Crate simd_aligned

NOTE - Do not use this crate for now. It has been reactivated to make FFSVM compile again, but needs some architectural work.

In One Sentence

You want to use std::simd but realized there is no simple, safe and fast way to align your f32x8 (and friends) in memory and treat them as regular f32 slices for easy loading and manipulation; simd_aligned to the rescue.

Highlights

• built on top of std::simd for easy data handling
• supports everything from u8x2 to f64x8
• think in flat slices (&[f32]), but get performance of properly aligned SIMD vectors (&[f32x16])
• defines u8s, …, f36s as “best guess” for current platform (WIP)
• provides N-dimensional VectorD and NxM-dimensional MatrixD.

Note: Right now this is an experimental crate. Features might be added or removed depending on how std::simd evolves. At the end of the day it’s just about being able to load and manipulate data without much fuzz.

Examples

Produces a vector that can hold 10 elements of type f64. Might internally allocate 5 elements of type f64x2, or 3 of type f64x4, depending on the platform. All elements are guaranteed to be properly aligned for fast access.

#![feature(portable_simd)]
use std::simd::*;
use simd_aligned::*;

// Create vectors of `10` f64 elements with value `0.0`.
let mut v1 = VectorD::<f64s>::with(0.0, 10);
let mut v2 = VectorD::<f64s>::with(0.0, 10);

// Get "flat", mutable view of the vector, and set individual elements:
let v1_m = v1.flat_mut();
let v2_m = v2.flat_mut();

// Set some elements on v1
v1_m[0] = 0.0;
v1_m[4] = 4.0;
v1_m[8] = 8.0;

// Set some others on v2
v2_m[1] = 0.0;
v2_m[5] = 5.0;
v2_m[9] = 9.0;

let mut sum = f64s::splat(0.0);

// Eventually, do something with the actual SIMD types. Does
// `std::simd` vector math, e.g., f64x8 + f64x8 in one operation:
sum = v1[0] + v2[0];

Benchmarks

There is no performance penalty for using simd_aligned, while retaining all the simplicity of handling flat arrays.

test vectors::packed       ... bench:          77 ns/iter (+/- 4)
test vectors::scalar       ... bench:       1,177 ns/iter (+/- 464)
test vectors::simd_aligned ... bench:          71 ns/iter (+/- 5)

FAQ

How does it relate to faster and std::simd?

• simd_aligned builds on top of std::simd. At aims to provide common, SIMD-aligned data structure that support simple and safe scalar access patterns.

• faster (as of today) is really good if you already have exiting flat slices in your code and want operate them “full SIMD ahead”. However, in particular when dealing with multiple slices at the same time (e.g., kernel computations) the performance impact of unaligned arrays can become a bit more noticeable (e.g., in the case of ffsvm up to 10% - 20%).

Re-exports

• pub use crate::matrix::AccessStrategy;
• pub use crate::matrix::Columns;
• pub use crate::matrix::Rows;
• pub use crate::arch::current::*;

Modules

• arch
  Contains vector definitions with a fixed bit width.
• cmpExperimental
  Traits for comparing and ordering vectors.
• numExperimental
  Traits for vectors with numeric elements.
• preludeExperimental
  The portable SIMD prelude.
• ptrExperimental
  Traits for vectors of pointers.
• traits
  Unified views on SIMD types.

Macros

• simd_swizzleExperimental
  Constructs a new SIMD vector by copying elements from selected elements in other vectors.

Structs

• LaneCountExperimental
  Specifies the number of lanes in a SIMD vector as a type.
• MaskExperimental
  A SIMD vector mask for N elements of width specified by Element.
• MatrixD
  A dynamic (heap allocated) matrix with one axis aligned for fast and safe SIMD access that also provides a flat view on its data.
• MatrixFlat
  Produced by MatrixD::flat, this allow for flat matrix access.
• MatrixFlatMut
  Provided by MatrixD::flat_mut, this allow for flat, mutable matrix access.
• SimdExperimental
  A SIMD vector with the shape of [T; N] but the operations of T.
• VectorD
  A dynamic (heap allocated) vector aligned for fast and safe SIMD access that also provides a flat view on its data.

Traits

• MaskElementExperimental
  Marker trait for types that may be used as SIMD mask elements.
• SimdCastExperimental
  Supporting trait for Simd::cast. Typically doesn’t need to be used directly.
• SimdElementExperimental
  Marker trait for types that may be used as SIMD vector elements.
• SimdExt
• StdFloatExperimental
  This trait provides a possibly-temporary implementation of float functions that may, in the absence of hardware support, canonicalize to calling an operating system’s math.h dynamically-loaded library (also known as a shared object). As these conditionally require runtime support, they should only appear in binaries built assuming OS support: std.
• SupportedLaneCountExperimental
  Statically guarantees that a lane count is marked as supported.
• SwizzleExperimental
  Create a vector from the elements of another vector.
• ToBytesExperimental
  Convert SIMD vectors to vectors of bytes

Functions

• packed_as_flat
  Converts an slice of SIMD vectors into a flat slice of elements.
• packed_as_flat_mut
  Converts a mutable slice of SIMD vectors into a flat slice of elements.

