Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[lang] Warn about non-contiguous gradient tensors #8450

Merged
merged 2 commits into from
Dec 27, 2023
Merged

[lang] Warn about non-contiguous gradient tensors #8450

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1eb57a2
Warn about non-contiguous gradient kernels
bobcao3 Dec 24, 2023
67fc59e
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot] Dec 24, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 10 additions & 0 deletions python/taichi/lang/kernel_impl.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -763,6 +763,16 @@ def call_back():
if v.requires_grad and v.grad is None:
v.grad = torch.zeros_like(v)

if v.requires_grad:
if not isinstance(v.grad, torch.Tensor):
raise ValueError(
f"Expecting torch.Tensor for gradient tensor, but getting {v.grad.__class__.__name__} instead"
)
if not v.grad.is_contiguous():
raise ValueError(
"Non contiguous gradient tensors are not supported, please call tensor.grad.contiguous() before passing it into taichi kernel."
)

tmp = v
if (str(v.device) != "cpu") and not (
str(v.device).startswith("cuda") and taichi_arch == _ti_core.Arch.cuda
Expand Down
Loading