Fair CUDA Streams benchmark for pinned memory, asynchronous copies, events, overlap and Nsight Systems.
-
Updated
Aug 31, 2026 - Cuda
#
cuda-streams
Here are 7 public repositories matching this topic...
The MNIST classification problem is a fundamental machine learning task that involves recognizing handwritten digits (0- 9) from a dataset of 70,000 grayscale images (28x28 pixels each). It serves as a benchmark for evaluating machine learning models, particularly neural networks.
benchmarking deep-learning parallel-computing cuda mnist neural-networks high-performance-computing gpu-acceleration profiling shared-memory openacc performance-optimization c-cpp nsight tensor-cores cuda-streams pinned-memory
-
Updated
Sep 12, 2025 - Cuda
Multi-stream GPU edge detection over whole image directories — CUDA NPP plus two custom kernels, 103 images / 46 MP in 0.64 s on an A100, verified against an independent CPU reference that agrees on 99.9998% of pixels.
cpp parallel-computing cuda image-processing nvidia high-performance-computing gpu-acceleration edge-detection npp cuda-kernels gpu-computing gaussian-filter sobel otsu-threshold cuda-streams
-
Updated
Jul 24, 2026 - C++
CUDA C++ practice project for RTX 4070 SUPER — explore GPU concurrency, pinned memory, and Nsight profiling. Includes SAXPY and 2D blur kernels to train optimization, stream overlap, and timing analysis for NVIDIA Developer Technology Engineering skillset.
cpp gpu cuda nvidia high-performance-computing cuda-kernels gpu-optimization nsight parrallel-computing deep-learning-inference gpu-profiling cuda-streams pinned-memory
-
Updated
Nov 18, 2025 - Cuda
Profile-first CUDA lab for killing GPU pipeline bubbles: stream overlap, CUDA Graphs, torch.compile, spec-decode overlap — Nsight-verified
performance-engineering gpu cuda pytorch inference-optimization microbenchmark speculative-decoding gpu-profiling torch-compile cuda-streams cuda-graphs nsign-systems
-
Updated
Jul 18, 2026 - Python
A CUDA C++ demo showing how to overlap data transfer and kernel execution using multiple streams and pinned (page-locked) host memory. This project illustrates asynchronous memcpy, event timing, and performance benefits of concurrent GPU execution — essential for building high-throughput pipelines.
gpu cuda performance-optimization parallel-programming asynchronous-execution cuda-streams pinned-memory
-
Updated
Nov 17, 2025 - Cuda
CUDA-accelerated multi-restart K-Means clustering with custom kernels and CUDA-stream concurrency, benchmarked against a CPU baseline (~22–55× speedup on a Tesla T4).
machine-learning cpp parallel-computing high-performance-computing cuda-kernels gpu-computing kmeans-clustering nsight-systems cuda-streams
-
Updated
Jun 5, 2026 - C++
Add this topic to your repo
To associate your repository with the cuda-streams topic, visit your repo's landing page and select "manage topics."