Type Aliases

• f32x1Experimental
  A SIMD vector with one element of type f32.
• f32x2Experimental
  A SIMD vector with two elements of type f32.
• f32x4Experimental
  A SIMD vector with four elements of type f32.
• f32x8Experimental
  A SIMD vector with eight elements of type f32.
• f32x16Experimental
  A SIMD vector with 16 elements of type f32.
• f32x32Experimental
  A SIMD vector with 32 elements of type f32.
• f32x64Experimental
  A SIMD vector with 64 elements of type f32.
• f64x1Experimental
  A SIMD vector with one element of type f64.
• f64x2Experimental
  A SIMD vector with two elements of type f64.
• f64x4Experimental
  A SIMD vector with four elements of type f64.
• f64x8Experimental
  A SIMD vector with eight elements of type f64.
• f64x16Experimental
  A SIMD vector with 16 elements of type f64.
• f64x32Experimental
  A SIMD vector with 32 elements of type f64.
• f64x64Experimental
  A SIMD vector with 64 elements of type f64.
• i8x1Experimental
  A SIMD vector with one element of type i8.
• i8x2Experimental
  A SIMD vector with two elements of type i8.
• i8x4Experimental
  A SIMD vector with four elements of type i8.
• i8x8Experimental
  A SIMD vector with eight elements of type i8.
• i8x16Experimental
  A SIMD vector with 16 elements of type i8.
• i8x32Experimental
  A SIMD vector with 32 elements of type i8.
• i8x64Experimental
  A SIMD vector with 64 elements of type i8.
• i16x1Experimental
  A SIMD vector with one element of type i16.
• i16x2Experimental
  A SIMD vector with two elements of type i16.
• i16x4Experimental
  A SIMD vector with four elements of type i16.
• i16x8Experimental
  A SIMD vector with eight elements of type i16.
• i16x16Experimental
  A SIMD vector with 16 elements of type i16.
• i16x32Experimental
  A SIMD vector with 32 elements of type i16.
• i16x64Experimental
  A SIMD vector with 64 elements of type i16.
• i32x1Experimental
  A SIMD vector with one element of type i32.
• i32x2Experimental
  A SIMD vector with two elements of type i32.
• i32x4Experimental
  A SIMD vector with four elements of type i32.
• i32x8Experimental
  A SIMD vector with eight elements of type i32.
• i32x16Experimental
  A SIMD vector with 16 elements of type i32.
• i32x32Experimental
  A SIMD vector with 32 elements of type i32.
• i32x64Experimental
  A SIMD vector with 64 elements of type i32.
• i64x1Experimental
  A SIMD vector with one element of type i64.
• i64x2Experimental
  A SIMD vector with two elements of type i64.
• i64x4Experimental
  A SIMD vector with four elements of type i64.
• i64x8Experimental
  A SIMD vector with eight elements of type i64.
• i64x16Experimental
  A SIMD vector with 16 elements of type i64.
• i64x32Experimental
  A SIMD vector with 32 elements of type i64.
• i64x64Experimental
  A SIMD vector with 64 elements of type i64.
• isizex1Experimental
  A SIMD vector with one element of type isize.
• isizex2Experimental
  A SIMD vector with two elements of type isize.
• isizex4Experimental
  A SIMD vector with four elements of type isize.
• isizex8Experimental
  A SIMD vector with eight elements of type isize.
• isizex16Experimental
  A SIMD vector with 16 elements of type isize.
• isizex32Experimental
  A SIMD vector with 32 elements of type isize.
• isizex64Experimental
  A SIMD vector with 64 elements of type isize.
• mask8x1Experimental
  A SIMD mask with one element for vectors with 8-bit element types.
• mask8x2Experimental
  A SIMD mask with two elements for vectors with 8-bit element types.
• mask8x4Experimental
  A SIMD mask with four elements for vectors with 8-bit element types.
• mask8x8Experimental
  A SIMD mask with eight elements for vectors with 8-bit element types.
• mask8x16Experimental
  A SIMD mask with 16 elements for vectors with 8-bit element types.
• mask8x32Experimental
  A SIMD mask with 32 elements for vectors with 8-bit element types.
• mask8x64Experimental
  A SIMD mask with 64 elements for vectors with 8-bit element types.
• mask16x1Experimental
  A SIMD mask with one element for vectors with 16-bit element types.
• mask16x2Experimental
  A SIMD mask with two elements for vectors with 16-bit element types.
• mask16x4Experimental
  A SIMD mask with four elements for vectors with 16-bit element types.
• mask16x8Experimental
  A SIMD mask with eight elements for vectors with 16-bit element types.
• mask16x16Experimental
  A SIMD mask with 16 elements for vectors with 16-bit element types.
• mask16x32Experimental
  A SIMD mask with 32 elements for vectors with 16-bit element types.
