Cuda Trending

karpathy /

llm.c

#14

LLM training in simple, raw C/CUDA

xlite-dev /

LeetCUDA

#2

📚LeetCUDA: Modern CUDA Learn Notes with PyTorch for Beginners🐑, 200+ CUDA Kernels, Tensor Cores, HGEMM, FA-2 MMA.🎉

deepseek-ai /

DeepGEMM

#3

DeepGEMM: clean and efficient BLAS kernel library on GPU

deepseek-ai /

DeepEP

#19

DeepEP: an efficient expert-parallel communication library

NVIDIA /

nccl-tests

#18

NCCL Tests

NVIDIA /

cuopt

#16

GPU accelerated decision optimization

NVlabs /

instant-ngp

#8

Instant neural graphics primitives: lightning fast NeRF and more

alibaba /

rtp-llm

#1

RTP-LLM: Alibaba's high-performance LLM inference engine for diverse applications.

NVIDIA /

cuvs

#15

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

brucefan1983 /

GPUMD

#10

Graphics Processing Units Molecular Dynamics

NVIDIA /

raft

#9

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.

NVIDIA /

nvbench

#7

CUDA Kernel Benchmarking Library

thu-ml /

SageAttention

#6

[ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

rapidsai /

cugraph

#17

cuGraph - RAPIDS Graph Analytics Library

HigherOrderCO /

HVM2

#13

A massively parallel, optimal functional runtime in Rust

Dao-AILab /

causal-conv1d

#12

Causal depthwise conv1d in CUDA, with a PyTorch interface

NVIDIA /

cub

#5

[ARCHIVED] Cooperative primitives for CUDA C++. See https://github.com/NVIDIA/cccl

baidu-research /

warp-ctc

#4

Fast parallel CTC.

karpathy /

llm.c

#7

LLM training in simple, raw C/CUDA

thu-ml /

SageAttention

#14

[ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

mirage-project /

mirage

#17

Mirage Persistent Kernel: Compiling LLMs into a MegaKernel

NVlabs /

instant-ngp

#19

Instant neural graphics primitives: lightning fast NeRF and more

alibaba /

rtp-llm

#16

RTP-LLM: Alibaba's high-performance LLM inference engine for diverse applications.

deepseek-ai /

DeepGEMM

#1

DeepGEMM: clean and efficient BLAS kernel library on GPU

NVIDIA /

cub

#13

[ARCHIVED] Cooperative primitives for CUDA C++. See https://github.com/NVIDIA/cccl

NVIDIA /

nvbench

#11

CUDA Kernel Benchmarking Library

HazyResearch /

ThunderKittens

#4

Tile primitives for speedy kernels

NVIDIA /

nccl-tests

#3

NCCL Tests

NVIDIA /

cuopt

#18

GPU accelerated decision optimization

NVIDIA /

raft

#15

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.

baidu-research /

warp-ctc

#10

Fast parallel CTC.

Dao-AILab /

causal-conv1d

#9

Causal depthwise conv1d in CUDA, with a PyTorch interface

siboehm /

SGEMM_CUDA

#8

Fast CUDA matrix multiplication from scratch

NVIDIA /

cuvs

#6

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

rapidsai /

cugraph

#5

cuGraph - RAPIDS Graph Analytics Library

deepseek-ai /

DeepEP

#2

DeepEP: an efficient expert-parallel communication library

HazyResearch /

ThunderKittens

#17

Tile primitives for speedy kernels

deepseek-ai /

DeepEP

#3

DeepEP: an efficient expert-parallel communication library

karpathy /

llm.c

#18

LLM training in simple, raw C/CUDA

deepseek-ai /

DeepGEMM

#1

DeepGEMM: clean and efficient BLAS kernel library on GPU

siboehm /

SGEMM_CUDA

#14

Fast CUDA matrix multiplication from scratch

NVIDIA /

nccl-tests

#5

NCCL Tests

alibaba /

rtp-llm

#2

RTP-LLM: Alibaba's high-performance LLM inference engine for diverse applications.

NVIDIA /

cuopt

#9

GPU accelerated decision optimization

NVIDIA /

nvbench

#19

CUDA Kernel Benchmarking Library

NVlabs /

instant-ngp

#16

Instant neural graphics primitives: lightning fast NeRF and more

HigherOrderCO /

HVM2

#11

A massively parallel, optimal functional runtime in Rust

thu-ml /

SageAttention

#8

[ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

Dao-AILab /

causal-conv1d

#4

Causal depthwise conv1d in CUDA, with a PyTorch interface

brucefan1983 /

GPUMD

#15

Graphics Processing Units Molecular Dynamics

NVIDIA /

cub

#13

[ARCHIVED] Cooperative primitives for CUDA C++. See https://github.com/NVIDIA/cccl

NVIDIA /

cudf-spark-jni

#12

RAPIDS Accelerator JNI For Apache Spark

baidu-research /

warp-ctc

#7

Fast parallel CTC.

