George S

Hello - the same error described in the original bug was fixed in #1416, which is included in the most recent Torch-TRT release (1.3.0). I do not have access to...

Hello - I have tested out the sample script with the latest release of Torch-TRT (1.3.0), and both of the previously-failing compilations are now succeeding. Please let me know if...

> @gs-olive / @bowang007 How does this relate to index_put_? I don't think this specific case is directly related to `index_put_` other than that both `append` and `index_put_` are in-place...

Windows build is now successful with this branch: [`windows_CI`](https://github.com/pytorch/TensorRT/tree/windows_CI), however the smoke-test fails due to a missing `.dll` file, [as here](https://github.com/pytorch/TensorRT/actions/runs/7280075715/job/19864883942?pr=2502)

@HolyWu - thank you for the suggestion on this. I will try out the manual copy in the smoke test to get the CI tests running in the meantime.

### Additional Ideas + Notes - Shard compileable subgraphs based on estimated memory cost during compilation - Better estimate the necessary workspace size

> would you ever want to set `construct_live=False` in the compile path `construct_live=False` would specify offloading the Engines to CPU whereas `construct_live=True` would be the current behavior. If one doesn't...

Hello - thanks for the report. Could you share the output of debug mode or the line in Torch-TRT where the error occurs? Additionally, I noticed the `torch.save(compiled, "compiled.ts")` call,...

The Dynamo paths both use a truncation mechanism for int64 inputs which takes effect prior to the `TRTInterpreter` to avoid that issue. Does Int32 not work for the test cases?

Ok got it - could the test then mimic the scheme used here: https://github.com/pytorch/TensorRT/blob/16c031349c6a1af5a8408a817f2ef8542aa6f176/tests/py/dynamo/backend/test_backend_compiler.py#L193 This uses `torch.compile` directly to allow the 64-bit repair code to run.