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

Attempting to fetch value instead of handling error Failed precondition: Could not load dynamic library 'libhip_hcc.so'; dlerror: libhip_hcc.so: cannot open shared object file: No such file or directory #1163

Closed
vuquocan1987 opened this issue Jun 24, 2020 · 7 comments

Comments

@vuquocan1987
Copy link

I installed the tensorflow_rocm with rocm stack on my computer, but when I tried to create a model I got such error.

My set up is:

ubuntu 18.04.04
rocm 3.5.1
tensorflow-rocm 2.2.0
gpu rx580
cpu intel 9100f

What is wrong?

@xuhuisheng
Copy link
Contributor

it seems that ldconfig cannot find the soft link of hip_hcc,
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/rocm/hip/lib
then it run well

Ubuntu 18.04.04
ROCm 3.5.1-34
tensorflow-rocm 2.2.0
Python 3.6.0 with anaconda

Hardware:
cpu I7 4770
gpu RX 580

@vuquocan1987
Copy link
Author

thank you very much, this does solve my problem. Do you know why though?

@ghost
Copy link

ghost commented Jun 25, 2020

Solves the problem partially but not fully

@xuhuisheng
Copy link
Contributor

thank you very much, this does solve my problem. Do you know why though?

At this moment, I dont know why caused this situation.
The libhip_hcc.so and related files are in the /opt/rocm/hip/lib directory, but even if I run sudo ldconfig, but tensorflow-rocm cannot find these .so files.

@papadako
Copy link

I am not sure either. If you delete libhiprtc.so then ldconfig will find libhip_hcc.so.
Check my post ROCm/tensorflow-upstream#1019 (comment) about another workaround

thank you very much, this does solve my problem. Do you know why though?

At this moment, I dont know why caused this situation.
The libhip_hcc.so and related files are in the /opt/rocm/hip/lib directory, but even if I run sudo ldconfig, but tensorflow-rocm cannot find these .so files.

@kk-1
Copy link

kk-1 commented Aug 21, 2020

Hello Folks,
I had the same problem.
I saw that the library libhip_hcc is moved to (as symlink!!!):
/usr/lib/aomp/lib

Here 2 ways:
1-) You can create symlink to "libamdhip64.so" in the PATH manually or
2-) You can install the aomp and set the LD_LIBRARY_PATH in your .profile like that (got some extras but just get the idea!):

export LD_LIBRARY_PATH="/usr/lib/aomp/lib:/usr/lib/aomp/lib64:/opt/rocm/lib:/opt/rocm/lib64:/opt/rocm/opencl/lib/x86_64:$LD_LIBRARY_PATH"

BTW I got:
openSUSE Tumbleweed: rpm based system
rocm 3.7.0
aomp 11.8

Cheers!!!

@ROCmSupport
Copy link

Hi @vuquocan1987 and all,
Thanks for reaching out.
This issue is fixed and no more observed with the recent ROCm release versions.
Recommend to try with 3.10.
Thank you.

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

5 participants