flashinfer-ai / flashinfer
FlashInfer: Kernel Library for LLM Serving
See what the GitHub community is most excited about this week.
FlashInfer: Kernel Library for LLM Serving
Quantized Attention achieves speedup of 2-5x and 3-11x compared to FlashAttention and xformers, without lossing end-to-end metrics across language, image, and video models.
CUDA accelerated rasterization of gaussian splatting
LLM training in simple, raw C/CUDA
DeepEP: an efficient expert-parallel communication library
DeepGEMM: clean and efficient FP8 GEMM kernels with fine-grained scaling
GPU accelerated decision optimization
NCCL Tests
Tile primitives for speedy kernels
A collection of examples for the ROCm software stack
[ARCHIVED] Cooperative primitives for CUDA C++. See https://github.com/NVIDIA/cccl
cuVS - a library for vector search and clustering on the GPU
RAFT contains fundamental widely-used algorithms and primitives for machine learning and information retrieval. The algorithms are CUDA-accelerated and form building blocks for more easily writing high performance applications.
A massively parallel, optimal functional runtime in Rust
CUDA Library Samples
SpargeAttention: A training-free sparse attention that can accelerate any model inference.
Instant neural graphics primitives: lightning fast NeRF and more