• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*!
2 # glam
3 
4 `glam` is a simple and fast linear algebra library for games and graphics.
5 
6 ## Features
7 
8 * [`f32`](mod@f32) types
9   * vectors: [`Vec2`], [`Vec3`], [`Vec3A`] and [`Vec4`]
10   * square matrices: [`Mat2`], [`Mat3`], [`Mat3A`] and [`Mat4`]
11   * a quaternion type: [`Quat`]
12   * affine transformation types: [`Affine2`] and [`Affine3A`]
13 * [`f64`](mod@f64) types
14   * vectors: [`DVec2`], [`DVec3`] and [`DVec4`]
15   * square matrices: [`DMat2`], [`DMat3`] and [`DMat4`]
16   * a quaternion type: [`DQuat`]
17   * affine transformation types: [`DAffine2`] and [`DAffine3`]
18 * [`i16`](mod@i16) types
19   * vectors: [`I16Vec2`], [`I16Vec3`] and [`I16Vec4`]
20 * [`u16`](mod@u16) types
21   * vectors: [`U16Vec2`], [`U16Vec3`] and [`U16Vec4`]
22 * [`i32`](mod@i32) types
23   * vectors: [`IVec2`], [`IVec3`] and [`IVec4`]
24 * [`u32`](mod@u32) types
25   * vectors: [`UVec2`], [`UVec3`] and [`UVec4`]
26 * [`i64`](mod@i64) types
27   * vectors: [`I64Vec2`], [`I64Vec3`] and [`I64Vec4`]
28 * [`u64`](mod@u64) types
29   * vectors: [`U64Vec2`], [`U64Vec3`] and [`U64Vec4`]
30 * [`bool`](mod@bool) types
31   * vectors: [`BVec2`], [`BVec3`] and [`BVec4`]
32 
33 ## SIMD
34 
35 `glam` is built with SIMD in mind. Many `f32` types use 128-bit SIMD vector types for storage
36 and/or implementation. The use of SIMD generally enables better performance than using primitive
37 numeric types such as `f32`.
38 
39 Some `glam` types use SIMD for storage meaning they are 16 byte aligned, these types include
40 `Mat2`, `Mat3A`, `Mat4`, `Quat`, `Vec3A`, `Vec4`, `Affine2` an `Affine3A`. Types
41 with an `A` suffix are a SIMD alternative to a scalar type, e.g. `Vec3` uses `f32` storage and
42 `Vec3A` uses SIMD storage.
43 
44 When SIMD is not available on the target the types will maintain 16 byte alignment and internal
45 padding so that object sizes and layouts will not change between architectures. There are scalar
46 math fallback implementations exist when SIMD is not available. It is intended to add support for
47 other SIMD architectures once they appear in stable Rust.
48 
49 Currently only SSE2 on x86/x86_64 is supported as this is what stable Rust supports.
50 
51 ## Vec3A and Mat3A
52 
53 `Vec3A` is a SIMD optimized version of the `Vec3` type, which due to 16 byte alignment results
54 in `Vec3A` containing 4 bytes of padding making it 16 bytes in size in total. `Mat3A` is composed
55 of three `Vec3A` columns.
56 
57 | Type       | `f32` bytes | Align bytes | Size bytes | Padding |
58 |:-----------|------------:|------------:|-----------:|--------:|
59 |[`Vec3`]    |           12|            4|          12|        0|
60 |[`Vec3A`]   |           12|           16|          16|        4|
61 |[`Mat3`]    |           36|            4|          36|        0|
62 |[`Mat3A`]   |           36|           16|          48|       12|
63 
64 Despite this wasted space the SIMD implementations tend to outperform `f32` implementations in
65 [**mathbench**](https://github.com/bitshifter/mathbench-rs) benchmarks.
66 
67 `glam` treats [`Vec3`] as the default 3D vector type and [`Vec3A`] a special case for optimization.
