[QST] Jetson Support/Documentation

[SeeRich]

What is your question?

Does rapidsai run/work on Jetson platforms? Specifically the newer platforms such as jetson orin nano and orin nx?

I am testing on a Jetson Orin NX with Jetpack 6.2 and CUDA 12.6 (also tested with cuda 12.9 and cuda compat package). I believe my setup meets all the system requirements listed on installation docs.

I was able to install the packages via pip using:

pip install \
    --extra-index-url=https://pypi.nvidia.com \
    "cudf-cu12==25.4.*" "dask-cudf-cu12==25.4.*" "cuml-cu12==25.4.*" \
    "cugraph-cu12==25.4.*" "nx-cugraph-cu12==25.4.*" "cuspatial-cu12==25.4.*" \
    "cuproj-cu12==25.4.*" "cuxfilter-cu12==25.4.*" "cucim-cu12==25.4.*" \
    "pylibraft-cu12==25.4.*" "raft-dask-cu12==25.4.*" "cuvs-cu12==25.4.*" \
    "nx-cugraph-cu12==25.4.*"

I tried to run the simple example from the user guide:

import os

import cudf
import cupy as cp
import dask_cudf
import pandas as pd

cp.random.seed(12)

s = cudf.Series([1, 2, 3, None, 4])
print(s)

However, I get the following error:

Traceback (most recent call last):
  File "/home/orinnx/Development/test.py", line 11, in <module>
    print(s)
  File "/home/orinnx/.local/lib/python3.10/site-packages/cudf/utils/performance_tracking.py", line 51, in wrapper
    return func(*args, **kwargs)
  File "/home/orinnx/.local/lib/python3.10/site-packages/cudf/core/frame.py", line 1908, in __str__
    return repr(self)
  File "/home/orinnx/.local/lib/python3.10/site-packages/cudf/core/series.py", line 1488, in __repr__
    output = repr(preprocess._pandas_repr_compatible().to_pandas())
  File "/home/orinnx/.local/lib/python3.10/site-packages/cudf/core/indexed_frame.py", line 4431, in _pandas_repr_compatible
    result = super()._pandas_repr_compatible()
  File "/home/orinnx/.local/lib/python3.10/site-packages/cudf/core/frame.py", line 828, in _pandas_repr_compatible
    self._data._from_columns_like_self(columns, verify=False)
  File "/home/orinnx/.local/lib/python3.10/site-packages/cudf/core/column_accessor.py", line 204, in _from_columns_like_self
    data=dict(data),
  File "/home/orinnx/.local/lib/python3.10/site-packages/cudf/core/frame.py", line 826, in <genexpr>
    columns = (col._prep_pandas_compat_repr() for col in self._columns)
  File "/home/orinnx/.local/lib/python3.10/site-packages/cudf/core/column/column.py", line 625, in _prep_pandas_compat_repr
    return self.astype(CUDF_STRING_DTYPE).fillna(self._PANDAS_NA_REPR)
  File "/home/orinnx/.local/lib/python3.10/site-packages/cudf/core/column/column.py", line 1638, in astype
    result = self.as_string_column()
  File "/home/orinnx/.local/lib/python3.10/site-packages/cudf/core/column/numerical.py", line 358, in as_string_column
    conv_func(self.to_pylibcudf(mode="read"))
  File "convert_integers.pyx", line 54, in pylibcudf.strings.convert.convert_integers.from_integers
  File "convert_integers.pyx", line 75, in pylibcudf.strings.convert.convert_integers.from_integers
RuntimeError: after determining tmp storage requirements for exclusive_scan: cudaErrorInvalidDevice: invalid device ordinal