• mask16x64Experimental
  A SIMD mask with 64 elements for vectors with 16-bit element types.
• mask32x1Experimental
  A SIMD mask with one element for vectors with 32-bit element types.
• mask32x2Experimental
  A SIMD mask with two elements for vectors with 32-bit element types.
• mask32x4Experimental
  A SIMD mask with four elements for vectors with 32-bit element types.
• mask32x8Experimental
  A SIMD mask with eight elements for vectors with 32-bit element types.
• mask32x16Experimental
  A SIMD mask with 16 elements for vectors with 32-bit element types.
• mask32x32Experimental
  A SIMD mask with 32 elements for vectors with 32-bit element types.
• mask32x64Experimental
  A SIMD mask with 64 elements for vectors with 32-bit element types.
• mask64x1Experimental
  A SIMD mask with one element for vectors with 64-bit element types.
• mask64x2Experimental
  A SIMD mask with two elements for vectors with 64-bit element types.
• mask64x4Experimental
  A SIMD mask with four elements for vectors with 64-bit element types.
• mask64x8Experimental
  A SIMD mask with eight elements for vectors with 64-bit element types.
• mask64x16Experimental
  A SIMD mask with 16 elements for vectors with 64-bit element types.
• mask64x32Experimental
  A SIMD mask with 32 elements for vectors with 64-bit element types.
• mask64x64Experimental
  A SIMD mask with 64 elements for vectors with 64-bit element types.
• masksizex1Experimental
  A SIMD mask with one element for vectors with pointer-sized element types.
• masksizex2Experimental
  A SIMD mask with two elements for vectors with pointer-sized element types.
• masksizex4Experimental
  A SIMD mask with four elements for vectors with pointer-sized element types.
• masksizex8Experimental
  A SIMD mask with eight elements for vectors with pointer-sized element types.
• masksizex16Experimental
  A SIMD mask with 16 elements for vectors with pointer-sized element types.
• masksizex32Experimental
  A SIMD mask with 32 elements for vectors with pointer-sized element types.
• masksizex64Experimental
  A SIMD mask with 64 elements for vectors with pointer-sized element types.
• u8x1Experimental
  A SIMD vector with one element of type u8.
• u8x2Experimental
  A SIMD vector with two elements of type u8.
• u8x4Experimental
  A SIMD vector with four elements of type u8.
• u8x8Experimental
  A SIMD vector with eight elements of type u8.
• u8x16Experimental
  A SIMD vector with 16 elements of type u8.
• u8x32Experimental
  A SIMD vector with 32 elements of type u8.
• u8x64Experimental
  A SIMD vector with 64 elements of type u8.
• u16x1Experimental
  A SIMD vector with one element of type u16.
• u16x2Experimental
  A SIMD vector with two elements of type u16.
• u16x4Experimental
  A SIMD vector with four elements of type u16.
• u16x8Experimental
  A SIMD vector with eight elements of type u16.
• u16x16Experimental
  A SIMD vector with 16 elements of type u16.
• u16x32Experimental
  A SIMD vector with 32 elements of type u16.
• u16x64Experimental
  A SIMD vector with 64 elements of type u16.
• u32x1Experimental
  A SIMD vector with one element of type u32.
• u32x2Experimental
  A SIMD vector with two elements of type u32.
• u32x4Experimental
  A SIMD vector with four elements of type u32.
• u32x8Experimental
  A SIMD vector with eight elements of type u32.
• u32x16Experimental
  A SIMD vector with 16 elements of type u32.
• u32x32Experimental
  A SIMD vector with 32 elements of type u32.
• u32x64Experimental
  A SIMD vector with 64 elements of type u32.
• u64x1Experimental
  A SIMD vector with one element of type u64.
• u64x2Experimental
  A SIMD vector with two elements of type u64.
• u64x4Experimental
  A SIMD vector with four elements of type u64.
• u64x8Experimental
  A SIMD vector with eight elements of type u64.
• u64x16Experimental
  A SIMD vector with 16 elements of type u64.
• u64x32Experimental
  A SIMD vector with 32 elements of type u64.
• u64x64Experimental
  A SIMD vector with 64 elements of type u64.
• usizex1Experimental
  A SIMD vector with one element of type usize.
• usizex2Experimental
  A SIMD vector with two elements of type usize.
• usizex4Experimental
  A SIMD vector with four elements of type usize.
• usizex8Experimental
  A SIMD vector with eight elements of type usize.
• usizex16Experimental
  A SIMD vector with 16 elements of type usize.
• usizex32Experimental
  A SIMD vector with 32 elements of type usize.
• usizex64Experimental
  A SIMD vector with 64 elements of type usize.