LibreCUDA is a project aimed at replacing the CUDA driver API to enable launching CUDA code on Nvidia GPUs without relying on the proprietary CUDA runtime. It achieves this by communicating directly with the hardware via ioctls, (specifically what Nvidia's open-gpu-kernel-modules refer to as the rmapi), as well as QMD, Nvidia's MMIO command queue structure. LibreCUDA is capable of uploading CUDA ELF binaries onto the GPU and launching them via the command queue.
Still in the early stages, it looks like.
Top, but for GPU processing on nVidia cards.
Go from graph data to a secure and interactive visual graph app in 15 minutes. Batteries-included self-hosting of graph data apps with Streamlit, Graphistry, RAPIDS, and more!
This open source effort puts together patterns the Graphistry team has reused across many graph projects as teams go from code-heavy Jupyter notebook experiments to deploying streamlined analyst tools. Whether building your first graph app, trying an idea, or wanting to check a reference, this project aims to simplify that process. It covers pieces like: Easy code editing and deployment, a project stucture ready for teams, built-in authentication, no need for custom JS/CSS at the start, batteries-included data + library dependencies, and fast loading & visualization of large graphs.
GPU enabled.
Seems to be Docker-only.
Utilities for cracking encrypted zip files that use weak encryption. CUDA enabled.
An open source brute-force passphrase cracker for attacking Truecrypt volumes. Can attack volumes that use the RIPEMD160 and AES-XTS cryptosystems by either throwing a dictionary at it or by generating pseudorandom passphrases from a user-defined set of characters. CUDA-aware, so nVidia graphics cards can be used to accelerate the process.