Various operators not currently supported on the ocl backend.

[ckuethe]

While trying to see about using OpenCL to make dSGP4 go faster on my laptop with just an integrated GPU, I came across at least the following unsupported operators:

• aten::all.all_out
• aten::any.all_out
• aten::atan2.out
• aten::cos.out
• aten::fmod.Tensor_out
• aten::linspace.out
• aten::pow.Tensor_Tensor_out
• aten::remainder.Tensor_out
• aten::sin.out
• aten::where.self

The examples gave a good example of how to use this library and were sufficient to trigger the unsupported operator warnings: https://esa.github.io/dSGP4/notebooks/tle_propagation.html

Using PyTorch 2.8.0 with device #0:Intel(R) Graphics [0x7d45] on Intel(R) OpenCL Graphics