Docs - radumarias/rvirt-gpu GitHub Wiki

libs:

• Vulkan bindings for Rust https://crates.io/crates/ash

• Implementation of the Khronos OpenCL 3.0 API and extensions https://crates.io/crates/opencl3

• CUDA bindings https://crates.io/crates/cuda

• Cross-platform, safe, pure-rust graphics API. It runs natively on Vulkan, Metal, D3D12, and OpenGL; and on top of WebGL2 and WebGPU on wasm. https://github.com/gfx-rs/wgpu

• https://developer.nvidia.com/gpudirect

• https://miro.com/welcomeonboard/NFZMNUJqOVU2TDV6cGFlRVpSaVdaTHQ5MGhOVHBlQTJvR3h2VUdlNlpGVWJwZk85MDl1NmRrd1FKblVKRTFaVXwzMDc0NDU3MzQ4MTM3MDIyMDU2fDI=?share_link_id=704793315398

• https://github.com/Rust-GPU/Rust-CUDA

• https://github.com/MWATelescope/mwa_hyperdrive

• https://www.reddit.com/r/rust/comments/18209in/gpu_programming_in_rust/

• cuCrypt

• cuBLAS

• libgcrypt: While libgcrypt itself doesn't use CUDA, you can use it as a reference for implementing AES or ChaCha20, and then port the computation-heavy parts to CUDA.

• CryptoGPU: Some researchers and developers have created libraries and projects that implement cryptographic algorithms on GPUs. Searching through academic papers and GitHub repositories with keywords like "CUDA AES-GCM" or "CUDA ChaCha20-Poly1305" might yield usable code or at least examples to build upon.

• CUDA SDK and Examples: NVIDIA's CUDA SDK and documentation provide examples on how to write efficient parallel code, which can be adapted for cryptographic algorithms like AES-GCM and ChaCha20-Poly1305.

CUDA Crypto Libraries: NVIDIA provides certain cryptographic functions through its CUDA libraries, such as cuBLAS and cuRAND. While these libraries don't directly implement AES-GCM, they offer primitives that can be used to build such an implementation.