Gpu Programming Python Packages

scalene

Scalene: a high-performance, high-precision CPU, GPU, and memory profiler for Python with AI-powered optimization proposals

taichi

Productive, portable, and performant GPU programming in Python.

geomstats

Computations and statistics on manifolds with geometric structures.

metal-linalg

QR, symmetric eigendecomposition and SVD for batches of matrices on Apple Silicon GPUs, for MLX and PyTorch. 2x to 10x gain in performance. Metal kernels per regime, routed to the GPU or the CPU by policies measured per Mac. C++, Python (pip: metal-linalg, metal-linalg-torch), Swift and C.

cuda-cccl

CUDA Core Compute Libraries

fragmentcolor

Universal GPU abstraction for Rust, Javascript, Python, Swift, and Kotlin.

fast-trimul

A drop-in, hardware-agnostic library for Fused Triangle Multiplicative Updates across AlphaFold3-family models, powered by CUTLASS CuTe kernels.

cthreads

Typed Python that compiles to C++ and runs on real OS threads. Same jobs/pools/sync feel as normal Python, without the GIL or multiprocessing overhead. Optional Vulkan @Gpu kernels via pip install cthreads-gpu

taichi-nightly

Productive, portable, and performant GPU programming in Python.

metal-linalg-torch

QR, symmetric eigendecomposition and SVD for batches of matrices on Apple Silicon GPUs, for MLX and PyTorch. 2x to 10x gain in performance. Metal kernels per regime, routed to the GPU or the CPU by policies measured per Mac. C++, Python (pip: metal-linalg, metal-linalg-torch), Swift and C.

cthreads-gpu

Typed Python that compiles to C++ and runs on real OS threads. Same jobs/pools/sync feel as normal Python, without the GIL or multiprocessing overhead. Optional Vulkan @Gpu kernels via pip install cthreads-gpu

nabla-ml

Nabla: High-Performance Scientific Computing

mlsys-lab

Auto-graded exercises in low-level ML systems: real C++, real CUDA-C on a software GPU (no NVIDIA hardware), deterministic measured gates.

cutile-gpt

A complete GPT built entirely on NVIDIA cuTile: declarative GPU kernels in ~20 lines instead of ~150 hand-tuned CUDA, matching PyTorch speed in a ~10MB footprint

triton-transformer

Implementation of a Transformer, but completely in Triton

gpu-wait

A package to run commands when GPU resources are available

chronos-gpu

Chronos is a time-based GPU partitioning utility that allows multiple users or applications to share a single GPU by creating exclusive time-limited partitions with automatic expiration. Built with OpenCL, it works across platforms including macOS (Apple Silicon & Intel), Linux, and Windows.

numba-timer

A helper package to easily time Numba CUDA GPU events ⌛

greyhound-kernels

CuTe DSL kernels for PyTorch training and inference workloads

stgraph

🌟 Compiler for vertex-centric programming of GNNs/TGNNs

cuda-parallel

CUDA Core Compute Libraries

cuda-cooperative

CUDA Core Compute Libraries