NVIDIA / nvshmemLinks
NVIDIA NVSHMEM is a parallel programming interface for NVIDIA GPUs based on OpenSHMEM. NVSHMEM can significantly reduce multi-process communication and coordination overheads by allowing programmers to perform one-sided communication from within CUDA kernels and on CUDA streams.
☆415Updated last month
Alternatives and similar repositories for nvshmem
Users that are interested in nvshmem are comparing it to the libraries listed below
Sorting:
- AMD RAD's multi-GPU Triton-based framework for seamless multi-GPU programming☆140Updated this week
- torchcomms: a modern PyTorch communications API☆309Updated this week
- Perplexity GPU Kernels☆539Updated last month
- A lightweight design for computation-communication overlap.☆196Updated 2 months ago
- Helpful kernel tutorials and examples for tile-based GPU programming☆456Updated this week
- NVSHMEM‑Tutorial: Build a DeepEP‑like GPU Buffer☆147Updated 3 months ago
- Github mirror of trition-lang/triton repo.☆109Updated this week
- MSCCL++: A GPU-driven communication stack for scalable AI applications☆444Updated last week
- Open ABI and FFI for Machine Learning Systems☆256Updated this week
- ☆151Updated 11 months ago
- Accelerating MoE with IO and Tile-aware Optimizations☆351Updated this week
- ☆332Updated this week
- DeeperGEMM: crazy optimized version☆73Updated 7 months ago
- ⚡️Write HGEMM from scratch using Tensor Cores with WMMA, MMA and CuTe API, Achieve Peak⚡️ Performance.☆138Updated 7 months ago
- Tritonbench is a collection of PyTorch custom operators with example inputs to measure their performance.☆301Updated this week
- Allow torch tensor memory to be released and resumed later☆191Updated 3 weeks ago
- Low overhead tracing library and trace visualizer for pipelined CUDA kernels☆127Updated 3 weeks ago
- This repository contains companion software for the Colfax Research paper "Categorical Foundations for CuTe Layouts".☆83Updated 3 months ago
- Perplexity open source garden for inference technology☆306Updated 2 weeks ago
- ☆69Updated 6 months ago
- ☆99Updated last year
- TritonParse: A Compiler Tracer, Visualizer, and Reproducer for Triton Kernels☆178Updated this week
- Thunder Research Group's Collective Communication Library☆43Updated 5 months ago
- ☆65Updated 7 months ago
- We invite you to visit and follow our new repository at https://github.com/microsoft/TileFusion. TiledCUDA is a highly efficient kernel …☆190Updated 10 months ago
- Ship correct and fast LLM kernels to PyTorch☆126Updated this week
- ☆52Updated 7 months ago
- extensible collectives library in triton☆91Updated 8 months ago
- Aims to implement dual-port and multi-qp solutions in deepEP ibrc transport☆72Updated 7 months ago
- A NCCL extension library, designed to efficiently offload GPU memory allocated by the NCCL communication library.☆66Updated last week