Repository navigation

Add fixed-size segment env overloads for DeviceSegmentedReduce - #8097

Merged

gonidelis merged 13 commits into

NVIDIA:mainfrom

gonidelis:seg_red_fixedsize_env

Apr 14, 2026

Merged

Add fixed-size segment env overloads for DeviceSegmentedReduce#8097
gonidelis merged 13 commits into
NVIDIA:mainfrom
gonidelis:seg_red_fixedsize_env

Commits on Apr 13, 2026
- Add fixed-size segment env overloads for DeviceSegmentedReduce
  gonidelis
  committed
  700eaef
- Fix critical issues plus<> specilization
  gonidelis
  committed
  b0784a8
- feedback address
  gonidelis
  committed
  55b86ba
- Refactor
  
  bernhardmgruber
  authored and
  gonidelis
  committed
  0ade842
- More refactoring
  
  bernhardmgruber
  authored and
  gonidelis
  committed
  12874fa
- Refactor: unify fixed-size ArgMin/ArgMax dispatch into a single function
  gonidelis
  committed
  091662c
- Make using offset_t = int explicit
  gonidelis
  committed
  daa85bf
- Substitute cudaStream_t with cuda::stream and use always stream.sync() but outside of the literalinclude guards
  gonidelis
  committed
  e65cf05
- Guard Sum with sfinae
  gonidelis
  committed
  f82642c
- Sfinae operator<<(os,...) for cuda::std::pairs
  gonidelis
  committed
  ea6cac4
- Do not specialize std::plus<>
  gonidelis
  committed
  3c44079
Commits on Apr 14, 2026
- Refine operator<<(os, pair<T, U) constraint
  bernhardmgruber
  committed
  548f40e
- Fix clang-tidy
  bernhardmgruber
  committed
  c7489c8