68 When methods need to return a 3D vector they will generally return [`Vec3`].
69 
70 There are [`From`] trait implementations for converting from [`Vec4`] to a [`Vec3A`] and between
71 [`Vec3`] and [`Vec3A`] (and vice versa).
72 
73 ```
74 use glam::{Vec3, Vec3A, Vec4};
75 
76 let v4 = Vec4::new(1.0, 2.0, 3.0, 4.0);
77 
78 // Convert from `Vec4` to `Vec3A`, this is a no-op if SIMD is supported.
79 let v3a = Vec3A::from(v4);
80 assert_eq!(Vec3A::new(1.0, 2.0, 3.0), v3a);
81 
82 // Convert from `Vec3A` to `Vec3`.
83 let v3 = Vec3::from(v3a);
84 assert_eq!(Vec3::new(1.0, 2.0, 3.0), v3);
85 
86 // Convert from `Vec3` to `Vec3A`.
87 let v3a = Vec3A::from(v3);
88 assert_eq!(Vec3A::new(1.0, 2.0, 3.0), v3a);
89 ```
90 
91 ## Affine2 and Affine3A
92 
93 `Affine2` and `Affine3A` are composed of a linear transform matrix and a vector translation. The
94 represent 2D and 3D affine transformations which are commonly used in games.
95 
96 The table below shows the performance advantage of `Affine2` over `Mat3A` and `Mat3A` over `Mat3`.
97 
98 | operation          | `Mat3`      | `Mat3A`    | `Affine2`  |
99 |--------------------|-------------|------------|------------|
100 | inverse            | 11.4±0.09ns | 7.1±0.09ns | 5.4±0.06ns |
101 | mul self           | 10.5±0.04ns | 5.2±0.05ns | 4.0±0.05ns |
102 | transform point2   |  2.7±0.02ns | 2.7±0.03ns | 2.8±0.04ns |
103 | transform vector2  |  2.6±0.01ns | 2.6±0.03ns | 2.3±0.02ns |
104 
105 Performance is much closer between `Mat4` and `Affine3A` with the affine type being faster to
106 invert.
107 
108 | operation          | `Mat4`      | `Affine3A`  |
109 |--------------------|-------------|-------------|
110 | inverse            | 15.9±0.11ns | 10.8±0.06ns |
111 | mul self           |  7.3±0.05ns |  7.0±0.06ns |
112 | transform point3   |  3.6±0.02ns |  4.3±0.04ns |
113 | transform point3a  |  3.0±0.02ns |  3.0±0.04ns |
114 | transform vector3  |  4.1±0.02ns |  3.9±0.04ns |
115 | transform vector3a |  2.8±0.02ns |  2.8±0.02ns |
116 
117 Benchmarks were taken on an Intel Core i7-4710HQ.
118 
119 ## Linear algebra conventions
120 
121 `glam` interprets vectors as column matrices (also known as column vectors) meaning when
122 transforming a vector with a matrix the matrix goes on the left.
123 
124 ```
125 use glam::{Mat3, Vec3};
126 let m = Mat3::IDENTITY;
127 let x = Vec3::X;
128 let v = m * x;
129 assert_eq!(v, x);
130 ```
131 
132 Matrices are stored in memory in column-major order.
133 
134 All angles are in radians. Rust provides the `f32::to_radians()` and `f64::to_radians()` methods to
135 convert from degrees.
136 
137 ## Direct element access
138 
139 Because some types may internally be implemented using SIMD types, direct access to vector elements
140 is supported by implementing the [`Deref`] and [`DerefMut`] traits.
141 
142 ```
143 use glam::Vec3A;
144 let mut v = Vec3A::new(1.0, 2.0, 3.0);
145 assert_eq!(3.0, v.z);
146 v.z += 1.0;
147 assert_eq!(4.0, v.z);
148 ```
149 
150 [`Deref`]: https://doc.rust-lang.org/std/ops/trait.Deref.html
151 [`DerefMut`]: https://doc.rust-lang.org/std/ops/trait.DerefMut.html
152 
153 ## glam assertions
154 
155 `glam` does not enforce validity checks on method parameters at runtime. For example methods that
156 require normalized vectors as input such as `Quat::from_axis_angle(axis, angle)` will not check
157 that axis is a valid normalized vector. To help catch unintended misuse of `glam` the
158 `debug-glam-assert` or `glam-assert` features can be enabled to add checks ensure that inputs to
159 are valid.
