113f3000)
>>>>>>>>>>>         libdl.so.2 => /lib64/libdl.so.2 (0x00002acf1174b000)
>>>>>>>>>>>         /lib64/ld-linux-x86-64.so.2 (0x0000003d90400000)
>>>>>>>>>>>         libutil.so.1 => /lib64/libutil.so.1 (0x00002acf1194f000)
>>>>>>>>>>>
>>>>>>>>>>> So to clarify, what I was trying to achieve is find which of
>>>>>>>>>>> numpy's shared libraries was trying to load libmkl_lapack.so, and if it
>>>>>>>>>>> could be found properly with ldd. Just to make sure it's not another .so,
>>>>>>>>>>> can you run from within your numpy directory:
>>>>>>>>>>>    find -name "*.so" -exec ldd \{} \; | grep lapack
>>>>>>>>>>>
>>>>>>>>>>> If the output is not empty, try to run ldd on all .so within
>>>>>>>>>>> numpy to see which one is linked against libmkl_lapack.so. Then hopefully
>>>>>>>>>>> something will stand out (I admit that's not guaranteed though ;).
>>>>>>>>>>>
>>>>>>>>>>> -=- Olivier
>>>>>>>>>>>
>>>>>>>>>>> 2011/11/2 akshar bhosale <akshar.bhosale@gmail.com>
>>>>>>>>>>>>
>>>>>>>>>>>> Hi,
>>>>>>>>>>>>
>>>>>>>>>>>> ldd
>>>>>>>>>>>> Python-2.6/lib/python2.6/site-packages/numpy/linalg/lapack_lite.so
>>>>>>>>>>>>     libstdc++.so.6 => /usr/lib64/libstdc++.so.6
>>>>>>>>>>>> (0x00002b1199349000)
>>>>>>>>>>>>     libmkl_def.so =>
>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/mkl/lib/em64t/libmkl_def.so
>>>>>>>>>>>> (0x00002b1199653000)
>>>>>>>>>>>>     libmkl_intel_lp64.so =>
>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/mkl/lib/em64t/libmkl_intel_lp64.so
>>>>>>>>>>>> (0x00002b119a1a8000)
>>>>>>>>>>>>     libmkl_intel_thread.so =>
>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/mkl/lib/em64t/libmkl_intel_thread.so
>>>>>>>>>>>> (0x00002b119a503000)
>>>>>>>>>>>>     libmkl_core.so =>
>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/mkl/lib/em64t/libmkl_core.so
>>>>>>>>>>>> (0x00002b119b22c000)
>>>>>>>>>>>>     libmkl_mc.so =>
>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/mkl/lib/em64t/libmkl_mc.so (0x00002b119b420000)
>>>>>>>>>>>>     libpthread.so.0 => /lib64/libpthread.so.0
>>>>>>>>>>>> (0x00002b119c176000)
>>>>>>>>>>>>     libimf.so =>
>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/lib/intel64/libimf.so (0x00002b119c392000)
>>>>>>>>>>>>     libsvml.so =>
>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/lib/intel64/libsvml.so (0x00002b119c6e9000)
>>>>>>>>>>>>     libm.so.6 => /lib64/libm.so.6 (0x00002b119c8a6000)
>>>>>>>>>>>>     libiomp5.so =>
>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/lib/intel64/libiomp5.so (0x00002b119cb2b000)
>>>>>>>>>>>>     libintlc.so.5 =>
>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/lib/intel64/libintlc.so.5 (0x00002b119ccbc000)
>>>>>>>>>>>>     libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00002b119cdf9000)
>>>>>>>>>>>>     libc.so.6 => /lib64/libc.so.6 (0x00002b119d007000)
>>>>>>>>>>>>     libdl.so.2 => /lib64/libdl.so.2 (0x00002b119d37e000)
>>>>>>>>>>>>     /lib64/ld-linux-x86-64.so.2 (0x0000003fb8000000)
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> On Wed, Nov 2, 2011 at 11:26 PM, Olivier Delalleau
>>>>>>>>>>>> <shish@keba.be> wrote:
>>>>>>>>>>>>>
>>>>>>>>>>>>> Doh I'm sorry, I forgot the problem was with lapack, what about
>>>>>>>>>>>>> ldd numpy/linalg/lapack_lite.so?
>>>>>>>>>>>>>
>>>>>>>>>>>>> 2011/11/2 akshar bhosale <akshar.bhosale@gmail.com>
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> Hi,
>>>>>>>>>>>>>> ldd _dotblas.so
>>>>>>>>>>>>>> libstdc++.so.6 => /usr/lib64/libstdc++.so.6
>>>>>>>>>>>>>> (0x00002b12f0692000)
>>>>>>>>>>>>>>     libmkl_def.so =>
>>>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/mkl/lib/em64t/libmkl_def.so
>>>>>>>>>>>>>> (0x00002b12f099c000)
>>>>>>>>>>>>>>     libmkl_intel_lp64.so =>
>>>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/mkl/lib/em64t/libmkl_intel_lp64.so
>>>>>>>>>>>>>> (0x00002b12f14f1000)
