NVIDIA/cudf

Is your feature request related to a problem? parquet::fetch_bloom_filters_to_device treats any size-zero range as “no filter” and skips it. Consequently, these chunks are never bloom-pruned. This behavior is correct but suboptimal: pruning is silently lost because the device caster conservatively returns membership true for empty spans in bloom_filter_reader.cu. The default read_parquet path already handles this case. read_bloom_filters substitutes a speculative 256-byte initial read, clamped to the datasource size, using speculative_read_size in: https://github.com/rapidsai/cudf/blob/3b3cea4d9669b9db1e35769ea727514eac3cb3e7/cpp/src/io/parquet/bloom_filter_reader.cu#L363-L366 Describe the solution you’d like Handle this case in fetch_bloom_filters_to_device_impl in cpp/src/io/parquet/io_utils/parquet_io_utils.cpp This function owns the datasources and can therefore clamp reads safely. The hybrid-scan metadata layer cannot do so because it has no access to the datasources. Phase

Report incorrect documentation Location of incorrect documentation cudf/source/user_guide/10min.ipynb Describe the problems or issues found in the documentation Currently this notebook (and other notebooks that use Dask) isn't being validated in our CI https://github.com/rapidsai/cudf/blob/62387458abe94670c386460a39c877c7530d544d/ci/test_notebooks.sh#L52 And it's likely that #20925 broke something in this notebook when run by-hand. Suggested fix for documentation Create 2, 10 minutes to [cuDF|Dask-cuDF] notebooks so we can at least validate a (probably commonly seen) introductory cuDF notebook Investigate other notebooks that might be running Dask and split them out into other notebooks if appropriate. (The CI currently only skips validating 10min.ipynb because of Dask usage)

Is your feature request related to a problem? Please describe. Regex features are available in cudf string APIs and nvtext APIs. grep is widely used utility. It would cool to try creating cudfgrep with cudf/nvtext and document the throughput. Describe the solution you'd like Create cudfgrep utility in our python codebase or a C++ example with libcudf APIs. Additional context Benchmark it without GDS (most common usecase). Benchmark on GPU in PCIE, and on Grace systems with fast C2C.