github rapidsai/cudf v25.12.00
on GitHub

one day ago

What's Changed

🚨 Breaking Changes

🐛 Bug Fixes

📖 Documentation

  • Add note that --rmm-async only affects distributed scheduler. by @bdice in #20129
  • Add profiling guide by @bdice in #20292
  • Find RMM before CCCL by @wence- in #20336
  • Use current system architecture in conda environment creation command by @bdice in #20500
  • Use uname -m instead of arch command by @bdice in #20502
  • Use RAPIDS_BRANCH file for documentation links by @bdice in #20494

🚀 New Features

  • Add memory resources to unary, transform, and filling modules by @vyasr in #20054
  • Add memory resources to binaryop, copying, and stream_compaction by @vyasr in #20059
  • Add memory resources to groupby, datetime, and lists modules by @vyasr in #20102
  • Add memory resources to search, reshape, and partitioning module by @vyasr in #20101
  • Add memory resources to rolling, sorting, and quantiles modules by @vyasr in #20099
  • [FEA] Implement JIT Filter for read_parquet by @lamarrr in #19831
  • Add memory resources to all nvtext APIs by @vyasr in #20119
  • Add memory resource to all strings modules by @vyasr in #20123
  • Add memory resources to reduce, column, column_factories, and contiguous_split by @vyasr in #20135
  • Add memory resources to I/O modules by @vyasr in #20136
  • Remove rounding from cudf java by @pmattione-nvidia in #20110
  • Add memory resources to replace, json, and hashing by @vyasr in #20150
  • Add support for maintain_order param in joins by @Matt711 in #17698
  • Add an example to inspect parquet files and dump row group and page level metadata information by @mhaseeb123 in #20117
  • Support forward/backward filling null values in a grouped window context by @Matt711 in #19907
  • Allow multiple calls to cudf::initialize and cudf::deinitialize by @vuule in #20111
  • Add remaining memory resources by @vyasr in #20197
  • Add memory resources to scalars by @vyasr in #20196
  • Add pylibcudf is_valid_reduce_aggregation API by @davidwendt in #20145
  • Support decimal literals in cudf-polars by @Matt711 in #20147
  • Support cum_sum(...).over(...) expressions in cudf-polars by @Matt711 in #19908
  • Passthrough unary ops through Parquet predicate pushdown by @mhaseeb123 in #20127
  • Implement ARGMIN and ARGMAX aggregations for reduction by @ttnghia in #20207
  • Skip decompression of pruned parquet pages by @mhaseeb123 in #20192
  • Add an example to demonstrate the use of next-gen parquet reader to read a parquet file with highly selective filters by @mhaseeb123 in #19469
  • Evaluate IS_NULL at row group and page level in Parquet filtering by @mhaseeb123 in #20144
  • [Java] Add optional native deps loader by @zpuller in #20414
  • Add cudf-polars + rapidsmpf CI check by @rjzamora in #20355
  • Add Python bindings for the hybrid scan reader by @vyasr in #20381
  • RapidsMPF streaming-engine translation by @rjzamora in #20161
  • [JNI] Use a read/write lock pattern in Rmm.class by @abellina in #20521
  • [Java] Supports output projection indices for contiguousSplitGroupsAndGenUniqKeys by @res-life in #20391
  • Support Series.at and Series.iat for pandas compatability by @Matt711 in #20529
  • Add COUNT_VALID aggregation support to groupby-scan by @davidwendt in #20531
  • Use RapidsMPF read_parquet in "rapidsmpf" runtime by @rjzamora in #20497
  • Support decimal128 SUM aggregation in hash-based groupby by @PointKernel in #20509
  • Add stream testing in pylibcudf by @vyasr in #20625

🛠️ Improvements

New Contributors

Full Changelog: v25.12.00a...v25.12.00

Check out latest releases or
releases around rapidsai/cudf v25.12.00

Don't miss a new cudf release

NewReleases is sending notifications on new releases.

Get notifications