GitHub Trending Cuda This Month

[ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

Cuda3,652483Built by

DeepEP: an efficient expert-parallel communication library

Cuda10,0161,389Built by

DeepGEMM: clean and efficient BLAS kernel library on GPU

Cuda7,7001,182Built by

RTP-LLM: Alibaba's high-performance LLM inference engine for diverse applications.

Cuda1,309259Built by

cuVS - a library for vector search and clustering on the GPU

Cuda834222Built by

NCCL Tests

Cuda1,623398Built by

GPU accelerated decision optimization

Cuda1,006219Built by

Mirage Persistent Kernel: Compiling LLMs into a MegaKernel

Cuda2,431239Built by

how to optimize some algorithm in cuda.

Cuda3,204290Built by

Instant neural graphics primitives: lightning fast NeRF and more

Cuda17,5222,066Built by

LLM training in simple, raw C/CUDA

Cuda30,8243,734Built by

Tile primitives for speedy kernels

Cuda3,633319Built by