This is an adaptation of SLOTH (slow-timed hash function) into a time-asymmetric permutation using a standard CBC block cipher.
This library consists of 3 major implementations:
- CPU
- CUDA (requires
cudafeature to be enabled) - OpenCL (requires
openclfeature to be enabled)
CPU
CPU Implementation contains 3 flavors:
- optimized assembly-assisted implementation for x86-64 processors with ADX ISA extension (Linux, macOS and Windows)
- any 64-bt platform with support for
__int128C type (modern GCC/Clang, but not MSVC) - fallback for other platforms
For more details, README.md under src/cpu can be referred.
CUDA
CUDA implementation is heavily using low-level PTX code to achieve the maximum performance.
Details of the CUDA implementation can be found in README.md under src/cuda.
OpenCL
OpenCL is a universal implementation that supports AMD (ROCm), Nvidia (CUDA) and Intel GPUs, including support for multiple GPU vendors on the same machine at the same time.
Details of the OpenCL implementation can be found in mod.rs under src/opencl.
How to test
Simply run cargo test at the root folder
- use
cargo test --features=cudafor enabling cuda tests (CUDA toolkit must be installed on your device) - use
cargo test --features=openclfor enabling opencl tests (openCL drivers must be installed on your device)
How to benchmark
Simply run cargo bench at the root folder
- use
cargo bench --features=cudafor enabling cuda tests (CUDA toolkit must be installed on your device) - use
cargo bench --features=openclfor enabling opencl tests (openCL drivers must be installed on your device)
License
Licensed under either of
- Apache License, Version 2.0 (LICENSE-APACHE or https://www.apache.org/licenses/LICENSE-2.0)
- MIT license (LICENSE-MIT or https://opensource.org/licenses/MIT)
at your option.
Contribution
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in cc-rs by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.