alexandrosstergiou / SoftPool

[ICCV 2021] Code for approximated exponential maximum pooling

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

cudaCheckError() failed : invalid device function

Cassie-CV opened this issue · comments

hi, ‘make install’ is succeed,but running ‘make test’ occurs some errors.
`cd test-files && python test.py
--- Initial checks for forward ---

Checking 1D CPU ...
PASSED
Checking 1D GPU ...
cudaCheckError() failed : invalid device function
Segmentation fault (core dumped)
Makefile:9: recipe for target 'test' failed
make: *** [test] Error 139
`

This seems like a CUDA or hardware related issue. Can you post some info about:
A. Your system (i.e. GPU model)
B. How you installed:

  • CUDA and cuDNN
  • PyTorch

Best,
Alex

This seems like a CUDA or hardware related issue. Can you post some info about:
A. Your system (i.e. GPU model)
B. How you installed:

  • CUDA and cuDNN
  • PyTorch

Best,
Alex

I found the same error here.
My environment info is:
GPU: 2080 Ti
CUDA: 10.0.130
cuDNN: 7.6.0
PyTorch: 1.6.0

Thank you.

Hi @rumsyx,

This sounds like an NVCC/CUDA runtime version mismatch for your system. Check #62 issue in PyTorch and ensure that your NVCC and CUDA runtime versions are correct.

Best,
Alex

may by your install more CUDA version.
so
make clean
CUDA_HOME=/mnt/lustre/share/cuda-10.2 make install

make clean
CUDA_HOME=/usr/local/cuda-10.2 make install
then can make test

Closing due to inactivity. Feel free to open a new issue for future problems with the installation process.