You are not logged in.
There was a big CUDA update recently, and since then I cannot run Python3, CUDA, Tensorflow.
The error I get is like the following one:
2019-04-05 16:22:16.819844: I tensorflow/stream_executor/dso_loader.cc:142] Couldn't open CUDA library libcublas.so.10.1. LD_LIBRARY_PATH:
2019-04-05 16:22:16.819876: F tensorflow/stream_executor/lib/statusor.cc:34] Attempting to fetch value instead of handling error Failed precondition: could not dlopen DSO: libcublas.so.10.1; dlerror: libcublas.so.10.1: cannot open shared object file: No such file or directory
Aborted (core dumped)
Last edited by joanmanel (2019-04-05 16:27:40)
Offline
I have the same problem.
cuda-10.0.130-2 with python-tensorflow-opt-cuda-1.13.1-2 work fine.
when those upgraded to cuda-10.1.105-6 and python-tensorflow-opt-cuda-1.13.1-4, problem happened.
ldconfig -p | grep libcublas
libcublasLt.so.10 (libc6,x86-64) => /opt/cuda/lib64/libcublasLt.so.10
libcublasLt.so (libc6,x86-64) => /opt/cuda/lib64/libcublasLt.so
libcublas.so.10 (libc6,x86-64) => /opt/cuda/lib64/libcublas.so.10
libcublas.so (libc6,x86-64) => /opt/cuda/lib64/libcublas.so
Last edited by code2501 (2019-04-06 04:26:49)
Offline
For some reasons ldconfig doesn't create link to libcublas.so.10.1
ldconfig -v | grep libcublas
libcublas.so.10 -> libcublas.so.10.1.0.105
libcublasLt.so.10 -> libcublasLt.so.10.1.0.105
My temporal workaround is passing it with LD_LIBRARY_PATH=/opt/cuda/lib64
Offline
Same problem here, on two different systems I am supporting.
Edit: https://bugs.archlinux.org/task/62287
Last edited by TheFiddler (2019-04-09 17:31:46)
The Open Toolkit: OpenGL, OpenAL, OpenCL and Vulkan for Mono/.Net.
Offline
The issue is upstream with CUDA. It is now manually patched in cuda-10.1.105-11:
$ ldconfig -v | grep cublas
libcublas.so.10 -> libcublas.so.10.1.0.105
libcublasLt.so.10 -> libcublasLt.so.10.1.0.105
libcublasLt.so.10.1 -> libcublasLt.so.10.1
libcublas.so.10.1 -> libcublas.so.10.1
Offline