ByteDance-Seed/Triton-distributed

DeepEP: an efficient expert-parallel communication library

Development repository for the Triton language and compiler

A fast communication-overlapping library for tensor/expert parallelism on GPUs.

cuTile is a programming model for writing parallel kernels for NVIDIA GPUs

DeepGEMM: clean and efficient BLAS kernel library on GPU

UCCL is an efficient communication library for GPUs, covering collectives, P2P (e.g., KV cache transfer, RL weight transfer), and EP (e.g., GPU-driven)

Optimized primitives for collective multi-GPU communication

Automatically Discovering Fast Parallelization Strategies for Distributed Deep Neural Network Training

Open-source CUDA, Triton and HIP compiler targeting multiple GPU and CPU architectures.