160 
161 ## Vector swizzles
162 
163 `glam` vector types have functions allowing elements of vectors to be reordered, this includes
164 creating a vector of a different size from the vectors elements.
165 
166 The swizzle functions are implemented using traits to add them to each vector type. This is
167 primarily because there are a lot of swizzle functions which can obfuscate the other vector
168 functions in documentation and so on. The traits are [`Vec2Swizzles`], [`Vec3Swizzles`] and
169 [`Vec4Swizzles`].
170 
171 Note that the [`Vec3Swizzles`] implementation for [`Vec3A`] will return a [`Vec3A`] for 3 element
172 swizzles, all other implementations will return [`Vec3`].
173 
174 ```
175 use glam::{swizzles::*, Vec2, Vec3, Vec3A, Vec4};
176 
177 let v = Vec4::new(1.0, 2.0, 3.0, 4.0);
178 
179 // Reverse elements of `v`, if SIMD is supported this will use a vector shuffle.
180 let wzyx = v.wzyx();
181 assert_eq!(Vec4::new(4.0, 3.0, 2.0, 1.0), wzyx);
182 
183 // Swizzle the yzw elements of `v` into a `Vec3`
184 let yzw = v.yzw();
185 assert_eq!(Vec3::new(2.0, 3.0, 4.0), yzw);
186 
187 // To swizzle a `Vec4` into a `Vec3A` swizzle the `Vec4` first then convert to
188 // `Vec3A`. If SIMD is supported this will use a vector shuffle. The last
189 // element of the shuffled `Vec4` is ignored by the `Vec3A`.
190 let yzw = Vec3A::from(v.yzwx());
191 assert_eq!(Vec3A::new(2.0, 3.0, 4.0), yzw);
192 
193 // You can swizzle from a `Vec4` to a `Vec2`
194 let xy = v.xy();
195 assert_eq!(Vec2::new(1.0, 2.0), xy);
196 
197 // And back again
198 let yyxx = xy.yyxx();
199 assert_eq!(Vec4::new(2.0, 2.0, 1.0, 1.0), yyxx);
200 ```
201 
202 ## SIMD and scalar consistency
203 
204 `glam` types implement `serde` `Serialize` and `Deserialize` traits to ensure
205 that they will serialize and deserialize exactly the same whether or not
206 SIMD support is being used.
207 
208 The SIMD versions implement the `core::fmt::Debug` and `core::fmt::Display`
209 traits so they print the same as the scalar version.
210 
211 ```
212 use glam::Vec4;
213 let a = Vec4::new(1.0, 2.0, 3.0, 4.0);
214 assert_eq!(format!("{}", a), "[1, 2, 3, 4]");
215 ```
216 
217 ## Feature gates
218 
219 All `glam` dependencies are optional, however some are required for tests
220 and benchmarks.
221 
222 * `std` - the default feature, has no dependencies.
223 * `approx` - traits and macros for approximate float comparisons
224 * `bytemuck` - for casting into slices of bytes
225 * `libm` - uses `libm` math functions instead of `std`, required to compile with `no_std`
226 * `mint` - for interoperating with other 3D math libraries
227 * `rand` - implementations of `Distribution` trait for all `glam` types.
228 * `rkyv` - implementations of `Archive`, `Serialize` and `Deserialize` for all
229   `glam` types. Note that serialization is not interoperable with and without the
230   `scalar-math` feature. It should work between all other builds of `glam`.
