HazyResearch / ThunderKittens
Tile primitives for speedy kernels
See what the GitHub community is most excited about this week.
Tile primitives for speedy kernels
CUDA Library Samples
LLM training in simple, raw C/CUDA
Causal depthwise conv1d in CUDA, with a PyTorch interface
FlashInfer: Kernel Library for LLM Serving
cuGraph - RAPIDS Graph Analytics Library
CUDA accelerated rasterization of gaussian splatting
cuVS - a library for vector search and clustering on the GPU
Instant neural graphics primitives: lightning fast NeRF and more
[ARCHIVED] Cooperative primitives for CUDA C++. See https://github.com/NVIDIA/cccl
A massively parallel, optimal functional runtime in Rust
RCCL Performance Benchmark Tests
A throughput-oriented high-performance serving framework for LLMs