rustycyclo

Strip Spectral Correlation Analyzer (SSCA) implementation CUDA wrapped inside a Rust interface.

Simple usage example (from docs)

Check examples folder.

Development environment

Since this repository compiles CUDA code, it needs a specific nvcc compiler to build along with cargo. The Dockerfile in the root of this repository can be used to create a container that can be used to develop and compile the library. Use docker compose up to bring up the container. Then docker exec -it <container_id> /bin/bash. cargo can be used as normal inside the container.

Benchmarks

cargo bench inside the analyzer folder to run the analyzer benchmarks. The provided benchmarks runs at ~40 MSamples/sec on an NVIDIA A10 GPU with a server-class Xeon CPU. On a NVIDIA GeForce GTX 1070 GPU, we get 21 MSamples/sec.