ggmath
A linear algebra library for games and graphics with generic SIMD types.
The library provides:
- Vectors:
Vec2<T>,Vec3<T>,Vec4<T>. - Square Matrices:
Mat2<T>,Mat3<T>,Mat4<T>. - Quaternions:
Quat<T>. - Affine Transforms:
Affine2<T>,Affine3<T>. - Masks:
Mask2<T>,Mask3<T>,Mask4<T>.
SIMD
Appropriate types have increased memory alignment in order to take advantage of
SIMD instructions that improve performance. For example, Vec3<f32>,
Vec4<f32>, Mat3<f32> and Mat4<f32> are aligned to 16 bytes on x86 targets
in order to take advantage of the SSE instruction set.
Although SIMD alignment generally results better performance, it can also result
in wasted space. For example, due to 16-byte alignment, Vec3<f32> has 4 bytes
of padding, and consequently Mat3<f32> has 12 bytes of padding. For scenarios
where better performance is not worth wasted space, math types have non-SIMD,
unaligned variants:
- Vectors:
Vec2U<T>,Vec3U<T>,Vec4U<T>. - Square Matrices:
Mat2U<T>,Mat3U<T>,Mat4U<T>. - Quaternions:
QuatU<T>. - Affine Transforms:
Affine2U<T>,Affine3U<T>. - Masks:
Mask2U<T>,Mask3U<T>,Mask4U<T>.
Unaligned types are optimal in memory-critical scenarios, for example when storing 3D models. In all other cases, aligned types are optimal and result in better performance than unaligned types.
Currently SIMD optimizations are only implemented for f32 types on x86
targets. These types are closely benchmarked against glam and generally
match its performance.
Integration with wide enables SoA (Structure of Arrays) SIMD, which lets
you perform operations concurrently on multiple values, for example with
Vec3<f32x4> which represents four values of Vec3<f32>. SoA requires modeling
algorithms in a very specific way, but can be much faster than normal types.
Generics
Because types are generic over T, they support non-primitive scalar types.
Integration with fixed enables support for fixed-point numbers, and
integration with wide enables support for SoA.
When Rust's type system is powerful enough, integration with num-primitive
will enable writing math code that is generic over primitive types, for example
functions generic over T: PrimitiveFloat will have access to float-vector
functionality.
Types relative to each other (e.g., Vec2<T>, Vec3<T>, Vec4<T> and
unaligned variants) are not distinct types, instead they are all type aliases to
these const-generic structs:
Vector<N, T, A>.Matrix<N, T, A>.Quaternion<T, A>.Affine<N, T, A>.Mask<N, T, A>.
Where:
Nis the length (2, 3, or 4).Tis the scalar type.Ais eitherAlignedorUnaligned.
Const generics eliminate the need for macros, making it easier to implement
functionality for all lengths (and both alignments). For example, instead of
defining seperate Ray2 and Ray3 types, it is possible to define a single
Ray<N, T, A> type then define type aliases for it.
Math conventions
ggmath is coordinate-system agnostic, and should work for both right-handed
and left-handed coordinate systems.
Vectors are treated as column matrices, meaning when transforming a vector with a matrix, the matrix goes on the left.
Matrices are stored in column-major order, meaning each column is continuous in memory.
Angles are in radians, but can be converted to and from degrees using standard-library functions.
Development status
ggmath is not mature yet but is under active development.
Feature List:
- Vectors
- Square Matrices
- Quaternions
- Affine Transforms
- Masks
- Sufficient Float-Vector functionality
- Sufficient Int-Vector functionality
- Sufficient Matrix functionality
- Sufficient Quaternion functionality
- Sufficient Affine functionality
Crate Support:
Performance:
-
f32SSE2 optimizations -
i32u32SSE2 optimizations -
f32NEON optimizations -
i32u32NEON optimizations -
f32WASM optimizations -
i32u32WASM optimizations - Niche
f32SSE4.2+ optimizations - Niche
i32u32SSE4.2+ optimizations - Niche
f64AVX+ optimizations - Niche
i8u8boolSSE2+ optimizations - Niche
i16u16SSE2+ optimizations - Niche
i8u8boolNEON optimizations - Niche
i16u16NEON optimizations - Niche
i8u8boolWASM optimizations - Niche
i16u16WASM optimizations
Usage
Rust must be updated to version 1.90.0 or later.
Add this to your Cargo.toml:
[]
= "0.16.3"
For no_std support, enable the libm feature:
[]
= { = "0.16.3", = ["libm"] }
For no_std without libm, disable default features:
[]
= { = "0.16.3", = false }
Without std or libm, the crate compiles but all float functionality that
relies on a backend is disabled.
Optional features
-
std(default feature): Usesstdas the backend for float functionality. -
assertions: Enables assertions in release mode. Assertions are panics that catch invalid input and are enabled by default in debug mode. -
no-assertions: Disables assertions in debug mode. Assertions should only be controlled by binary crates. Library crates should not set this flag directly. -
bytemuck: Implementsbytemucktraits forggmathtypes. -
fixed: ImplementsScalarfor fixed-point numbers. -
fixp: ImplementsScalarfor fixed-point numbers. -
libm: Useslibmas the backend for float functionality. This makes the crateno_stdeven if thestdfeature is not disabled. -
mint: Implements conversions betweenggmathandminttypes. -
serde: ImplementsSerializeandDeserializeforggmathtypes. -
wide: ImplementsScalarfor SIMD types.
License
Licensed under either Apache License Version 2.0 or MIT license at your option.
Contribution
See CONTRIBUTING.md.
Contributions in any form (issues, pull requests, etc.) to this project must adhere to Rust's Code of Conduct.
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.
Attribution
ggmath is heavily inspired by glam and ports most of its code from it, as
it serves the same purpose as glam but with generics.