Issues

• [FEA] Investigate CUB tunings for whole column and segmented reductions

  feature requestlibcudfPerformance
  NVIDIA/cudf#24646 · Filter by author wence- opened on Oct 16, 2026
  1

• [FEA] Share the row/field walk between the CSV type inference and parse kernels

  NVIDIA/cudf#24642 · Filter by author thirtiseven opened on Oct 16, 2026

• Support KX NYSE TAQ benchmark in cudf-polars fully on GPU

  cudf-polarsfeature request
  NVIDIA/cudf#24634 · Filter by author pentschev opened on Oct 15, 2026

• [BUG] JAVA: ColumnVector constructor leaks the column if a call inside its constructor throws

  ? - Needs Triagebug
  NVIDIA/cudf#24631 · Filter by author pmattione-nvidia opened on Oct 15, 2026
  

• [FEA] Support hybrid scan row group pass construction for sparse payload I/O mode

  cuIOfeature requestlibcudf
  NVIDIA/cudf#24620 · Filter by author mhaseeb123 opened on Oct 15, 2026
  

• [FEA] libcudf_streaming: fast path in unpack_and_concat for a single partition

  feature requestlibcudfPerformance
  NVIDIA/cudf#24619 · Filter by author nirandaperera opened on Oct 14, 2026

• [Story] Fuzz testing of the libcudf Parquet writer and readers

  ? - Needs TriagecuIOfeature requestlibcudftests
  NVIDIA/cudf#24604 · Filter by author mattgara opened on Oct 14, 2026

• [FEA] Improve performance of streaming group by

  feature requestlibcudfPerformance
  NVIDIA/cudf#24600 · Filter by author scanhex12 opened on Oct 14, 2026
  1

• [Story] Kratos Requirements for Presto GPU

  NVIDIA/cudf#24556 · Filter by author singhmanas1 opened on Oct 12, 2026

• [FEA] Support cudf-polars benchmarking with the polars.SQLContext

  cudf-polarsfeature request
  NVIDIA/cudf#24555 · Filter by author mroeschke opened on Oct 12, 2026
  1

• [FEA] Modify assert_gpu_result_equal to assert .flags equality

  cudf-polarsfeature request
  NVIDIA/cudf#24540 · Filter by author mroeschke opened on Oct 9, 2026

• [BUG] Correctly size retained composition tables in unicode_normalizer constructor

  ? - Needs Triagebuglibcudf
  NVIDIA/cudf#24529 · Filter by author KyleFromNVIDIA opened on Oct 9, 2026
  1

  

• [BUG]: cudf.read_parquet with columns=[] gives a shape-(0, 0) dataframe

  bugcudf.pandas
  NVIDIA/cudf#24523 · Filter by author TomAugspurger opened on Oct 9, 2026
  3

• [BUG] floor_datetimes/ceil_datetimes/round_datetimes silently overflow

  buglibcudf
  NVIDIA/cudf#24521 · Filter by author pentschev opened on Oct 9, 2026

• [BUG] import cudf fails with numpy 2.5

  ? - Needs Triagebug
  NVIDIA/cudf#24519 · Filter by author Intron7 opened on Oct 9, 2026
  3

• [FEA] Bump Polars minimum verison to 2.0

  cudf-polarsfeature requestimprovement
  NVIDIA/cudf#24513 · Filter by author mroeschke opened on Oct 8, 2026

• [FEA] Expose programmatic JIT compilation and kernel dispatch statistics

  feature requestimprovementPerformanceVelox
  Enhancement

  NVIDIA/cudf#24500 · Filter by author lamarrr opened on Oct 8, 2026 · libcudf JIT Enhancements

• [FEA] Use polars.Config.set_engine_affinity in cudf-polars Polars tests

  cudf-polarsfeature request
  NVIDIA/cudf#24478 · Filter by author mroeschke opened on Oct 6, 2026

• [FEA] Partial unpack: per-column byte ranges and column_view construction from packed_metadata_view

  feature requestPerformanceproposal
  NVIDIA/cudf#24477 · Filter by author nirandaperera opened on Oct 6, 2026
  1

• Synchronize cudf-polars testing with upstream Polars

  cudf-polarstests
  NVIDIA/cudf#24470 · Filter by author vyasr opened on Oct 6, 2026
  2

• [FEA] Reduce AST cold JIT compilation time with nvJitLink and type-erased row operators

  ? - Needs Triagefeature request
  Enhancement

  NVIDIA/cudf#24464 · Filter by author lamarrr opened on Oct 6, 2026 · libcudf JIT Enhancements

• [FEA] Improve cudf_polars distributed sort algorithm

  cudf-polarsfeature requestPerformance
  NVIDIA/cudf#24463 · Filter by author nirandaperera opened on Oct 6, 2026
  1

• [FEA] Support GROUPS frames in rolling windows

  feature requestJavalibcudf
  NVIDIA/cudf#24462 · Filter by author piotrrzysko opened on Oct 6, 2026

• [FEA] Per-row frame offsets for grouped rolling windows

  bugfeature requestJavalibcudf
  NVIDIA/cudf#24456 · Filter by author piotrrzysko opened on Oct 6, 2026
  1

• [FEA] Refactor binary and unary operations to use transform_kernel

  feature requestlibcudf
  Enhancement

  NVIDIA/cudf#24453 · Filter by author lamarrr opened on Oct 6, 2026 · libcudf JIT Enhancements