Skip to content
This repository has been archived by the owner on Sep 21, 2019. It is now read-only.

include<cuda.h> in thread_local #11

Open
iotamudelta opened this issue Jul 24, 2018 · 4 comments
Open

include<cuda.h> in thread_local #11

iotamudelta opened this issue Jul 24, 2018 · 4 comments

Comments

@iotamudelta
Copy link
Contributor

#include <cuda.h>

Needs to be fixed.

@dagamayank
Copy link

@sriharikarnam @NEELMCW - Can you please look at it ASAP.

@iotamudelta
Copy link
Contributor Author

#include <cuda.h>

And there is another one. Please fix.

@sriharikarnam
Copy link
Contributor

@iotamudelta
We use branch https://github.com/ROCmSoftwarePlatform/cub-hip/tree/hip_port_1.7.3 for HIP ports. Please check if your issue is fixed with above branch.

@iotamudelta
Copy link
Contributor Author

@sriharikarnam Caffe2/PyTorch have used 1.7.4 in the past and now want to use master since that branch got merged. Please simply remove the CUDA imports - I can also make a PR for that.

If you look at the branch you reference, it still has the #include<cuda.h> in thread_load.cuh...

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

No branches or pull requests

3 participants