A General-purpose Task-parallel Programming System using Modern C++

Sample codes for my CUDA programming book

📚150+ Tensor/CUDA Cores Kernels, 🔥toy-hgemm library with WMMA, MMA and CuTe (99%~100%+ TFLOPS of cuBLAS 🎉🎉).

CUDA Core Compute Libraries

Thin, unified, C++-flavored wrappers for the CUDA APIs

Adan: Adaptive Nesterov Momentum Algorithm for Faster Optimizing Deep Models

TinyChatEngine: On-Device LLM Inference Library

Safe rust wrapper around CUDA toolkit

🚀 你的YOLO部署神器。TensorRT Plugin、CUDA Kernel、CUDA Graphs三管齐下，享受闪电般的推理速度。| Your YOLO Deployment Powerhouse. With the synergy of TensorRT Plugins, CUDA Kernels, and CUDA Graphs, experience lightning-fast inference speeds.

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

A self-learning tutorail for CUDA High Performance Programing.

This is an archive of materials produced for an introductory class on CUDA programming at Stanford University in 2010

From zero to hero CUDA for accelerating maths and machine learning on GPU.

μ-Cuda, COVER THE LAST MILE OF CUDA. With features: intellisense-friendly, structured launch, automatic cuda graph generation and updating.

An implementation of HIP that works on CPUs, across OSes.

CUDA kernel author's tools

CudaPAD is a PTX/SASS viewer for NVIDIA Cuda kernels and provides an on-the-fly view of the assembly.

CUDA Guide

Install CUDA on Windows11 using WSL2

Speed up image preprocess with cuda when handle image or tensorrt inference