Lib.rs

Highlights

• Parallel Zip
• Parallel elementwise (no order) iterator
• Parallel .axis_iter() (and _mut)
• .par_map_inplace() for arrays.

Status and Lookout

• Still iterating on and evolving the crate
  • A separate crate is less convenient (doesn't use rayon IntoParallelIterator trait, but a separate trait) but allows rapid iteration and we can follow the evolution of rayon's internals. This crate is double pace: For every ndarray or rayon major version, this crate goes up one major version.
• Performance:
  • TBD. Tell me about your experience.
  • You'll need a big chunk of data (or an expensive operation per data point) to gain from parallelization.

How to use with cargo:

[dependencies]
ndarray-parallel = "0.4"

Recent Changes (ndarray-parallel)

• 0.9.1
  • Mark crate as deprecated, replaced by direct support in ndarray.
• 0.9.0
  • Upgrade for ndarray 0.12.0
• 0.8.0
  • Upgrade for rayon 1.0!
• 0.7.0
  • Upgrade for ndarray 0.11.0
• 0.6.0
  • Upgrade for rayon 0.9.0 by @cuviper
  • Fix unused mut warning when using par_azip!()
• 0.5.0
  • Upgrade for ndarray 0.10.0
  • Add par_azip!() macro by @msiglreith
• 0.4.0
  • Update Rayon dependency to 0.8.2
• 0.3.0
  • ParallelIterator for Zip, including .par_apply.
  • .par_map_inplace and .par_mav_inplace for arrays
  • Require ndarray 0.9 and rayon 0.7
  • Fix bug with array par_iter()'s .collect()
• 0.2.0
  • Require for ndarray 0.8
• 0.1.1
  • Clarify docs
  • Add categories
• 0.1.0
  • Initial release
  • Elementwise parallel iterator (no order) and parallel axis iterators.