NVIDIA / NVIDIA/apex

change __shfl_down_sync() to __shfl_down

Open
#423 3 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

Hello I'm new commer in apex nvidia.

I supposed to install apex but my cuda version is 8.0 so cannot installed it.
and in my situation I cannot change my cuda version to 9.0 or over(because of my project doing now)

so I tried to install the apex using
**pip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./**

command but since I cannot use cuda 9.0 or over, error is occurred like this

**csrc/type_shim.h(170): error: identifier "__shfl_down_sync" is undefined

    1 error detected in the compilation of "/tmp/tmpxft_00000a0e_00000000-7_multi_tensor_scale_kernel.cpp1.ii".
    error: command u'/usr/local/cuda/bin/nvcc' failed with exit status 2**

so I changed shfl_down_sync to shfl_down to install apex but the error message is similar like before, only changed message is "__shfl_down" is undefined

how can I solve this problem?
my linux version is 16.04 and cuda is 8.0
I really appreciate your answer. 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 with csrc/type_shim.h around line 170 and reproduce the reported build using CUDA 8.0 and the listed pip command. Trace how the shuffle intrinsic is selected for the CUDA extension, then use the compilation result to define a compatible change and verify that the extension builds successfully on the stated environment.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp, linux, python, pytorch
Domain
build-system, machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.