231   Endian conversion is currently not supported
232 * `bytecheck` - to perform archive validation when using the `rkyv` feature
233 * `serde` - implementations of `Serialize` and `Deserialize` for all `glam`
234   types. Note that serialization should work between builds of `glam` with and without SIMD enabled
235 * `scalar-math` - disables SIMD support and uses native alignment for all types.
236 * `debug-glam-assert` - adds assertions in debug builds which check the validity of parameters
237   passed to `glam` to help catch runtime errors.
238 * `glam-assert` - adds assertions to all builds which check the validity of parameters passed to
239   `glam` to help catch runtime errors.
240 * `cuda` - forces `glam` types to match expected cuda alignment
241 * `fast-math` - By default, glam attempts to provide bit-for-bit identical
242   results on all platforms. Using this feature will enable platform specific
243   optimizations that may not be identical to other platforms. **Intermediate
244   libraries should not use this feature and defer the decision to the final
245   binary build**.
246 * `core-simd` - enables SIMD support via the portable simd module. This is an
247   unstable feature which requires a nightly Rust toolchain and `std` support.
248 
249 ## Minimum Supported Rust Version (MSRV)
250 
251 The minimum supported Rust version is `1.58.1`.
252 
253 */
254 #![doc(html_root_url = "https://docs.rs/glam/0.25.0")]
255 #![cfg_attr(not(feature = "std"), no_std)]
256 #![cfg_attr(target_arch = "spirv", feature(repr_simd))]
257 #![deny(
258     rust_2018_compatibility,
259     rust_2018_idioms,
260     future_incompatible,
261     nonstandard_style
262 )]
263 // clippy doesn't like `to_array(&self)`
264 #![allow(clippy::wrong_self_convention)]
265 #![cfg_attr(
266     all(feature = "core-simd", not(feature = "scalar-math")),
267     feature(portable_simd)
268 )]
269 
270 #[macro_use]
271 mod macros;
272 
273 mod align16;
274 mod deref;
275 mod euler;
276 mod features;
277 
278 #[cfg(target_arch = "spirv")]
279 mod spirv;
280 
281 #[cfg(all(
282     target_feature = "sse2",
283     not(any(feature = "core-simd", feature = "scalar-math"))
284 ))]
285 mod sse2;
286 
287 #[cfg(all(
288     target_feature = "simd128",
289     not(any(feature = "core-simd", feature = "scalar-math"))
290 ))]
291 mod wasm32;
292 
293 #[cfg(all(feature = "core-simd", not(feature = "scalar-math")))]
294 mod coresimd;
295 
296 #[cfg(all(
297     target_feature = "sse2",
298     not(any(feature = "core-simd", feature = "scalar-math"))
299 ))]
300 use align16::Align16;
301 
302 /** `bool` vector mask types. */
303 pub mod bool;
304 pub use self::bool::*;
305 
306 /** `f32` vector, quaternion and matrix types. */
307 pub mod f32;
308 pub use self::f32::*;
309 
310 /** `f64` vector, quaternion and matrix types. */
311 pub mod f64;
312 pub use self::f64::*;
313 
314 /** `i16` vector types. */
315 pub mod i16;
316 pub use self::i16::*;
317 
318 /** `u16` vector types. */
319 pub mod u16;
320 pub use self::u16::*;
321 
322 /** `i32` vector types. */
323 pub mod i32;
324 pub use self::i32::*;
325 
326 /** `u32` vector types. */
327 pub mod u32;
328 pub use self::u32::*;
329 
330 /** `i64` vector types. */
331 pub mod i64;
332 pub use self::i64::*;
333 
334 /** `u64` vector types. */
335 pub mod u64;
336 pub use self::u64::*;
337 
338 /** Traits adding swizzle methods to all vector types. */
339 pub mod swizzles;
340 pub use self::swizzles::{Vec2Swizzles, Vec3Swizzles, Vec4Swizzles};
341 
342 /** Rotation Helper */
343 pub use euler::EulerRot;
344 
345 /** A trait for extending [`prim@f32`] and [`prim@f64`] with extra methods. */
346 mod float;
347 pub use float::FloatExt;
348