NumPy & SciPy for GPU

CUDA Library Samples

Hooked CUDA-related dynamic libraries by using automated code generation tools.

Escoin: Efficient Sparse Convolutional Neural Network Inference on GPUs

A complete example of batched refactorization in cuSOLVER.

A really old project that implemented the Stable Fluids using CUDA, cuBLAS and cuSPARSE

Compare preconditioned conjugate gradient algorithm with different sparse matrices, floating point precision levels, and implementations

Minimal examples to get CUDA linear algebra programs working with Rust using CC & FFI.

Repository for benchmarking linear solvers on GPU.

Collection of CUDA wrappers for a simplified kernel call

Compact Software Suite for Scientific Development - Linear Algebra

Generate narrated CUDA course videos with animated slides and AI avatars using Remotion, Gemini, and ElevenLabs TTS for automated production.

High-performance GPU SpMV kernels in Python/Numba CUDA. Achieves 96.6 GB/s (53% of cuSPARSE) with 5,432x speedup over CPU. Includes optimization analysis.

CUDA SpMV kernels (scalar, warp-per-row, ELL) on NVIDIA A100 benchmarked against cuSPARSE on SuiteSparse matrices, plus AVX2 + cache-tiled CPU baselines on Intel Xeon Gold. Vector kernel reaches 98-110% of HBM2 peak, beating cuSPARSE by 24-56% on regular matrices.

cuda、cublas、cublaslt、cusparse...

Sparse matrix operations for AWS Trainium via NKI (cuSPARSE-equivalent) — CSR/COO formats, SpMV and SpMM via gather-matmul-scatter, Schwarz integral screening for quantum chemistry.

Benchmarks for the sparse matrix–vector multiplication (SpMV) with matrix formats in TNL.

High-performance GPU-accelerated linear algebra library for scientific computing. Custom kernels outperform cuBLAS+cuSPARSE by 2.4x in iterative solvers. Built for circuit simulation workloads.

Reproducible Instruction Roofline analysis of cuSPARSE and Ginkgo SpMM on RTX 4090 using Nsight Compute metrics.