Lists (1)
Sort Name ascending (A-Z)
Stars
Decoded per-architecture SASS ISA tables from NVIDIA nvdisasm (CUDA 13.4), SM75 through SM121
Reverse engineering NVIDIA SASS instruction dictionary, kernel audits and pattern recognition across GPU architectures.
Minimalistic 4D-parallelism distributed training framework for education purpose
Third party assembler and GEMM library for NVIDIA Kepler GPU
Comprehensive collection of reinforcement learning algorithms implemented from scratch in PyTorch. Features tabular methods, value-based , policy gradient, actor-critic, offline RL, and multi-agent…
An agentic framework for GPU kernel optimization, combining a customizable benchmark SDK, correctness validation, and Agentic Nsight Compute analysis.
We aim to redefine Data Parallel libraries portabiliy, performance, programability and maintainability, by using C++ standard features, instead of creating new compilers.
bikrammajhi / cuLA
Forked from inclusionAI/cuLACUDA kernels for linear attention variants, written in CuTe DSL and CUTLASS C++.
Understanding and Implementing CUDA Kernels For Deep Learning Models
B200 Attention from Scratch: understand one of the hardest GPU kernels on the latest hardware
Header-only C++17 library with utilities for CUDA runtime code
100 days of LLM inference engineering — daily posts, experiments, and visualizations
A curated resource list for learning AI perf 8000 ormance engineering, from GPU fundamentals to production inference.
Custom inference pipeline for GPT2 using CUDA.
Training code for "SSTVOS: Sparse Spatiotemporal Transformers for Video Object Segmentation"
Complete solutions to the Programming Massively Parallel Processors Edition 4
A simple environment for writing and experimenting with hand-written CUDA PTX kernels.
Learn about PTX instructions ldmatrix and stmatrix
Training materials associated with NVIDIA's CUDA Training Series (www.olcf.ornl.gov/cuda-training-series/)