-
Notifications
You must be signed in to change notification settings - Fork 884
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[BUG] GPU-accelerated file reads fail on WSL2 #9861
Comments
This issue has been labeled |
To provide some additional context here, the generated cufile.log:
|
Same problem here when running import cudf
df = cudf.DataFrame({"A": [1, 2, 3], "B": [4, 5, 6]}) got Traceback (most recent call last):
File "/home/sgiraldo/miniconda3/envs/rapids-22.02/lib/python3.8/site-packages/cudf/core/column/column.py", line 2051, in as_column
memoryview(arbitrary), dtype=dtype, nan_as_null=nan_as_null
TypeError: memoryview: a bytes-like object is required, not 'list'
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/home/sgiraldo/tesis/EarthQuakeGNN/scripts/cudf_test.py", line 12, in <module>
df = cudf.DataFrame({"A": [1, 2, 3], "B": [4, 5, 6]})
File "/home/sgiraldo/miniconda3/envs/rapids-22.02/lib/python3.8/contextlib.py", line 75, in inner
return func(*args, **kwds)
File "/home/sgiraldo/miniconda3/envs/rapids-22.02/lib/python3.8/site-packages/cudf/core/dataframe.py", line 622, in __init__
self._init_from_dict_like(
File "/home/sgiraldo/miniconda3/envs/rapids-22.02/lib/python3.8/site-packages/cudf/core/dataframe.py", line 800, in _init_from_dict_like
data[col_name] = column.as_column(
File "/home/sgiraldo/miniconda3/envs/rapids-22.02/lib/python3.8/site-packages/cudf/core/column/column.py", line 2142, in as_column
data = as_column(
File "/home/sgiraldo/miniconda3/envs/rapids-22.02/lib/python3.8/site-packages/cudf/core/column/column.py", line 1816, in as_column
col = ColumnBase.from_arrow(arbitrary)
File "/home/sgiraldo/miniconda3/envs/rapids-22.02/lib/python3.8/site-packages/cudf/core/column/column.py", line 300, in from_arrow
result = libcudf.interop.from_arrow(data, data.column_names)[0]["None"]
File "cudf/_lib/interop.pyx", line 167, in cudf._lib.interop.from_arrow
RuntimeError: CUDA error encountered at: ../src/bitmask/null_mask.cu:93: 801 cudaErrorNotSupported operation not supported |
Getting this with README example, idk if the installation is broken import cudf, io, requests
from io import StringIO
url = "https://github.com/plotly/datasets/raw/master/tips.csv"
content = requests.get(url).content.decode('utf-8')
tips_df = cudf.read_csv(StringIO(content))
tips_df['tip_percentage'] = tips_df['tip'] / tips_df['total_bill'] * 100
# display average tip by dining party size
print(tips_df.groupby('size').tip_percentage.mean()) Traceback (most recent call last):
File "/home/sgiraldo/tesis/EarthQuakeGNN/scripts/cudf_test.py", line 24, in <module>
tips_df = cudf.read_csv(StringIO(content))
File "/home/sgiraldo/miniconda3/envs/rapids-22.02/lib/python3.8/contextlib.py", line 75, in inner
return func(*args, **kwds)
File "/home/sgiraldo/miniconda3/envs/rapids-22.02/lib/python3.8/site-packages/cudf/io/csv.py", line 73, in read_csv
return libcudf.csv.read_csv(
File "cudf/_lib/csv.pyx", line 430, in cudf._lib.csv.read_csv
MemoryError: std::bad_alloc: CUDA error at: /home/sgiraldo/miniconda3/envs/rapids-22.02/include/rmm/mr/device/cuda_memory_resource.hpp |
Looks like this no longer fails on the latest nightlies? Though I still do get a cufile.log:
Would be interested in pinpointing what exactly changed here, as I can still reproduce this failure using 22.02 on my setup. |
@cnewburn has cuFILE/GDS been tested on WSL2 ? |
It looks like this issue was due to a broken WSL2 CUDA toolkit installation? I uninstalled CTK apt packages and retried with stable 22.02 conda packages (including the conda CTK package) and things were resolved - this can be closed. cc @taureandyernv in case you have any additional context to provide here |
Describe the bug
When trying to read a dataframe from file on Windows Subsystem for Linux (WSL) using a GPU-accelerated reader method (e.g. all the reader methods but
to_feather
), the operation fails and crashes the Python kernel:Steps/Code to reproduce bug
Working with the latest cuDF source on WSL2 Ubuntu 20.04:
Expected behavior
A dataframe to successfully be loaded from the file(s).
Environment overview (please complete the following information)
Environment details
Click here to see environment details
Additional context
Related to #9859, which deals with writing dataframes to file.
The text was updated successfully, but these errors were encountered: