Skip to content

Latest commit

 

History

History
63 lines (63 loc) · 2.26 KB

CHANGELOG.org

File metadata and controls

63 lines (63 loc) · 2.26 KB
Raw

0.4.3

Features

  • Significantly speed up automatic iterators (huge thanks to Osveron!)

0.4.2

Features

  • Add simd_for_each
  • Add equality comparison via PackedEq

0.4.1

Bugfixes & Minor Improvements

  • Fix an issue with zipping even collections
  • Fix an upcast on AVX2 machines

0.4.0

Announcing faster 0.4.0 - a 4,500+ line diff from 0.3.0

Big Changes

  • Support #![no_std]
  • Add striping, gathers, and scatters.
  • Add vector merging
  • Add vectorized endianness operations
  • Add limited vector swizzling
  • Add lockstep packed iterators

Features

  • Add tons of docstrings
  • Allow FnMut closures in simd_map and simd_reduce
  • Vectorize operations on last elements of an uneven collection
  • Implement compound assignment operators for architectures without hardware SIMD
  • Add large vectors for architectures without hardware SIMD

Bugfixes & Minor Improvements

  • Add a changelog
  • Fix a correctness issue when mapping over uneven collections
  • Vectorize min/max for SSE4.1
  • Vectorize Upcast for SSE4.1
  • Implement Downcast polyfills on many more vector types
  • Implement and test [saturating_]{hadd,hsub} on more vector types.
  • Undeprecate halfs and interleave

0.3.0

Announcing Faster 0.3.0, a 3,500+ line diff from 0.1.1

Big Changes

  • Support for targets without hardware SIMD
  • Support for architectures other than x86
  • Documentation & examples for most objects
  • Intuitive support for uneven collections
  • Support SIMD-accelerated reductive operations
  • Add upcasting, casting, and downcasting

Features

  • Add summation and product calculations for all vectors
  • Add default initializer for vectors
  • Add many more polyfills and feature gates
  • Allow scalar iteration of SIMD iterators with `map` and `fold`
  • Add vector constructors for interleaved and half-and-half patterns
  • Expose saturating addition and subtraction intrinsics

Bugfixes & Minor Improvements

  • Add tests for sqrt, transmute, abs, recip, upcast, downcast, cast, and more
  • Make [saturating]_{hadd,hsub} portable

0.2.0

Announcing Faster 0.2.0, a 2,500+ line diff from 0.1.1

Big Changes

  • Change license to MPL 2.0
  • Support for uneven collections
  • Add packed transmutations

Features

  • Implement many more operations and polyfills

Maintainence & Bugfixes:

  • Require only SSE for