Code

TL/MLX5: various optimizations#1012

Code

Closed

samnordmann wants to merge 9 commits intoopenucx:masteropenucx/ucc:masterfrom samnordmann:tl/mlx5/pr/main_clean_historysamnordmann/ucc:tl/mlx5/pr/main_clean_historyCopy head branch name to clipboard

+384-181

Commits

Commits on Dec 18, 2024

• TL/MLX5: implement knomial fanin

  Show description for 29416ff
  samnordmann

  committed
  29416ff

  Copy full SHA for 29416ff

Commits on Dec 30, 2024

• TL/MLX5: configurable batch_size, ser, pollings

  Show description for 4bbd39a
  samnordmann

  committed
  4bbd39a

  Copy full SHA for 4bbd39a

• TL/MLX5: BFS iter, visit nodes before blocks

  Show description for 5f33297
  samnordmann

  committed
  5f33297

  Copy full SHA for 5f33297

• TL/MLX5: support rectangular blocks

  Show description for 5917644
  samnordmann

  committed
  5917644

  Copy full SHA for 5917644

• TL/MLX5: clean and fix after rebase

  Show description for 0bf47f8
  samnordmann

  committed
  0bf47f8

  Copy full SHA for 0bf47f8

• CODESTYLE: fix alignments and minor comments

  samnordmann

  committed
  6cdea8d

  Copy full SHA for 6cdea8d

• CODESTYLE: git-clang-format

  samnordmann

  committed
  0e39489

  Copy full SHA for 0e39489

• REVIEW: add assert to fix clang-tidy

  samnordmann

  committed
  8312300

  Copy full SHA for 8312300

Commits on Jan 2, 2025

• REVIEW: use PTR_OFFSET for dm ptr arith

  samnordmann

  committed
  4bd5995

  Copy full SHA for 4bd5995