项目文件夹

文件
nv-dlasalle 1f2e696080 Prevent users from attempting to pin PyTorch non-contiguous tensors or views only encompassing part of tensor. (#3992)
* Disable pinning non-contiguous memory

* Prevent views from being converted for write

* Fix linting

* Add unit tests

* Improve error message for users

* Switch to pytest function

* exclude mxnet and tensorflow from inplace pinning

* Add skip

* Restrict to pytorch backend

* Use backend to retrieve device

* Fix capitalization in decorator

Co-authored-by: Quan (Andy) Gan <coin2028@hotmail.com>
2022-05-16 14:49:36 -07:00

23 行
622 B
Python

import backend as F
import dgl
import pytest
@pytest.mark.skipif(F._default_context_str == 'cpu', reason="Need gpu for this test")
def test_pin_unpin():
t = F.arange(0, 100, dtype=F.int64, ctx=F.cpu())
assert not F.is_pinned(t)
if F.backend_name == 'pytorch':
dgl.utils.pin_memory_inplace(t)
assert F.is_pinned(t)
dgl.utils.unpin_memory_inplace(t)
assert not F.is_pinned(t)
else:
with pytest.raises(dgl.DGLError):
# tensorflow and mxnet should throw an erro
dgl.utils.pin_memory_inplace(t)
if __name__ == "__main__":
test_pin_unpin()