viiccwen opened a new issue, #920: URL: https://github.com/apache/mahout/issues/920
### Summary In `qdp/qdp-python/src/lib.rs`, the CUDA tensor device is validated in two places: - Via PyTorch APIs in `validate_cuda_tensor_for_encoding(...)`, which checks that the tensor device matches `QdpEngine`’s device. - Via DLPack metadata inside `extract_dlpack_tensor(...)`, which reads `dl_tensor.device.device_id` from the `DLManagedTensor`. Currently, there is no check that these two views of the device ID are consistent. If, for any reason (framework bug, interop change, or misuse), the PyTorch-validated device ID diverges from the DLPack `device_id`, this could lead to hard‑to‑debug failures (e.g., using memory from the wrong device). reference: [comment](https://github.com/apache/mahout/pull/814#discussion_r2719818904) -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
