RustMP - Implicit Parallelism in Rust
Go to file
Jack Yu 0b22ea2395 Keyword change again: shared, shared_mut, shared_unsafe 2021-04-30 19:35:22 -04:00
omp Updated README, added rayon and tests for comparision 2021-04-30 19:24:15 -04:00
src Keyword change again: shared, shared_mut, shared_unsafe 2021-04-30 19:35:22 -04:00
.gitignore Removed imports for multiple calls to par_for!, renamed variables to 2021-04-18 15:22:03 -04:00
Cargo.toml Updated README, added rayon and tests for comparision 2021-04-30 19:24:15 -04:00
README.md Updated README, added rayon and tests for comparision 2021-04-30 19:24:15 -04:00

README.md

RustMP

This repository contains the source code for Alex Bowman, Paul Ouellette, Raffi Sanna, and Jack Yu's research project for CSC200H: an implicit parallelism library written in Rust. For more details, please see the project report.

Authors

Building/Running

RustMP requires the following dependencies to run:

  • rand-v0.8.3: Random number generation used for benchmarking examples
  • rayon-v1.5.0: Rayon parallel iterator library used in RustMP/Rayon.. comparision tests
  • hwloc2-v2.2.0: C HWLoc wrapper used for NUMA aware process pinning
  • lazy_static-v1.4.0: Lazy static macro used for delayed singleton init

src/ contains all code for the RustMP library, including lib.rs, sysinfo.rs, and threadpool.rs. src/bin contains benchmarking programs demonstrated in our paper. To run one of the benchmarks with cargo, execute the following command:

$ cargo run --release --bin <testname>

Additional comparison programs are located in ext/, where sequential, OpenMP and Rayon comparison benchmarks can be found.

Known issues

Due to Rust compiler limitations, Rust is only able to support up to 32 layers of nested macros by default. If an alternative test program that exceeds this nested depth for macros is used, please consider increasing this limit in Cargo.toml.