Intel® Distribution for Python*
Engage in discussions with community peers related to Python* applications and core computational packages.

K-means with Scikit-Learn

velvia
Beginner
1,209 Views

Hi,

The following link ( https://software.intel.com/en-us/articles/intelr-distribution-for-python-2017-update-2 ) states that the Intel Python Distribution 2017 Update 2 uses the DAAL as a backend for K-means clustering in Scikit-Learn.

Using the following code, on Linux, does not seem to use the DAAL as reported by vTune. It still runs the original Scikit-Learn code written in Cython. Is there anything to do to enable the DAAL?

import os
from time import time
import numpy as np
from sklearn.cluster import KMeans

dim = 784
nb_points = 60000
nb_clusters = 10
nb_iterations = 20
nb_threads = 1

os.environ['OMP_NUM_THREADS'] = '28'

points = np.random.rand(nb_points, dim)

estimator = KMeans(init = "random", verbose = 1, max_iter = nb_iterations,
                   algorithm = "full", precompute_distances = False, tol = 0.0,
                   n_clusters = nb_clusters, n_init = 1)
print("Before computing")
time_begin = time()
estimator.fit(points)
time_end = time()
print("After computing")
print((time_end - time_begin) / nb_iterations)

 

0 Kudos
7 Replies
gaston-hillar
Valued Contributor I
1,209 Views

Hi Velvia,

Check the following comment related to KMeans and DAAL: https://software.intel.com/en-us/forums/intel-distribution-for-python/topic/731351#comment-1903917

 

0 Kudos
gaston-hillar
Valued Contributor I
1,209 Views

Hi Velvia,

I think that the entire thread will be useful because in this thread it was suggested to disable daal optimizations. So, probably the same problem applies to the issue you are reporting. I haven't tested your code in my configs. However, I remember the last comments of the thread I mention: https://software.intel.com/en-us/forums/intel-distribution-for-python/topic/731351

0 Kudos
DavidLiu
Employee
1,209 Views

Hi Velvia,

Can you give me the output of your 'conda list', 'conda info', and your system specs?

Thanks,

-David

0 Kudos
velvia
Beginner
1,209 Views

Hi David.

Sure. Here it is:

[fayard@grisbouille Digits]$ conda info
Current conda install:

               platform : linux-64
          conda version : 4.2.12
       conda is private : False
      conda-env version : 4.2.12
    conda-build version : not installed
         python version : 3.5.2.final.0
       requests version : 2.11.1
       root environment : /opt/intel/intelpython3  (writable)
    default environment : /opt/intel/intelpython3
       envs directories : /opt/intel/intelpython3/envs
          package cache : /opt/intel/intelpython3/pkgs
           channel URLs : https://conda.anaconda.org/intel/linux-64
                          https://conda.anaconda.org/intel/noarch
                          https://repo.continuum.io/pkgs/free/linux-64
                          https://repo.continuum.io/pkgs/free/noarch
                          https://repo.continuum.io/pkgs/pro/linux-64
                          https://repo.continuum.io/pkgs/pro/noarch
            config file : /opt/intel/intelpython3/.condarc
           offline mode : False

