A massively parallel, optimal functional runtime in Rust

DeepEP: an efficient expert-parallel communication library

CUDA Library Samples

cuGraph - RAPIDS Graph Analytics Library

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

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 …

Code from the "CUDA Crash Course" YouTube series by CoffeeBeforeArch

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

Fast k nearest neighbor search using GPU

State of the art sorting and segmented sorting, including OneSweep. Implemented in CUDA, D3D12, and Unity style compute shaders. Theoretically portable to all wave/warp/subgroup sizes.

CUDA Data Parallel Primitives Library

A simple GPU hash table implemented in CUDA using lock free techniques

GPU implementation of a fast generalized ANS (asymmetric numeral system) entropy encoder and decoder, with extensions for lossless compression of numerical and other data types in HPC/ML applications.

HierarchicalKV is a part of NVIDIA Merlin and provides hierarchical key-value storage to meet RecSys requirements. The key capability of HierarchicalKV is to store key-value feature-embeddings on h…

Efficient Top-K implementation on the GPU

Custom PTX Instruction Benchmark

GPUfs - File system support for NVIDIA GPUs

RAPIDS Accelerator JNI For Apache Spark

Harmonia is an algorithm that allows for the implementation of operations on B+ trees using parallelization. As a part of my GPU project, I implemented the Harmonia paper published in 2019 in CUDA.

Comparison of regression line calculation in CUDA GPU code vs. AVX-512 code