NVIDIA /

cuvs

#6

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

karpathy /

llm.c

#2

LLM training in simple, raw C/CUDA

deepseek-ai /

DeepEP

#3

DeepEP: an efficient expert-parallel communication library

deepseek-ai /

DeepGEMM

#1

DeepGEMM: clean and efficient BLAS kernel library on GPU

HazyResearch /

ThunderKittens

#8

Tile primitives for speedy kernels

NVIDIA /

nccl-tests

#4

NCCL Tests

rapidsai /

cugraph

#17

cuGraph - RAPIDS Graph Analytics Library

NVlabs /

instant-ngp

#13

Instant neural graphics primitives: lightning fast NeRF and more

BBuf /

how-to-optim-algorithm-in-cuda

#6

how to optimize some algorithm in cuda.

NVIDIA /

cuopt

#16

GPU accelerated decision optimization

brucefan1983 /

GPUMD

#15

Graphics Processing Units Molecular Dynamics

NVIDIA /

cudf-spark-jni

#12

RAPIDS Accelerator JNI For Apache Spark

thu-ml /

SageAttention

#11

[ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

baidu-research /

warp-ctc

#9

Fast parallel CTC.

Dao-AILab /

causal-conv1d

#7

Causal depthwise conv1d in CUDA, with a PyTorch interface

NVIDIA /

cuvs

#5

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

karpathy /

llm.c

#16

LLM training in simple, raw C/CUDA

deepseek-ai /

DeepGEMM

#2

DeepGEMM: clean and efficient BLAS kernel library on GPU

Dao-AILab /

causal-conv1d

#13

Causal depthwise conv1d in CUDA, with a PyTorch interface

HazyResearch /

ThunderKittens

#12

Tile primitives for speedy kernels

NVIDIA /

nccl-tests

#9

NCCL Tests

alibaba /

rtp-llm

#18

RTP-LLM: Alibaba's high-performance LLM inference engine for diverse applications.

siboehm /

SGEMM_CUDA

#17

Fast CUDA matrix multiplication from scratch

thu-ml /

SageAttention

#4

[ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

deepseek-ai /

DeepEP

#1

DeepEP: an efficient expert-parallel communication library

NVIDIA /

cub

#10

[ARCHIVED] Cooperative primitives for CUDA C++. See https://github.com/NVIDIA/cccl

NVIDIA /

raft

#20

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.

mirage-project /

mirage

#19

Mirage Persistent Kernel: Compiling LLMs into a MegaKernel

rapidsai /

cugraph

#15

cuGraph - RAPIDS Graph Analytics Library

NVIDIA /

cuvs

#14

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

baidu-research /

warp-ctc

#8

Fast parallel CTC.

NVIDIA /

nvbench

#7

CUDA Kernel Benchmarking Library

rahul-goel /

fused-ssim

#6

Lightning fast differentiable SSIM.

NVIDIA /

cuopt

#5

GPU accelerated decision optimization

NVlabs /

instant-ngp

#3

Instant neural graphics primitives: lightning fast NeRF and more

karpathy /

llm.c

#2

LLM training in simple, raw C/CUDA

NVlabs /

instant-ngp

#11

Instant neural graphics primitives: lightning fast NeRF and more

thu-ml /

SageAttention

#9

[ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

NVIDIA /

cuvs

#13

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

deepseek-ai /

DeepGEMM

#4

DeepGEMM: clean and efficient BLAS kernel library on GPU

deepseek-ai /

DeepEP

#17

DeepEP: an efficient expert-parallel communication library

mirage-project /

mirage

#14

Mirage Persistent Kernel: Compiling LLMs into a MegaKernel

HazyResearch /

ThunderKittens

#7

Tile primitives for speedy kernels

NVIDIA /

cuopt

#5

GPU accelerated decision optimization

NVIDIA /

cub

#3

[ARCHIVED] Cooperative primitives for CUDA C++. See https://github.com/NVIDIA/cccl

rapidsai /

cugraph

#1

cuGraph - RAPIDS Graph Analytics Library

Dao-AILab /

causal-conv1d

#16

Causal depthwise conv1d in CUDA, with a PyTorch interface

rahul-goel /

fused-ssim

#15

Lightning fast differentiable SSIM.

supranational /

sppark

#12

Zero-knowledge template library

baidu-research /

warp-ctc

#6

Fast parallel CTC.

NVIDIA /

nccl-tests

#10

NCCL Tests

karpathy /

llm.c

#8

LLM training in simple, raw C/CUDA

mirage-project /

mirage

#14

Mirage Persistent Kernel: Compiling LLMs into a MegaKernel

NVlabs /

instant-ngp

#12

Instant neural graphics primitives: lightning fast NeRF and more

HazyResearch /

ThunderKittens

#6

Tile primitives for speedy kernels

BBuf /

how-to-optim-algorithm-in-cuda

#13

how to optimize some algorithm in cuda.

NVIDIA /

raft

#11

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.

thu-ml /

SageAttention

#7

[ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

alibaba /

rtp-llm

#16

RTP-LLM: Alibaba's high-performance LLM inference engine for diverse applications.

Dao-AILab /

causal-conv1d

#15

Causal depthwise conv1d in CUDA, with a PyTorch interface

NVIDIA /

nvbench

#10

CUDA Kernel Benchmarking Library

supranational /

sppark

#5

Zero-knowledge template library

NVIDIA /

cuopt

#4

GPU accelerated decision optimization

NVIDIA /

cub

#3

[ARCHIVED] Cooperative primitives for CUDA C++. See https://github.com/NVIDIA/cccl

baidu-research /

warp-ctc

#2

Fast parallel CTC.

NVIDIA /

cuvs

#1

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