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

Conda package #409

Open
mimre25 opened this issue Jul 31, 2019 · 9 comments
Open

Conda package #409

mimre25 opened this issue Jul 31, 2019 · 9 comments

Comments

@mimre25
Copy link

mimre25 commented Jul 31, 2019

Is there a conda package for this or can I only install it via pip?

@ptrblck
Copy link
Contributor

ptrblck commented Aug 7, 2019

We are currently in the process of releasing a conda package.
The pip package is not related to NVIDIA/apex, but refers to this package.

@mimre25
Copy link
Author

mimre25 commented Aug 7, 2019

Oh! My bad, I thought it's the pip package since your installation steps mention pip.

Any chance that you know a rough release date of the conda package?

@miguelvr
Copy link

+1 on the conda package, would make things much easier.

@ttpro1995
Copy link

+1, really need conda package to solve the CUDA dependency mess

@ttpro1995
Copy link

with pip install, I cannot use cudatoolkit from conda (I don't have nvcc and seperated CUDA installation because it is conflict with versioning, and so much bug, trouble going)

So, I am urgent to have CUDA for full feature apex support with CUDA

@jramapuram
Copy link

Any update on this? There are numerous issues related to building apex which could easily be avoided by a conda package that tightly couples:

pytorch <--> apex <--> cudatoolkit

This would make life a whole lot easier for a lot of folks.

@saverymax
Copy link

I would also be interested to know about any updates.

@tstandley
Copy link

Pytorch 1.6 will have native mixed precision training, and I assume the native sync-batch-norm will be supported.

@mcarilli
Copy link
Contributor

We decided to focus on getting the most important utilities from apex into upstream, so no separate installs are needed. Pytorch 1.6 will have native mixed precision support with a better API.

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

No branches or pull requests

8 participants