>>>>>>>>>>>>>>     libmkl_intel_thread.so =>
>>>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/mkl/lib/em64t/libmkl_intel_thread.so
>>>>>>>>>>>>>> (0x00002b12f184c000)
>>>>>>>>>>>>>>     libmkl_core.so =>
>>>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/mkl/lib/em64t/libmkl_core.so
>>>>>>>>>>>>>> (0x00002b12f2575000)
>>>>>>>>>>>>>>     libmkl_mc.so =>
>>>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/mkl/lib/em64t/libmkl_mc.so (0x00002b12f2769000)
>>>>>>>>>>>>>>     libpthread.so.0 => /lib64/libpthread.so.0
>>>>>>>>>>>>>> (0x00002b12f34bf000)
>>>>>>>>>>>>>>     libimf.so =>
>>>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/lib/intel64/libimf.so (0x00002b12f36db000)
>>>>>>>>>>>>>>     libsvml.so =>
>>>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/lib/intel64/libsvml.so (0x00002b12f3a32000)
>>>>>>>>>>>>>>     libm.so.6 => /lib64/libm.so.6 (0x00002b12f3bef000)
>>>>>>>>>>>>>>     libiomp5.so =>
>>>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/lib/intel64/libiomp5.so (0x00002b12f3e74000)
>>>>>>>>>>>>>>     libintlc.so.5 =>
>>>>>>>>>>>>>> /opt/intel/Compiler/11.0/069/lib/intel64/libintlc.so.5 (0x00002b12f4005000)
>>>>>>>>>>>>>>     libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00002b12f4142000)
>>>>>>>>>>>>>>     libc.so.6 => /lib64/libc.so.6 (0x00002b12f4350000)
>>>>>>>>>>>>>>     libdl.so.2 => /lib64/libdl.so.2 (0x00002b12f46c7000)
>>>>>>>>>>>>>>     /lib64/ld-linux-x86-64.so.2 (0x0000003fb8000000)
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> On Wed, Nov 2, 2011 at 10:14 PM, Olivier Delalleau
>>>>>>>>>>>>>> <shish@keba.be> wrote:
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> Ok, can you print the output of ldd numpy/core/_dotblas.so?
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> -=- Olivier
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> 2011/11/2 akshar bhosale <akshar.bhosale@gmail.com>
>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>> HI,
>>>>>>>>>>>>>>>> It is already added in the LD_LIBRARY_PATH, thenalso it is
>>>>>>>>>>>>>>>> generating the same error.
>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>> On Wed, Nov 2, 2011 at 10:01 PM, Olivier Delalleau
>>>>>>>>>>>>>>>> <shish@keba.be> wrote:
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> Locate your libmkl_lapack.so and try to add the directory
>>>>>>>>>>>>>>>>> that contains it to your LD_LIBRARY_PATH environment variable.
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> -=- Olivier
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> 2011/11/2 akshar bhosale <akshar.bhosale@gmail.com>
>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>> Hi,
>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>> i am getting following error.
>>>>>>>>>>>>>>>>>>    python -c 'import numpy;numpy.matrix([[1, 5, 10], [1.0,
>>>>>>>>>>>>>>>>>> 3j, 4]],
>>>>>>>>>>>>>>>>>>    numpy.complex128).T.I.H'
>>>>>>>>>>>>>>>>>>    MKL FATAL ERROR: Cannot load libmkl_lapack.so
>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>>    have installed numpy 1.6.0 with python 2.6.
>>>>>>>>>>>>>>>>>>    i have intel cluster toolkit installed on my system.
>>>>>>>>>>>>>>>>>> (11/069 version and mlk=10.1). i have machine having intel xeon processor
>>>>>>>>>>>>>>>>>> and rhel 5.2 x86_64
>>>>>>>>>>>>>>>>>>    platform.
>>>>>>>>>>>>>>>>>> Kindly help
>>>>>>>>>>>>>>>>>>

You have to isolate the specific source yourself if you want further help.
For example:
1) Does Numpy install and pass tests with gcc?
2) Does Numpy install and pass tests with gcc and Atlas?
3) Have you correctly installed or reinstalled the latest Python and
Intel compiler/libraries? This specific thread suggests that it was
not.
4) Which part of the  test_einsum_sums_cfloat128 fails? (Previously asked.)

Bruce


More information about the NumPy-Discussion mailing list