csarofeen / pytorch

Tensors and Dynamic neural networks in Python with strong GPU acceleration
http://pytorch.org
Other
26 stars 7 forks source link

Fix downcast_ptr != nullptr in RandomOp Record Functor #2547

Closed rdspring1 closed 1 year ago

rdspring1 commented 1 year ago
----------------------------------------------------------------------
ERROR: test_uniform (__main__.TestNvFuserFrontend)
----------------------------------------------------------------------
Traceback (most recent call last):
File "/home/rspring/workspace/nvf_pytorch/third_party/nvfuser/python_tests/test_python_frontend.py", line 935, in
test_uniform nvf_out, _ = self.exec_nvfuser(fusion_func, inputs)

File "/home/rspring/workspace/nvf_pytorch/third_party/nvfuser/python_tests/test_python_frontend.py", line 46, in
exec_nvfuser fusion_func(fd) 

File "/home/rspring/workspace/nvf_pytorch/nvfuser/__init__.py", line 10, in __exit__
self._finalize_definition() 

RuntimeError: downcast_ptr != nullptr INTERNAL ASSERT FAILED
at "/home/rspring/workspace/nvf_pytorch/third_party/nvfuser/csrc/utils.h":165, please report a bug to PyTorch.