[fayard@grisbouille Digits]$ conda list
# packages in environment at /opt/intel/intelpython3:
#
backports                 1.0                py35_intel_6  [intel]  intel
boost                     1.61.0             py35_intel_3  [intel]  intel
caffe                     1.0.0             py35_intel_10  [intel]  intel
conda                     4.2.12             py35_intel_0  [intel]  intel
conda-env                 2.6.0                         0    intel
cycler                    0.10.0             py35_intel_5  [intel]  intel
cython                    0.24              py35_intel_16  [intel]  intel
dbus                      1.10.10                       0  
decorator                 4.0.10             py35_intel_5  [intel]  intel
distarray                 0.6.0              py35_intel_4  [intel]  intel
entrypoints               0.2.2              py35_intel_1  [intel]  intel
expat                     2.1.0                         0  
fontconfig                2.11.1                  intel_7  [intel]  intel
freetype                  2.5.5                   intel_9  [intel]  intel
get_terminal_size         1.0.0              py35_intel_5  [intel]  intel
gflags                    2.1.2                   intel_3  [intel]  intel
glib                      2.50.2                        1  
glog                      0.3.4                   intel_2  [intel]  intel
gst-plugins-base          1.8.0                         0  
gstreamer                 1.8.0                         0  
h5py                      2.6.0           np111py35_intel_1  [intel]  intel
hdf5                      1.8.17                  intel_0  [intel]  intel
icc_rt                    16.0.3                  intel_5  [intel]  intel
icu                       54.1                    intel_2  [intel]  intel
impi_rt                   2017.0.1                intel_2  [intel]  intel
intelpython               2017.0.2                      1    intel
ipykernel                 4.5.0              py35_intel_4  [intel]  intel
ipyparallel               5.2.0              py35_intel_1  [intel]  intel
ipython                   4.2.0              py35_intel_8  [intel]  intel
ipython_genutils          0.1.0              py35_intel_5  [intel]  intel
ipywidgets                5.2.2              py35_intel_1  [intel]  intel
jbig                      2.1                           0  
jinja2                    2.8                py35_intel_5  [intel]  intel
jpeg                      8d                      intel_3  [intel]  intel
jsonschema                2.5.1              py35_intel_5  [intel]  intel
jupyter                   1.0.0              py35_intel_5  [intel]  intel
jupyter_client            4.4.0              py35_intel_2  [intel]  intel
jupyter_console           5.0.0              py35_intel_5  [intel]  intel
jupyter_core              4.2.0              py35_intel_5  [intel]  intel
keras                     2.0.2                    py35_0  
leveldb                   1.18                    intel_2  [intel]  intel
libffi                    3.2.1                         1  
libgcc                    5.2.0                         0  
libiconv                  1.14                          0  
libpng                    1.6.28                 intel_10  [intel]  intel
libsodium                 1.0.10                  intel_6  [intel]  intel
libtiff                   4.0.7                   intel_8  [intel]  intel
libxcb                    1.12                          1  
libxml2                   2.9.4                  intel_10  [intel]  intel
llvmlite                  0.15.0             py35_intel_0  [intel]  intel
lmdb                      0.9.18                  intel_2  [intel]  intel
lmdb                      0.92                      <pip>
markupsafe                0.23               py35_intel_6  [intel]  intel
matplotlib                1.5.3              py35_intel_3  [intel]  intel
mistune                   0.7.3              py35_intel_1  [intel]  intel
mkl                       2017.0.2                intel_1  [intel]  intel
mkl-dnn                   0.3dev                  intel_0  [intel]  intel
mock                      2.0.0                    py35_0  
mpi4py                    2.0.0              py35_intel_9  [intel]  intel
mpmath                    0.19               py35_intel_5  [intel]  intel
nbconvert                 4.2.0              py35_intel_5  [intel]  intel
nbformat                  4.1.0              py35_intel_1  [intel]  intel
networkx                  1.11               py35_intel_0  [intel]  intel
nose                      1.3.7             py35_intel_12  [intel]  intel
notebook                  4.2.3              py35_intel_7  [intel]  intel
numba                     0.30.1            py35_intel_11  [intel]  intel
numexpr                   2.6.1           np111py35_intel_11  [intel]  intel
numpy                     1.11.2            py35_intel_51  [intel]  intel
olefile                   0.44                     py35_0  
openssl                   1.0.2k                  intel_3  [intel]  intel
pandas                    0.19.2             py35_intel_1  [intel]  intel
path.py                   8.2.1              py35_intel_7  [intel]  intel
pbr                       1.10.0                   py35_0  
pcre                      8.39                          1  
pexpect                   4.0.1              py35_intel_5  [intel]  intel
pickleshare               0.7.4              py35_intel_1  [intel]  intel
pillow                    3.4.2              py35_intel_4  [intel]  intel
pip                       8.1.2              py35_intel_5  [intel]  intel
prompt_toolkit            1.0.3              py35_intel_5  [intel]  intel
protobuf                  3.0.0              py35_intel_1  [intel]  intel
ptyprocess                0.5.1              py35_intel_5  [intel]  intel
pycosat                   0.6.1              py35_intel_6  [intel]  intel
pydaal                    2017.0.2.20170126    py35_intel_0  [intel]  intel
pygments                  2.1.3              py35_intel_6  [intel]  intel
pyparsing                 2.1.4              py35_intel_5  [intel]  intel
pytables                  3.2.3.1            py35_intel_8  [intel]  intel
python                    3.5.2                  intel_27  [intel]  intel
python-dateutil           2.5.3              py35_intel_5  [intel]  intel
python-gflags             3.1.0              py35_intel_1  [intel]  intel
python-leveldb            0.193              py35_intel_0  [intel]  intel
pytz                      2016.6.1           py35_intel_5  [intel]  intel
pyyaml                    3.12               py35_intel_1  [intel]  intel
pyzmq                     15.4.0             py35_intel_3  [intel]  intel
qt                        5.6.2                         0  
readline                  6.2                           2  
requests                  2.11.1             py35_intel_1  [intel]  intel
ruamel_yaml               0.11.14            py35_intel_1  [intel]  intel
scikit-image              0.12.3          np111py35_intel_2  [intel]  intel
scikit-learn              0.18.1             py35_intel_4  [intel]  intel
scipy                     0.18.1          np111py35_intel_10  [intel]  intel
setuptools                23.0.0             py35_intel_4  [intel]  intel
simplegeneric             0.8.1              py35_intel_5  [intel]  intel
six                       1.10.0             py35_intel_7  [intel]  intel
snappy                    1.1.3                   intel_2  [intel]  intel
sqlite                    3.13.0                 intel_13  [intel]  intel
sympy                     1.0                py35_intel_5  [intel]  intel
tbb                       2017.0.4           py35_intel_0  [intel]  intel
tcl                       8.6.4                  intel_16  [intel]  intel
tensorflow                0.10.0rc0           np111py35_0  
terminado                 0.6                py35_intel_6  [intel]  intel
theano                    1.0                py35_intel_0  [intel]  intel
tk                        8.6.4                  intel_18  [intel]  intel
tornado                   4.4.1              py35_intel_3  [intel]  intel
traitlets                 4.2.2              py35_intel_3  [intel]  intel
wcwidth                   0.1.7              py35_intel_5  [intel]  intel
werkzeug                  0.12.1                   py35_0  
wheel                     0.29.0             py35_intel_5  [intel]  intel
widgetsnbextension        1.2.6                    py35_0    intel
xz                        5.2.2                  intel_15  [intel]  intel
yaml                      0.1.6                  intel_10  [intel]  intel
zeromq                    4.1.4                   intel_5  [intel]  intel
zlib                      1.2.11                  intel_0  [intel]  intel

 

0 Kudos
DavidLiu
Employee
1,209 Views

Hi Velvia, 

Thanks for your reply.  I've asked engineering for further clarification on the DAAL functionality under scikit-learn, which will hopefully shed some light as to why your example is reverting back to normal Cython.

Thanks,

David

0 Kudos
Denis_N_Intel
Employee
1,209 Views

Hi Velvia, 

There were some updates on our channel. You can try to update scikit-learn with 

conda update -c intel scikit-learn

Denis.

 

0 Kudos
Denis_N_Intel
Employee
1,209 Views

Hello,

We've released Update 3, so you can try with it too.

Denis.

 

 

0 Kudos
Reply