Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

PPC64LE Conda Install #633

Closed
DNA2RNA opened this issue Jul 1, 2020 · 8 comments
Closed

PPC64LE Conda Install #633

DNA2RNA opened this issue Jul 1, 2020 · 8 comments

Comments

@DNA2RNA
Copy link

DNA2RNA commented Jul 1, 2020

Hi,

Is there any chance Open Force Field might get a PPC64LE or noarch installation option?

Thanks!

@mattwthompson
Copy link
Member

We'd like to do one or both of these at some point in the future but, unfortunately, it's probably not something we can do in the short term. The OpenFF toolkit is pure Python, so it comes down tot he dependencies. Most are distributed as noarch and/or ppc64le, but some important ones don't seem to be: RDKit and OpenMM stick out to me at a glance. They can easy probably be compiled locally but I have not tried myself and don't have access to a PowerPC machine at the moment to test. I'm not sure about the OpenEye toolkits, if you happen to have a license.

@jchodera
Copy link
Member

jchodera commented Jul 1, 2020

(cc @j-wags @jaimergp @peastman)

We are actively working to migrate OpenMM to conda-forge, where we hope to be able to support both ppc64le and ARM packages. Hopefully this can be wrapped up quickly.

We might open an issue on the https://github.com/conda-forge/rdkit-feedstock to see if they could add ppc64le and ARM support as well, and at least find out what the roadblocks are.

@DNA2RNA
Copy link
Author

DNA2RNA commented Jul 1, 2020

Hi Matt and John,

Thank you for the information. I see the RDKit issue you mention. As for OpenMM (and ParmEd), I managed to get them working on a Summit-style cluster with these Conda installations:

conda install https://anaconda.org/omnia-dev/openmm/7.4.0/download/linux-ppc64le/openmm-7.4.0-py37_cuda101_0.tar.bz2
conda install -c jchodera parmed

It looks like I will be able to pre-parametrize our ligands on an x86 system then move them over to ppc64le systems using the magic of serialization.

As a side note, I'd just like to say that the work you folks are doing is absolutely outstanding :)

Thank you all again,
Gabriel

@jchodera
Copy link
Member

jchodera commented Jul 1, 2020

I had built those manually using recipes I created for building on Summit:
https://github.com/inspiremd/conda-recipes-summit

Now that conda-forge has a mechanism for building for ppc64le and arm, however, we're very much hoping to get this working ASAP.

@mattwthompson
Copy link
Member

Opened an issue on the RKDit feedstock: conda-forge/rdkit-feedstock#52

@jaimergp
Copy link
Contributor

jaimergp commented Jul 3, 2020

You can track the progress for ppc64le and aarch CF packages in this PR:
conda-forge/openmm-feedstock#23

Note these are not released, so you would need to run conda build recipe on that repo on Summit to manually compile it yourself.

There are some issues with the CPU platform on ppc64le if it's compiled with GCC 8.x, but GCC 7.x works fine. Other platforms are not affected.

@jaimergp
Copy link
Contributor

Updates on this:

Having OpenMM on conda-forge is still ongoing, but we will get there. Is any other critical dependency I am missing?

@mattwthompson
Copy link
Member

This is now a bit stale - this probably works now given the improvements to upstream packaging, but it still might not.

If this is still desired and it does not work, please re-open.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

4 participants