fast.cu
A collection of GPU kernels written from scratch in CUDA, including a matrix multiplication kernel for H100 that is reported to match or beat cuBLAS. Useful as a learning reference for GPU performance.
Share on XLicense: MIT
Overview
fast.cu is a collection of GPU kernels written from scratch in CUDA. It includes bf16 matrix multiplication on H100, NVFP4 matrix multiplication on GB300 with ten kernel snapshots, and a sum reduction kernel on H100. The README reports the H100 matrix multiplication and sum kernels at or above cuBLAS and cub in its own tests. Each example builds with make, and a worklog explains the matmul work.
Key features
- H100 bf16 matrix multiplication kernels with fp32 accumulation
- GB300 NVFP4 matmul with ten kernel snapshots and a benchmark runner
- H100 sum reduction over 2^30 elements
- Simple make targets to build and run each example
Best for
People learning GPU performance tuning who want complete, readable kernels to study. The NVFP4 example needs a GB300 or B300 and CUDA 13.1.
- Upstream
- pranjalssh/fast.cu
- Fork on GitHub
- Guo-astro/fast.cu
- Upstream stars
- 628
- Category
- Developer tools and infrastructure
- Language
- Cuda
- License
- MIT
- Forked
- 2025-12-09
- Sync status
- In syncLast synced 2026-10-13
More in Developer tools and infrastructure
iPhone Use lets Codex control a real iPhone over USB through WebDriverAgent for app navigation, page reading, gestures, and text entry. It automates phone tasks and shows a live screen preview.
Forked 2026-10-11Last synced 2026-10-13License: MITDeveloper tools and infrastructureGitHub
AnyPS5 ports PS5 executables to Linux and Windows by converting them to native formats and providing system PRX libraries for dynamic linking. It is for running compatible games without emulation or a separate runtime process.
Forked 2026-10-11Last synced 2026-10-13License: GPL-2.0Developer tools and infrastructureGitHub
Creates full-stack project scaffolds from compatible stack parts, reducing setup and integration work. Use the CLI, browser builder, or MCP server to generate a project.
Forked 2026-10-09Last synced 2026-10-13License: MITDeveloper tools and infrastructureGitHub
A Python SDK for adding local face detection, landmark extraction, and face comparison to applications on Windows and Linux, without sending images off-device.
Forked 2026-10-09Last synced 2026-10-13No license declaredDeveloper tools and infrastructureGitHub