I don't see how to install pyqt into intelpython environment:
conda install pyqt --channel=intel
Using Anaconda Cloud api site https://api.anaconda.org
Fetching package metadata: ..........
Solving package specifications: ....
The following specifications were found to be in conflict:
- bzip2 (target=bzip2-1.0.6-intel_py3.5_1.tar.bz2)
- db (target=db-4.7.25-intel_py3.5_1.tar.bz2)
- icc_rt (target=icc_rt-15.0.6-intel_py3.5_1.tar.bz2)
- libffi (target=libffi-3.2.1-intel_py3.5_1.tar.bz2)
- openssl (target=openssl-1.0.2g-intel_py3.5_1.tar.bz2)
- pip (target=pip-7.1.2-intel_py3.5_1.tar.bz2) -> python 3.5*
- pip (target=pip-7.1.2-intel_py3.5_1.tar.bz2) -> setuptools
- pyqt -> python 3.5*
- python 3.5*
- setuptools (target=setuptools-18.2-intel_py3.5_1.tar.bz2) -> python 3.5*
- sqlite (target=sqlite-3.8.11.1-intel_py3.5_1.tar.bz2)
- tcl (target=tcl-8.6.4-intel_py3.5_1.tar.bz2)
- tix (target=tix-8.4.3-intel_py3.5_1.tar.bz2)
- tk (target=tk-8.6.4-intel_py3.5_1.tar.bz2)
- xz (target=xz-5.2.1-intel_py3.5_1.tar.bz2)
- zlib (target=zlib-1.2.8-intel_py3.5_1.tar.bz2)
Use "conda info <package>" to see the dependencies for each package.
(intelpython3)[nbecker@nbecker2 envs]$ conda install pyqt --channel=intel --override-channel
Using Anaconda Cloud api site https://api.anaconda.org
Fetching package metadata: ..
Solving package specifications: .
Error: Package missing in current linux-64 channels:
- pyqt
Did you mean one of these?
pytz
You can search for this package on anaconda.org with
anaconda search -t conda pyqt
Link Copied
A workaround is to:
conda remove freetype2
You will switch some packages from intel to continuum, including matplotlib, but you will still have intel numpy/scipy
We are working with continuum to resolve depedences that make it difficult to mix packages.
OK, I did
conda remove freetype2
Then:
conda install pyqt
Using Anaconda Cloud api site https://api.anaconda.org
Fetching package metadata: ........
Solving package specifications: ....
The following specifications were found to be in conflict:
- bzip2 (target=bzip2-1.0.6-intel_py3.5_1.tar.bz2)
- db (target=db-4.7.25-intel_py3.5_1.tar.bz2)
- icc_rt (target=icc_rt-15.0.6-intel_py3.5_1.tar.bz2)
- libffi (target=libffi-3.2.1-intel_py3.5_1.tar.bz2)
- openssl (target=openssl-1.0.2g-intel_py3.5_1.tar.bz2)
- pip (target=pip-7.1.2-intel_py3.5_1.tar.bz2) -> python 3.5*
- pyqt -> python 3.5*
- python 3.5*
- setuptools (target=setuptools-18.2-intel_py3.5_1.tar.bz2) -> python 3.5*
- sqlite (target=sqlite-3.8.11.1-intel_py3.5_1.tar.bz2)
- tcl (target=tcl-8.6.4-intel_py3.5_1.tar.bz2)
- tix (target=tix-8.4.3-intel_py3.5_1.tar.bz2)
- tk (target=tk-8.6.4-intel_py3.5_1.tar.bz2)
- xz (target=xz-5.2.1-intel_py3.5_1.tar.bz2)
- zlib (target=zlib-1.2.8-intel_py3.5_1.tar.bz2)
The problem is a version conflict between the packages we provide and the requirements of pyqt from anaconda. If you start from miniconda and create an intel python environment:
conda create -n idp3m intelpython3_core
Then I am able to install pyqt with no conflict. If I start with Intel python installer, then there is a conflict on the versions of libpng, but uninstalling freetype2 removes the conflict and installs pyqt. I think you have a different combination of packages and the conflict is different. If you can tell me how to create an conda environment that shows the problem, then I can reproduce and track it down.
We have asked Continuum to update their package to also accept libpng 1.6.18, instead of only 1.6.17, which would fix the problem that I see. We are also looking at more general solutions to avoid version conflicts in the future.
The original problem was that when I tried to plot with matplotlib, I got no screen output (and now error message).
I did notice that
%matplotlib
said setting output to Agg.
Usually, I would see
In [32]: %matplotlib
Using matplotlib backend: Qt5Agg
So I thought maybe the problem was caused by pyqt not being installed.
Otherwise, my original problem is still that matplotlib doesn't seem to produce any screen output.
If you download intel python from intel and activate the root environment, it works. If you create a conda environment, with intelpython3_core and matplotlib then plotting does not work--it silently exits for me.
I think there is a missing dependence that is only uncovered when you install packages individually. I need to investigate some more.
For more complete information about compiler optimizations, see our Optimization Notice.