A RuntimeError (Size mismatch between request values and groupby keys) occurs when using the GPU engine #25575
Description
Checks
- I have checked that this issue has not already been reported.
- I have confirmed this bug exists on the latest version of Polars.
Reproducible example
import polars as pl
ctx = pl.SQLContext()
t1 = pl.LazyFrame(
{
"c0": [True, False],
}
)
ctx.register("t1", t1)
result = t1.group_by("c0").agg(pl.lit(1).is_not_null())
print("CPU Result:")
print(result.collect())
print("GPU Result:")
print(result.collect(engine=pl.GPUEngine(raise_on_fail=True)))Log output
CPU Result:
shape: (2, 2)
┌───────┬─────────┐
│ c0 ┆ literal │
│ --- ┆ --- │
│ bool ┆ bool │
╞═══════╪═════════╡
│ false ┆ true │
│ true ┆ true │
└───────┴─────────┘
GPU Result:
Traceback (most recent call last):
File "/xxx/bug.py", line 19, in <module>
print(result.collect(engine=pl.GPUEngine(raise_on_fail=True)))
File "/xxx/anaconda3/envs/polars/lib/python3.10/site-packages/polars/_utils/deprecation.py", line 97, in wrapper
return function(*args, **kwargs)
File "/xxx/anaconda3/envs/polars/lib/python3.10/site-packages/polars/lazyframe/opt_flags.py", line 328, in wrapper
return function(*args, **kwargs)
File "/xxx/anaconda3/envs/polars/lib/python3.10/site-packages/polars/lazyframe/frame.py", line 2415, in collect
return wrap_df(ldf.collect(engine, callback))
File "/xxx/anaconda3/envs/polars/lib/python3.10/site-packages/polars/_utils/scan.py", line 27, in _execute_from_rust
return function(with_columns, *args)
File "/xxx/anaconda3/envs/polars/lib/python3.10/site-packages/cudf_polars/callback.py", line 263, in _callback
return evaluate_streaming(ir, config_options)
File "/xxx/anaconda3/envs/polars/lib/python3.10/site-packages/cudf_polars/experimental/parallel.py", line 287, in evaluate_streaming
return get_scheduler(config_options)(graph, key).to_polars()
File "/xxx/anaconda3/envs/polars/lib/python3.10/site-packages/cudf_polars/experimental/scheduler.py", line 149, in synchronous_scheduler
cache[k] = _execute_task(graph[k], cache)
File "/xxx/anaconda3/envs/polars/lib/python3.10/site-packages/cudf_polars/experimental/scheduler.py", line 51, in _execute_task
return arg[0](*(_execute_task(a, cache) for a in arg[1:]))
File "/xxx/anaconda3/envs/polars/lib/python3.10/site-packages/cudf_polars/dsl/ir.py", line 1814, in do_evaluate
group_keys, raw_tables = grouper.aggregate(requests, stream=df.stream)
File "pylibcudf/groupby.pyx", line 163, in pylibcudf.groupby.GroupBy.aggregate
File "pylibcudf/groupby.pyx", line 198, in pylibcudf.groupby.GroupBy.aggregate
RuntimeError: CUDF failure at: /tmp/conda-bld-output/bld/rattler-build_libcudf/work/cpp/src/groupby/groupby.cu:228: Size mismatch between request values and groupby keys.Issue description
A RuntimeError will occur when combining .agg() with .is_not_null() and using the GPU engine.
Expected behavior
No error will occur when using the GPU engine.
The output is consistent when using the CPU and GPU engines.
Installed versions
--------Version info---------
Polars: 1.34.0
Index type: UInt32
Platform: Linux-6.14.0-35-generic-x86_64-with-glibc2.39
Python: 3.10.19 (main, Oct 21 2025, 16:43:05) [GCC 11.2.0]
Runtime: rt32
----Optional dependencies----
Azure CLI <not installed>
adbc_driver_manager <not installed>
altair <not installed>
azure.identity <not installed>
boto3 <not installed>
cloudpickle <not installed>
connectorx <not installed>
deltalake <not installed>
fastexcel <not installed>
fsspec 2025.10.0
gevent <not installed>
google.auth <not installed>
great_tables <not installed>
matplotlib <not installed>
numpy 1.26.4
openpyxl <not installed>
pandas 2.3.3
polars_cloud <not installed>
pyarrow <not installed>
pydantic <not installed>
pyiceberg <not installed>
sqlalchemy <not installed>
torch <not installed>
xlsx2csv <not installed>
xlsxwriter <not installed>
cuda-bindings 12.9.4
cuda-pathfinder 1.3.2
cuda-python 12.9.4
cudf-polars 26.2.0a104
nvidia-ml-py 13.580.82
nvtx 0.2.13