Trending Cuda Projects

Top 44 Trending Cuda Projects

1. FlashMLA

   1 11,492 99.4%

   FlashMLA: Efficient MLA decoding kernels

   

2. DeepEP

   2 7,454 95.1%

   DeepEP: an efficient expert-parallel communication library

   

3. nunchaku

   3 1,532 51.6%

   [ICLR2025 Spotlight] SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models

   

4. BenchmarkCustomPTX

   4 123 36.6%

   Custom PTX Instruction Benchmark

   

5. SpargeAttn

   5 491 22.0%

   SpargeAttention: A training-free sparse attention that can accelerate any model inference.

   

6. AlexNet-Source-Code

   6 2,564 17.4%

   This package contains the original 2012 AlexNet code.

   

7. SageAttention

   7 1,397 12.5%

   Quantized Attention achieves speedup of 2-3x and 3-5x compared to FlashAttention and xformers, without lossing end-to-end metrics across language, image, and video models.

   

8. flash-attention-minimal

   8 790 11.3%

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

   

9. causal-conv1d

   9 438 10.0%

   Causal depthwise conv1d in CUDA, with a PyTorch interface

   

10. ThunderKittens

    10 2,301 10.0%

    Tile primitives for speedy kernels

    

11. Parallel-Computing-Cuda-C

    11 359 7.5%

    CUDA Learning guide

    

12. NATTEN

    12 443 7.2%

    Neighborhood Attention Extension. Bringing attention to a neighborhood near you!

    

13. Nanoflow

    13 799 6.8%

    A throughput-oriented high-performance serving framework for LLMs

    

14. nccl-tests

    14 1,075 6.6%

    NCCL Tests

    

15. CUDALibrarySamples

    15 1,902 5.9%

    CUDA Library Samples

    

16. cuda_programming

    16 826 5.6%

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

    

17. HRNet-Human-Pose-Estimation

    17 283 4.9%

    This repo is copied from https://github.com/leoxiaobin/deep-high-resolution-net.pytorch

    

18. array-language-comparisons

    18 173 4.6%

    A comparison of array languages & libraries: APL, J, BQN, Uiua, Q, Julia, R, NumPy, Nial, Futhark, Dex, Ivy, SaC & ArrayFire.

    

19. cugraph

    19 1,956 4.3%

    cuGraph - RAPIDS Graph Analytics Library

    

20. cuhnsw

    20 155 3.9%

    CUDA implementation of Hierarchical Navigable Small World Graph algorithm

    

21. cuspatial

    21 664 2.9%

    CUDA-accelerated GIS and spatiotemporal algorithms

    

22. raft

    22 871 2.6%

    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 high performance applications. (by rapidsai)

    

23. llm.c

    23 26,399 2.4%

    LLM training in simple, raw C/CUDA

    

24. GPUSorting

    24 327 1.8%

    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.

    

25. TorchPQ

    25 219 1.8%

    Approximate nearest neighbor search with product quantization on GPU in pytorch and cuda

    

26. k2

    26 1,189 1.8%

    FSA/FST algorithms, differentiable, with PyTorch compatibility.

    

27. cuda-convnet2

    27 799 1.8%

    Automatically exported from code.google.com/p/cuda-convnet2

    

28. Gpufit

    28 324 1.5%

    GPU-accelerated Levenberg-Marquardt curve fitting in CUDA

    

29. CGBN

    29 209 1.4%

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

    

30. RWKV-CUDA

    30 222 1.4%

    The CUDA version of the RWKV language model ( https://github.com/BlinkDL/RWKV-LM )

    

31. instant-ngp

    31 16,534 1.3%

    Instant neural graphics primitives: lightning fast NeRF and more

    

32. HVM

    32 10,972 1.2%

    A massively parallel, optimal functional runtime in Rust

    

33. Lantern

    33 170 1.2%

    

34. megalodon

    34 521 1.0%

    Reference implementation of Megalodon 7B model (by XuezheMax)

    

35. dietgpu

    35 325 0.9%

    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.

    

36. kilonerf

    36 484 0.8%

    Code for KiloNeRF: Speeding up Neural Radiance Fields with Thousands of Tiny MLPs

    

37. blocksparse

    37 1,037 0.7%

    Efficient GPU kernels for block-sparse matrix multiplication and convolution

    

38. unet.cu

    38 602 0.7%

    UNet diffusion model in pure CUDA

    

39. MegBA

    39 463 0.6%

    MegBA: A GPU-Based Distributed Library for Large-Scale Bundle Adjustment

    

40. deep-high-resolution-net.pytorch

    40 4,394 0.4%

    The project is an official implementation of our CVPR2019 paper "Deep High-Resolution Representation Learning for Human Pose Estimation"

    

41. nvParse

    41 561 0.2%

    Fast, gpu-based CSV parser

    

42. deep-painterly-harmonization

    42 6,073 0.0%

    Code and data for paper "Deep Painterly Harmonization": https://arxiv.org/abs/1804.03189

    

43. instant-ngp-Windows

    43 503 0.0%

    Instant neural graphics primitives: lightning fast NeRF and more

    

44. SENet

    44 3,443 0.0%

    Squeeze-and-Excitation Networks

    

Index