tccl

extensible collectives library in triton

built for CUDA MODE IRL Sep 21, 2024 by eric zhang, sage moore, clive chan

presentation slides: link

Name		Name	Last commit message	Last commit date
Latest commit History 3 Commits
.gitignore		.gitignore
README.md		README.md
__init__.py		__init__.py
plot.ipynb		plot.ipynb
presentation.pdf		presentation.pdf
tree.py		tree.py
triton_all_gather.py		triton_all_gather.py
triton_barrier.py		triton_barrier.py
triton_barrier2.py		triton_barrier2.py
triton_double_tree_allreduce.py		triton_double_tree_allreduce.py
triton_one_shot_all_reduce_no_128.py		triton_one_shot_all_reduce_no_128.py
triton_reduce_scatter.py		triton_reduce_scatter.py

Provide feedback