ymwangg opened a new pull request #5902:
URL: https://github.com/apache/incubator-tvm/pull/5902
The following code will fail when using CUDA 11 and causes test cases of
`nvptx` target to fail. This PR fixed this issue.
```
tvm.contrib.nvcc.find_libdevice_path('7.0')
ValueError: Traceback (most recent call last):
[bt] (1) /home/ubuntu/src/tvm/build/libtvm.so(TVMFuncCall+0x65)
[0x7fc26a6064b5]
[bt] (0) /home/ubuntu/src/tvm/build/libtvm.so(+0xe10dbb) [0x7fc26a602dbb]
File "/home/ubuntu/src/tvm/python/tvm/_ffi/_ctypes/packed_func.py", line
78, in cfun
rv = local_pyfunc(*pyargs)
File "/home/ubuntu/src/tvm/python/tvm/contrib/nvcc.py", line 182, in
find_libdevice_path
ver = int(fn.split(".")[-3].split("_")[-1])
ValueError: invalid literal for int() with base 10: 'libdevice'
```
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
