NVIDIA / NVIDIA/apex

[Feature Request] Support for linux-ppc64le

Open
#273 8 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
9k
Forks
1.5k
Avg merge
2d 4h
Merged PRs (30d)
3

Description

Hi, I'm a trying to use apex on linux-ppc64le machines. There is no error when compile and install. However, it still falls back to python mode with the error:

Warning:  multi_tensor_applier fused unscale kernel is unavailable, possibly because apex was installed without --cuda_ext --cpp_ext. Using Python fallback.  Original ImportError was: ImportError('/home/xxx/.conda/envs/pt/lib/python3.6/site-packages/amp_C.cpython-36m-powerpc64le-linux-gnu.so: undefined symbol: _ZN3c105ErrorC1ENS_14SourceLocationERKSs',)
Warning:  apex was installed without --cpp_ext.  Falling back to Python flatten and unflatten.

Thanks!

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reproducing the linux-ppc64le installation with --cuda_ext --cpp_ext and inspect the reported amp_C ImportError and undefined PyTorch symbol. The issue names no source file or test, so locate the extension build and platform handling first; done means the CUDA and C++ extensions load on linux-ppc64le without falling back to Python mode.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp, python
Domain
machine-learning
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.