LLM training in simple, raw C/CUDA

Instant neural graphics primitives: lightning fast NeRF and more

A massively parallel, optimal functional runtime in Rust

Flash Attention in ~100 lines of CUDA (forward pass only)

Pytorch Bindings for warp-ctc

llama3.cuda is a pure C/CUDA implementation for Llama 3 model.

an implementation of parallel linear BVH (LBVH) on GPU

High-speed GEMV kernels, at most 2.7x speedup compared to pytorch baseline.

GPU-Accelerated Lossless Data Compressors Survey

High-Performance SGEMM on CUDA devices

A GPU Accelerated Binary Vector Store