LLM training in simple, raw C/CUDA

A massively parallel, optimal functional runtime in Rust

Efficient GPU kernels for block-sparse matrix multiplication and convolution

Step-by-step optimization of CUDA SGEMM

Optimizing SGEMM kernel functions on NVIDIA GPUs to a close-to-cuBLAS performance.

CGBN: CUDA Accelerated Multiple Precision Arithmetic (Big Num) using Cooperative Groups

Zero-knowledge template library

Cuda implementation of EVM bytecode executor