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.

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

filtered_lrelu called with parameters that have no optimized CUDA kernel, using generic fallback #622

Open
ulucsahin opened this issue Oct 27, 2023 · 1 comment

Comments

@ulucsahin
Copy link

Describe the bug
A clear and concise description of what the bug is.

Output of _plugin.filtered_lrelu(x, fu, fd, b, si, up, down, px0, px1, py0, py1, sx, sy, gain, slope, clamp, flip_filter, write_signs) returns the return code -1, therefore it uses generic fallback. I want to use optimized kernels and not generic fallback. What might be the issue here for getting -1 return code?

I believe code enters

    if (!test_spec.exec)
    {
        // No kernel found - return empty tensors and indicate missing kernel with return code of -1.
        return std::make_tuple(torch::Tensor(), torch::Tensor(), -1);
    }

But I am not sure why this happens.

Desktop

  • OS: Ubuntu 16.04,
  • PyTorch version 1.9.1
  • CUDA toolkit version: CUDA 11.1
  • NVIDIA driver version: 510.47.03
  • GPU: Tesla T4
  • Docker: No
@ulucsahin
Copy link
Author

I disabled the code block in c++, and getting an issue as expected: internal error - CUDA kernel not found

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant